Introduction According to the National Association

According to the National Association of Social Workers Code of Ethics (2008),

social workers are expected to counter oppression and fight for expanding the rights of

clients to basic social and economic resources. The skills and responsibilities expected of

practitioners include advocacy for their clients, and social justice for all (Sunley,1983;

Hoefer, 1993; Michelson, 1995; Gravin & Seabury, 1997; Ezell, 2001).

In view of these professional norms, one would expect that the profession would

have a clear definition of what constitutes advocacy, and that the prevalence and practice

of social work advocacy would be well documented. However, social work advocacy is

defined in a myriad of ways in the social work literature and elsewhere (Ezell, 1991), and

advocacy appears to be less prevalent than one might expect (Hagen, 1994; Mickelson,

1995; Spetch & Courtney, 1994; Donaldson, 2007). While the body of knowledge

generated by research examining the organizational determinants of advocacy has grown,

the organizational factors that support or hinder nonprofit advocacy are not yet fully

understood (Salamon, 1995; Rees, 2000; Berry, 2003; Chaves, Stephens & Galaskiewicz,

2004; Donaldson, 2007; Child & Gronbjerg, 2007). More research is needed on these

topics, and further development of social work theory that explains advocacy in the

nonprofit context is needed.

Advancing advocacy in the field of social work demands that we illuminate

specific ways that social work practice supports advocacy for social change on all levels

of society. Promoting social work advocacy requires more clear definitions of advocacy

and social change strategies, as well as greater understanding of the organizational

characteristics of human service agencies that support advocacy. Addressing the dearth

of both theory and empirical research on advocacy is of utmost importance to the field.

This study examines social work advocacy, and advocacy, in general, among

nonprofit human service and other nonprofit social welfare organizations. Advocacy is a

central component of social work practice, and thus one would expect to find significant

levels of advocacy being conducted by these nonprofits (Mickelson, 1995; Sunley, 1983).

Furthermore, nonprofit advocacy is an essential force in present day American democratic

society which strengthens civic engagement in policy issues, and provides alternative

views to those expressed by government and the market, and ideally offers venues for

citizen empowerment and engagement (Reid, 2000; Putnam, 2000; Dahl, 1998). Both

personal and political empowerment processes are seen by many leading scholars and

practitioners as being integral to health and personal growth (Gutierrez & Lewis, 1994;

Simon, 1994; Putnam & DeWeever Jones, 2001; Klar & Kasser, 2009). In this study,

empowerment and democratic theories guided an exploratory and descriptive study of

empowerment-based advocacy among a sample of local nonprofit human service

providers that are likely to conduct advocacy activities.

In the next section, the purpose of the study, definitions of advocacy and a brief

review of the knowledge base to date on advocacy will be covered. This material will

provide the background necessary for understanding the intentions and aims of this

research project.

Background: Studying advocacy in the nonprofit sector

One of the foremost national nonprofit leaders once stated that advocacy and

facilitating citizen involvement in public policy making is the “quintessential role” of the

nonprofit sector” (O‟Connell, 1997, p. 103). Nonprofits are also seen as a key force in the

building of social capital, which is viewed by Putnam (1995) as consisting of the trust and

social connections between people that enable a democracy to function. Social capital is

also seen as “the cooperative networks that permit individuals to work together for

mutual goals” (Boris, 1999:18), that provide the building blocks for creating civil society.

In fact, the term “civil society” is now most often regarded as encompassing the nonprofit

sector (Burbidge, 1997; Boris, 1999), which Rueschemeyer et al, (1992:49) describe as

the “totality of social institutions and associations, both formal and informal that are not

strictly production-related nor governmental or familial in character.” The nonprofit

sector serves a central function in weaving and maintaining the social fabric of our lives,

and in particular, offers vital avenues for civic engagement (Saidel, 2002; Boris, 1999;

Dahl, 1998).

Nonprofits are seen as the preferred delivery mechanism in America for social and

other public services rather than government for a number of reasons, including the

American emphasis on local control and preference that services are ideally delivered in a

manner tailored to the local community (Salamon, 1995; Gibelman & Schervish, 1993).

In short, the congruence between the social work profession and the nonprofit sector in

regards to advocacy is evident in the overlap of their similar functions and goals in

American society, i.e. striving toward social justice and charitable or public benefit

purposes.

A wide review of social work articles containing the word advocacy or advocate

indicates that a great many social work practitioners and scholars call for increased

advocacy, yet few studies empirically document such efforts. Even when advocacy is

viewed as an appropriate function, it is not necessarily a method that is fully employed by

the social work practitioner (Hagen, 1994; Sunley, 1997). While social change efforts for

justice and equality are espoused in general for social work practice (Germain &

Gitterman, 1996), social work theory that encompasses advocacy at all socio-political

levels is sparse (MacNair, 1996; Payne, 1997), and most social work practices focus on

the individual or family as the target of intervention, instead of societal structures

(Everett, Homstead and Drisko, 2007; Ezell, 1991; Spetch & Courtney, 1994; Mickelson,

1995; Sunley, 1997).

There are several reasons for focusing on advocacy conducted within the

nonprofit sector by charitable, or 501(c)(3) nonprofits. As more social workers are

employed in the nonprofit sector than in government or business (Gibelman & Schervish,

1993), the examination of both the extent and types of advocacy conducted in this sector

is of importance to the field. And most importantly, advocacy is considered to be an

essential function of the nonprofit sector, since the sector is positioned outside of

government, and is independent of proprietary interests (Jenkins, 1987; Salamon, 1995;

O‟Connell, 1997). Some leading nonprofit scholars state that nonprofit

advocacy activities provide a major avenue for citizen participation (Reid, 2000;

Boris & MosherWilliams, 1998), and citizen participation is one social work

practice arena. As Gamble and Weil note:

Social workers who engage in empowerment-focused practice seek to develop the

capacity of individuals to understand their environment, make choices, take

responsibility for their choices, and influence their life situations through

organization and advocacy....[nonprofits] provide an organizational arena in

which to practice democracy and to influence representative government.

(1995:483)

Research is needed to reveal the full extent to which social workers are engaging

in advocacy in professional contexts. Some researchers have found that the beliefs,

attitudes and self-reports of social workers imply that they conduct (either on their own or

within their agencies) much more advocacy than would be expected of social workers

(Ezell, 1991; Reeser & Epstein, 1990). Ezell (1991) found in a survey of social workers,

most of whom worked in the nonprofit sector, that 68% of direct practice or clinical

social workers and 45.4 % of macro practitioners engaged in little or no advocacy on their

job. Even with the implementation of the JOBS program which was structured to

maximize client choices and supportive services, and also subsidized case management

services, case advocacy was rarely conducted (Hagen, 1994). While it is heartening that

Ezell found that many macro level social workers engage in advocacy, the lower rate of

advocacy among clinical social workers is troubling since the great majority of social

workers are in direct practice.

In the past few decades the promotion of empowerment-based approaches has

increased in the social work field (Gutierrez et. al., 2005; Weil, 2005). However, the

frequency and intensity of bona fide political empowerment activities that would indicate

the full integration of empowerment approaches in social work practice have not been

documented in social science research (Everett, Homstead and Drisko, 2007; Donaldson,

2007).

The low level of advocacy some researchers have found in nonprofits and among

social workers (Ezell, 1991; Salamon, 1995) may be due in part to the problem of

differing definitions used to describe advocacy and related activities. In addition, the lack

of attention researchers might give to the political realities of conducting advocacy within

an organizational context could also be confusing the picture. For example, the

measurement of advocacy activities of social workers employed by state or local

government agencies may not take into account the political and legal constraints of

operating as an agent of the executive branch. Another critical reason for lower levels of

advocacy is the lack of time caused by work overload and cut backs to many human

service organizations (Abramovitz, 2005; Donaldson, 2007).

The problem of defining and examining advocacy

To describe the extent and kinds of advocacy conducted in the field, first one must

define the term. Frequently, very general measures of advocacy are used in studies that

do not differentiate between types or socio-political levels of advocacy (Kramer, 1983;

Mondros and Wilson, 1994; Salamon, 1995; Weed, 1997), or advocacy is left mostly

undefined (Sosin, 1986). These variations mean that consistent, comparable measures of

the extent of advocacy in the nonprofit sector and social work advocacy remain elusive

(Ezell, 1991;1994; Salamon, 1995). Some nonprofit scholars assert that a great number of

nonprofits engage in public policy advocacy and other activities that support civil society

and democracy (Boris and Mosher-Williams, 1998; Reid, 2000), but to date, there is not

much empirical support for their claim.

Advocacy is broadly defined as the “promotion of measures to support and

enhance human life individually and generally” (Sunley, 1983, p.9). The Social Work

Dictionary defines advocacy as action that empowers individuals or communities (Barker,

1991), implying that social work advocacy is something done only with a client, not for a

client. However, Mickelson (1995:95) notes that some vulnerable groups such as

children and the severely mentally ill cannot be fully empowered. With this in mind, he

provides us with a more specific definition: “In social work, advocacy can be defined as

the act of directly representing, defending, intervening, supporting, or recommending a

course of action on behalf of one or more individuals, groups, or communities, with the

goal of securing or retaining social justice.”

In the social work and public policy literature, advocacy is typically viewed as

consisting of two basic categories, case advocacy and class (sometimes referred to as

cause) advocacy (Reisch, 1986b; Mickelson, 1995; Heimovics, 1998). Most research has

shown that the most frequently conducted social work advocacy function is case

advocacy, which means advocating for individuals to have adequate goods and services,

or for other individual-level interventions on the micro level (Piccard, 1983; Ezell, 1991;

Klein and Cnaan, 1995). Cause (also known as class) advocacy consists of those actions

taken to ameliorate some social problem or issue that are directed toward a macro target

that can change existing policies in order to impact a broader group, or class, or people

such as the mentally ill or poor.

Case advocacy alone will rarely result in the creation of new resources or

entitlement, since the impact of successful case advocacy is by definition limited to the

people involved. The one exception to this is that litigation based on case advocacy can

establish or expand a constitutional or legal right that will benefit others. Cause advocacy

involves advocating for or with a group of individuals to access services, resources and

opportunities or establish legal rights through the creation of new laws or entitlements,

including government budget allocations. Policy and political science scholars typically

define advocacy as cause-related activities that solely target policy makers, yet this macro

definition ignores other empowerment-related activities that informs and mobilizes

individuals and the public. Since case and cause advocacy do not encompass such

efforts, perhaps a third category of advocacy that captures activities such as education,

deliberation, and skills in democratic participation and collective action may be helpful in

studying advocacy. An expanded notion of advocacy outlined as a “tripartite concept of

empowerment”, described by Beck (1983:10) may be useful in resolving “the dilemma of

how to interrelate personal strength with social participation”.

Previous Relevant Research on Advocacy

The research to date on organizational factors that may inhibit or support

advocacy is growing, but yet still scholarship in this area has not yet definitively

addressed many key questions (Kramer, 1994; Salamon, 1995; McCarthy & Castelli,

2002; Berry, 2003; Child & Gronjberg, 2007). The idea that professionalism decreases

advocacy is in concert with organizational and institutional theories that claim that the

contractual relationship between government and nonprofits has a deleterious effect on

nonprofit advocacy. Earlier research found support for this notion that because nonprofits

aim to attract government support, they are seen as being loathe to bite the hand that

feeds them (Kramer, 1981, Smith & Lipsky,1993; Salamon, 1995).

In additional to the aforementioned organizational-level variables that impact

advocacy, Kramer (1994) notes that the diversity of an organization‟s funding base plays

a role in an organization‟s perception of relative risk and their degree of confidence to

take a position counter to government. Their position in the marketplace (i.e. whether or

not they are the sole provider of a needed service in an area) and their perceived political

clout are additional factors that bolster their position relative to government‟s influence.

A growing body of research findings on the impact of government funding counter claims

that contracting with government lessens the ability of nonprofits to advocate, and instead

these findings bolster resource theory that explains advocacy as being the result of greater

organizational capacity (Ostrander, 1989; Salamon, 1995; Berry, 2003; Chaves,

Galaskiewicz & Stephens, 2004; Donaldson, 2007).

Few studies document how professionalism and the educational level of staff

affect advocacy. Donaldson (2007) found that leadership was the most important factor

that enhanced the prospects of a nonprofit human service agency engaging in advocacy.

A survey of 41 community organizers found that none of these organizers had a social

work degree (O‟Donnell, 1995). Diversity and active representation of underrepresented

groups in nonprofit sector leadership and management may also be an important factor in

promoting advocacy. In one rare and unduplicated study, Reisch (1986) found that

multiracial organizations (defined as organizations with 20% or more of their staff and

leadership positions filled with minorities) not only engaged in more advocacy activity,

but they were more likely to achieve recognition and success. The multi-racial

organizations were on the average larger nonprofits (more than 20 staff), so the

availability of resources and larger size of the organization may be the critical variable

here, as it appeared that size was not used as a control variable in the analysis. These

results provide support for Reeser and Epstein‟s (1990) finding that black social workers

were more activist and engaged in more advocacy than white social workers. Clearly

more research is needed to reveal the factors impacting social work advocacy in the

nonprofit sector.

While studies have found varying rates of lobbying and other kinds of political

activity among nonprofits, most find that generally less than 50% of nonprofits lobby

policy makers. Two older national studies of nonprofit organizations using accepted

sampling techniques which included human service organizations found that the

percentage of organizations conducting advocacy (under 20%) was lower than might be

anticipated in nonprofits (Sosin, 1986; Salamon, 1995). A recent study using a sample of

mostly local nonprofits in Indiana yielded similar rates (Child & Gronjberg, 2007). While

Berry 2003) found in his large, relatively representative sample that 74% of nonprofits

lobbied, only 31% did so frequently. Chaves et. al. (2004) found that only 11% of

congregations that did not receive any government funding lobbied, while 39% of

congregations receiving government funding lobbied. Studies using broader definitions

of advocacy that encompass policy change activities other than lobbying find higher rates

of participation (Abramovitz, 2005; Donaldson, 2007). Abramovitz (2005) found that

62% of New York City agencies were actively trying to change welfare polices after

welfare reform, and more than 50% of these nonprofits engaged in all six of the change

efforts the author measured, which included rallies, educational meetings and trying to

educate policy makers. Rates of lobbying found in studies also are likely to vary

significantly depending on the manner in which the nonprofits were selected, as aside

from compiling a list from the ground up (an extremely intensive and expensive

proposition), all lists of nonprofits are biased (Gronjberg, 2002; Child & Gronjberg,

2007).

Definitions of Advocacy Used in This Study

While the theory guiding this study will be presented more thoroughly in the next

section, in order to set the stage and context for discussing theory in the next sections, the

empowerment theory and the definitions of advocacy used in this study are presented

here. These were developed as part of a study on nonprofit organizations by the

Nonprofit Education Initiative. The empowerment theoretical approach used in this study

is consonant with the definitions of advocacy also used in this study (defined below) of

case advocacy, public policy education and legislative issue advocacy, as empowerment

theorists have identified this sequence of activities as being crucial to the empowerment

process (Beck, 1983; Gutierrez and Lewis, 1999). For example one of the earliest writers

on the topic, Bertram Beck (1983:4) noted that: “three interlocking dimensions are

included in the concept of empowerment,” and these can be roughly correlated

respectively to case advocacy, public policy education and legislative issue advocacy:

1) Development of a more positive and potent sense of self.

2) Construction of knowledge and capacity for more critical comprehension

of the web of social and political realities which comprise one‟s experienced

environment; and

3) Cultivation of resources and strategies, or more functional competence, for

efficacious and proactive attainment of personal and collective social goals.

Categories of advocacy were developed by this researcher in an attempt to define

a full spectrum of advocacy activities relating to advocacy conducted in the nonprofit

sector within human service or social welfare advocacy nonprofits. This definition of

advocacy was used in this research project in order to capture the types of advocacy in

greater detail than most previous studies. These three categories: case advocacy, public

policy education, and legislative issue advocacy, are defined below.

Case Advocacy

Case advocacy involves advocating for an individual to increase access to services

or ensure legal rights. It differs from class advocacy in that the activity is done on an

individual or case by case basis, as opposed to advocacy on behalf of a group. Case

advocacy may include such activities as accompanying a client to court and working with

or for her/him to obtain due process and legal rights, or to assist in obtaining public

assistance or other goods and services, and conducting empowermentbased interventions

to encourage people to advocate for themselves.

Public Policy Education

Public policy education includes a number of educational activities and methods,

with the overarching principle that the information must be provided in such a way as to

allow the recipient of the education to make up their own mind. If information is provided

in a one-sided, partisan manner, the communication or education could be considered

lobbying by the IRS. Public policy education may include the provision of nonpartisan

analysis, study or research on issues of concern to policy makers or the general public, as

well as providing examinations and discussion of broad social, economic, and similar

problems to policy makers or the general public. Nonprofits are able under the IRS code

to conduct an unlimited amount of public policy education with either policy makers or

the general public.

Legislative Issue Advocacy

Legislative issue advocacy refers to advocacy on behalf of a class of people, or

cause advocacy on an issue affecting the public. Specifically it refers to when an

organization takes a specific position on a piece of pending legislation, and

communicates that view to policy makers and asks policy makers to take this same

position. Legislative issue advocacy also includes grassroots lobbying, which involves

encouraging the general public to take a certain view and contact their legislators to

communicate that view on legislation. It is legal for 501(c)(3) nonprofit organizations to

lobby, as long as their efforts are not substantial.

Again, these three definitions describe a span of advocacy activities that attempt

to capture empowerment-based advocacy conducted by human service organizations. The

third category above, legislative issue advocacy, incorporates the legal definitions of

lobbying used by the I.R.S. While public policy education is fundamental in the

advocacy process, it is not fully captured in the traditional definitions of case and class

advocacy found in the social work literature, and is often overlooked in research on

advocacy.

In the next section, a brief review of relevant theories that inform this study is

presented, followed by a summation of the theoretical approach used in this study.

Theories Related to Topic

The landscape of social work theory in advocacy and community practice is wide

ranging and fragmented, and a prominent or unifying framework has yet to emerge,

although generally a systems theory approach prevails (Payne, 1997; MacNair, 1996a).

Several theoretical approaches are used in this study of advocacy. Primarily these are the

empowerment set of theories developed in the social work field, and the theories derived

from democratic and civil society theories from political science. After a brief review of

both groups of theories, the congruence between the definitions of advocacy used in this

study and Rothman‟s community practice model is examined. To conclude this section,

theory that combines these two approaches, as well as the insights gleaned from

Rothman‟s community practice model, is presented as the theoretical framework for this

study.

Democratic/Civil Society theories

Much of the rationale for nonprofit advocacy can be found in democratic political

theory. Dahl‟s (1998) development of the basic elements that make up a democracy is an

empirically derived theoretical framework that states there are five basic criteria needed

to create a democratic society. These are: (1) effective participation; (2) voting equality;

(3) control of the agenda; (4) inclusion of all adults; and (5) enlightened understanding.

Through examining the histories of democratic states, Dahl (1998, p.85-86) identified six

political “institutions” (or rather, principles and practices that must be institutionalized)

that provide what is needed to implement a large-scale democracy meeting the above five

criteria. These are: elected officials; free fair and frequent elections; freedom of

expression; alternate sources of information; associational autonomy; and inclusive

citizenship. These analyses underscore the importance of advocacy conducted in the

nonprofit sector as part of a healthy functioning democracy.

All of Dahl‟s political institutions listed above relate in some fashion to the

advocacy function of public policy education. For example, nonprofits are a recognized

alternative source of information that nurtures public debate, and nonprofits provide

avenues for political expression and opposition to government and market forces, and

offer opportunities for associational autonomy. Nonprofits that educate voters on the

political stances of candidates ensure that free and fair elections are made possible, and

less beholden to the influences of corporate donations or advertising. The first of Dahl‟s

institutions, freedom of expression, creates the rationale especially for political or

legislative issue advocacy to occur, and indeed the U.S. Constitution establishes this

principle as being integral to a democratic society. Inclusive citizenship is related closely

to case advocacy, since healthy, strong and empowered individuals are needed to build a

healthy democracy.

Dahl‟s populist approach to democratic theory has been criticized for its idealism

and perceived lack of attention to the political realities of how money and other powerful

interests hold sway in the U.S. policy making process (Berry, 2003). With recognizing

this theoretical limitation, Dahl‟s ideals remain ever more pressing as corporations

increase their power and openly subvert democracy (Phillips, 2002; Covington, 2001;

Paget, 1997; Bass et al, 2007). Dahl‟s theory supports the proposition laid out in this

study that various forms of advocacy are crucial for ensuring a healthy, civil society.

Empowerment-related theories

To engage in the process of empowerment requires that one assume that human

beings bear significant responsibility, in league with other human beings, for the

enhancement of their condition on earth. Empowerment thinking is deeply

antifatalistic. It insists that human agency is the central resource to call upon to

improve the human condition. It is not a search that people can conduct

alone...To the contrary, the pursuit of self-empowerment is a campaign in which

persons, families, and groups call upon the democratic state, the surrounding

neighborhood, and a variety of professions to assist them in marshaling the skills,

the knowledge, and the resources that they need if they are to enjoy the full fruits

of citizenship and of community and family membership while carrying out the

many obligations associated with those complex statuses. (Simon, 1994, 34)

The above quote serves to illuminate the close fit between empowerment theory

and nonprofit advocacy. As Simon articulates, without an empowered citizenry acting in

concert with each other, the attainment of social justice is impossible. Thus, advocacy

conducted under the aegis of nonprofit organizations, also known as the voluntary or

independent sector, obviously plays a critical role in the empowerment process of people

and communities.

Simon (1994) has traced the evolution of empowerment visions and practices back

to the very roots of the social work profession, and found social work practice and theory

divided into two main branches. The first view, embodied in the casework tradition, sees

empowerment as primarily an individual affair that is facilitated and nurtured through

professional intervention and other means. The other more political approach, rooted in

the settlement movement, views empowerment as a process of democratic political

mobilization set in motion to achieve social justice aims. More recent and fully

developed empowerment and advocacy theory does not hold to this divide, as it

recognizes that especially in working with oppressed communities, one must intervene on

all levels, personal, community and structural (Gutierrez et al., 2005). Consciousness

raising, education and participation in reflective discourse that culminates in collective

engagement and political action are used to facilitate empowerment and the ability to

achieve political power for social justice (Beck, 1983; Gutierrez, 1995; Carr, 2003). Most

scholars view empowerment as a process, rather than a static state of being (Staples,

1990; Gutierrez & Lewis, 1999).

Anti-discriminatory theories are closely related to empowerment, as they are

directly the result of the many social movements of the 80s and 90s that address

oppression on multiple socio-political and cultural levels (Payne, 1997).

Antidiscriminatory and anti-oppression theories provide another structural analysis

similar to radical theories, except that these theories emphasize the role of cultural

assumptions and biases over oppression based solely on class. These theories focus on

the process of how racism and sexism are socially constructed and perpetuated through

social ideologies and structures (Payne, 1997; Gutierrez and Lewis, 1994). Through

consciousness-raising, questioning basic assumptions, separatism and self-help strategies,

and an assumption of the need for client-directed advocacy, anti-discrimination theory has

the achievement of social justice as a central aim (Solomon, 1976; Payne, 1997; Gutierrez

and Lewis, 1999). The main contribution of anti-discriminatory/oppression theories to

social work community practice is that the multiple means of oppression are brought into

focus (Solomon, 1976). Advocacy in this context must include empowering clients to

advocate, and advocacy conducted on behalf of, instead of with or in support of, a group

is considered less valid than advocacy conducted by the oppressed themselves (Payne,

1997; Gutierrez and Lewis, 1999).

Advocacy in anti-oppression contexts must include empowering clients to

advocate, and also providing not just a voice for clients, but organizational supports so

that they may raise their own voices as well. When people advocate on their own behalf,

there is a validity that is missing from more professionalized advocacy that does not

include clients as citizens. Mondros and Wilson (1994) describe some of the means that

nonprofit organizations can employ to ensure access to participation and engagement in

social action. Structural elements include outreach programs, orientations, training,

buddy or mentoring systems, and the availability of active roles for volunteers and

consumers within the organizational setting. Specifically identifying and assisting

consumers in playing a lead role as an advocate is a key indicator of such an approach.

Process or cultural strategies include encouraging conscientiousness about not using

professional jargon, ensuring that decision making processes are as open as possible, and

employing two-way and horizontal communication exchanges.

An anti-discrimination/empowerment approach to advocacy questions the power

relationship of the professional to the client, and posits that clients should be viewed

instead as citizens or consumers rather than patients, as the former descriptors imply more

power and rights (Gutierrez and Lewis, 1999). This perspective is especially important

for work with poor and oppressed populations whose problems often stem primarily from

a lack of power (Solomon, 1976; Gutierrez and Lewis, 1999). Several proponents of

empowerment theory fold an anti-discriminatory perspective into their frameworks

(Payne, 1997; Gutierrez and Lewis, 1999), and thus offer a more comprehensive

theoretical base for advocacy theory than original iterations of empowerment theory.

This expanded theoretical framework that targets individuals and socio-political systems

also simultaneously positions the empowerment process itself as the central “healing

intervention” (Gutierrez and Lewis, 1999).

Following in this vein, Germain & Gitterman‟s (1996:29) Life Model of practice,

a well-established ecological social work theory, includes “diversity-sensitive,

empowering, and ethical practice” as one of the ten defining features of the model. They

state that “sensitivity to difference and empowerment practice are inseparably and

intimately connected to ethical practice,” and affirm the following skills as being critical

for social workers engaged in empowerment practice: “accepting people‟s definitions of

their life issues, identifying and building on existing strengths, engaging in a power

analysis of their situation, mobilizing resources and advocating with and on their behalf”

(Germain & Gitterman;1996:29). The combined anti-discriminatory/oppression

empowerment/advocacy theoretical perspective has gained ascendancy among social

workers especially committed to social justice for oppressed populations and the poor,

and thus it represents a strong theoretical base from which to approach social work

advocacy (Solomon, 1976; Beck, 1983; Staples, 1990; Simon, 1994; Rose, 1990;

Mondros & Wilson, 1994; Gutierrez, 1994; Gutierrez & Lewis, 1999).

Empowerment Advocacy Theory for This Study on Nonprofit Advocacy

The theoretical framework selected for this study is the empowerment/advocacy

theory, informed by democratic/civil society theory. Both empowerment/advocacy theory

and democratic/civil society theory recognize the critical role of advocacy by individuals

and organizations in building a vibrant democratic society. The democratic theory Dahl

(1998) lays out clearly articulates the need for advocacy at various levels of society, such

as the three definitions encompass. Empowered individuals, solid policyrelated

information, and the expression of political positions and views are all essential elements

to making democracy work, and organizations operating at various sociopolitical levels

are needed to mobilize and support these activities.

Study Purpose & Research Questions

Study purpose

The purpose of this study is to apply a new system of defining and describing

advocacy in a survey research field project, one that seeks to illuminate the values of

social work advocacy and the principles embodied in a democratic society. Through

examining advocacy within one type of organizational context, the nonprofit or

independent sector, the researcher endeavored to gain a better understanding of the

organizational characteristics that inhibit or support social work advocacy.

A brief summary of how the theoretical framework supports the central research

questions and hypotheses used in this study is provided below to set the stage for the

proposed research plan and methodology found in the next chapter. This summary is

followed by a list of the research questions and hypotheses that are the focus of this study.

The first two research questions guided the descriptive phase of this study, and the third

and fourth questions explored the theoretical claims presented earlier. To explore and

refine the theory used in this study, three hypotheses are tested; these are also explained

later in .this section.

To recap the essential elements of an empowerment advocacy framework, a brief

summary of four central empowerment principles follows that specifically relates to the

topic of social work advocacy conducted under the aegis of nonprofit organizations.

1) Advocacy conducted by nonprofit organizations on multiple socio-political levels

promotes both personal and collective empowerment and strengthens civil

society.

2) Client empowerment means that clients are regarded as citizens and consumers,

with the power for self-determination and the right to participate in shaping and

implementing interventions.

3) The social work value of working toward social justice must be preeminent in

empowerment-based social work practice and advocacy.

4) Diversity and anti-discrimination practices are both essential to, and indicative of,

the empowerment process.

The theoretical suppositions articulated in the end of the last section support the

use in this study of the definitions (described on pages 12-13) of case advocacy, public

policy education, and legislative issue advocacy in several ways. First, the theory

postulates that advocacy conducted by nonprofit organizations on multiple socio-political

levels promotes empowerment and strengthens civil society. Secondly, nonprofit human

service providers that embody the social work value of working toward social justice

must engage in empowerment-based social work advocacy that works on all levels of

society to effect social change that leads to social justice. The tripartite definition of

advocacy embodies the social work perspective and captures those activities that are

essential for true empowerment to emerge and organizes them into three categories that

span from the individual level to socio-political levels. Again, by using the three

definitions developed for this study that capture advocacy targeted at varying levels of

society, as well as the qualitatively different aims of the different types of advocacy, a

fuller spectrum of advocacy was studied in this research project.

Research questions

The first two research questions guided the descriptive phase of this study, which

explores the nature and mission of the organizations in the sample. The third question

explores other bivariate associations with the advocacy activity variables in order to

reveal the factors that support or result in an organization utilizing all three types of

advocacy. The last question involves the testing of three hypotheses in order to explore

how well the theory used in this study explains the findings on advocacy.

Research Question 1: Do the organizational characteristics of the nonprofits in

this sample vary by mission (i.e. anti-violence organizations and anti-poverty

organizations)?

Research Question 2: What kinds of advocacy are conducted by the nonprofits in

this sample, and do these activities vary by mission?

Research Question 3: How are advocacy activities associated with each other, and

with other organizational characteristics?

Research Question 4: How useful is an empowerment-based advocacy theoretical

framework in understanding the phenomena of nonprofit advocacy among human

services organizations?

This last research question used multivariate techniques to test several

assumptions derived from the theoretical framework to explain the patterns of advocacy

found among the nonprofits in this sample. The hypotheses in this study are based on

empowerment/advocacy theory, which asserts that empowerment-based advocacy

organizations (that employ all three types of advocacy) have consumers on the staff and

boards, a commitment to diversity, and more social workers as advocates.

Again, the theoretical premise articulated earlier proposes that when an

organization is found to have an advocacy pattern of an “Empowerment-Based Advocacy

Organization” (EBA), (i.e. it conducts all three types of advocacy) that organization will

be viewed as having an empowerment philosophy and practice approach. This phase

includes associational analyses that tested three hypotheses through appropriate analytic

techniques. The three hypotheses presented below are based on empowerment/advocacy

theory that posits that an organization that employs all three types of advocacy is an

empowerment-based organization that has a higher level of consumers on the staff and

boards, a greater commitment to diversity, and more social workers on staff.

The three hypotheses that were explored and tested are:

1) An organization with an advocacy pattern of an empowerment-based

advocacy organization (EBA) is more likely to have consumers on the board

and staff (empowerment of consumers) than non-EBA organizations.

2) An organization with an advocacy pattern of an empowerment-based

advocacy organization is more likely to have a greater commitment to

diversity (a more diverse staff & greater connections to diverse communities)

than non-EBA organizations.

3) An organization with an advocacy pattern of an empowerment-based

advocacy organization is more likely to have social workers on their staff

conducting advocacy than non-EBA organizations.

In the next chapter, the methodology and more detail about the variables and

definitions used in this study are presented.

CHAPTER II. METHODS

Introduction: Research Design and Chapter Organization

The research design is a naturalistic field study that is primarily descriptive and

exploratory in nature, and also develops several multivariate models to explore the data

and test hypotheses in order to refine empowerment theory. The study employs

quantitative methods to analyze data collected at one point in time from a mailed survey

to a purposive sample of nonprofit organizations with questions about their nonprofit

advocacy activities, as well as some of the characteristics of the organization.

A purposive sample is used to ensure a robust number of nonprofits that engage in

different types of advocacy activities. Three research questions guide the initial

univariate and bivariate phases of the data analysis. These three questions are:

Research Question 1:

Do the organizational characteristics of the nonprofits in this sample vary by

mission (i.e. anti-violence organizations and anti-poverty organizations)?

Research Question 2:

What kinds of advocacy are conducted by the nonprofits in this sample, and do

these activities vary by mission?

Research Question 3:

How are advocacy activities associated with each other, and with other

organizational characteristics?

In the multivariate phase of the analyses, nonparametric statistics are used to

explore the relationship of key variables hypothesized to be associated with nonprofit

advocacy activities. The hypotheses to be explored and tested are:

4) An organization with an empowerment-based advocacy pattern (EBA) is more

likely to have consumers on the board and staff (empowerment of consumers)

than non-EBA organizations.

5) An organization with an empowerment-based advocacy pattern is more likely

to have a greater commitment to diversity (a more diverse staff & greater

connections to diverse communities) than non-EBA organizations.

6) An organization with an empowerment-based advocacy pattern is more likely

to have social workers on their staff conducting advocacy than non-EBA

organizations.

This chapter starts with an explanation of the sample, and then covers the survey

design, survey administration, data collection, and then concludes with a brief discussion

of the main limits of the methodology used. A more detailed review of the limitations of

this study is found in the conclusion chapter that covers the implications of the findings.

Sample

The researcher was focused on examining the advocacy activities of the typical,

charitable nonprofit in which both micro and macro social work services are provided.

Since the number of local nonprofits that conduct all forms of advocacy can often be

quite low (Gronjberg & Childs, 2007), the researcher designed a purposive sampling plan

that would be more likely to identify human service or local nonprofits that engage in

advocacy, so as to create a larger pool of advocating nonprofits for the analyses. This

sample was made up of nonprofit organizations that are members, or are on the mailing

list, of three statewide advocacy organizations. All of the cases (in this study the

organizations are the unit of analysis) fall into one of two groups of organizations: either

anti-poverty, or anti-violence nonprofits.

Since these organizations are more likely to conduct advocacy by virtue of their

membership or affiliation in these statewide advocacy organizations, the researcher

presumed that this approach would provide a better opportunity to collect data on

nonprofit advocacy in both the anti-violence and anti-poverty fields. Indeed, as the

results in the next chapter show, this strategy did create a sample with an unusually high

percentage of nonprofits that advocate.

A total of 1462 surveys were mailed to each nonprofit that was identified from

mailing lists provided by three statewide nonprofit associations in New York State: the

New York State Coalition Against Sexual Assault (NYSCASA), the New York State

Coalition Against Domestic Violence (NYSCADV), and the Statewide Emergency

Network for Social and Economic Security (SENSES). The nonprofits in this sample are

defined as either anti-violence organizations (members from NYSCASA and

NYSCADV) or anti-poverty organizations (affiliates of SENSES). Of the 1462 mailed,

215 were mailed to anti-violence organizations, and the rest were mailed to anti-poverty

organizations. The anti-poverty nonprofit organizations that were either large hospitals,

institutions of higher education, or unions were removed from the original SENSES

mailing list. From the first mailing to all of the nonprofits in the study, 116 were either

returned in the mail, or determined to be duplicates, resulting in 1346 total valid surveys

mailed.

Of the 215 mailed to anti-violence organizations, 41 were either returned or

determined to be duplicates, and of the 1247 mailed to anti-poverty organizations, 75

were either returned or duplicates. The total sample (N=205) consists of organizations

that are from these two distinct sub-samples. Roughly two thirds are given the ascribed

attribute of being anti-poverty nonprofits, while one third are ascribed as being

antiviolence organizations. Further descriptions of the two sub-samples are found in the

next two sections.

Sub-sample of Anti-poverty Nonprofit Organizations

This sub-sample (N= 135) consisted of nonprofit organizations that are members

of, or are affiliated with, a major grassroots, anti-poverty statewide advocacy organization

active at the time of this survey, the Statewide Emergency Network for Social and

Economic Security (SENSES). These nonprofits work on variety of povertyrelated issues

such as welfare reform, job development and training, community economic

development, housing, and other related issues such as general human service

organizations.

This sub-sample was culled from a larger database of SENSES‟s mailing list by

removing all organizations that were clearly not human service nonprofits focused

primarily on either violence against women or relieving/ending poverty or other human

service needs. This included all government entities, educational institutions, unions, and

foundations. However, responses indicated that a few organizations that are not

nonprofits may have been left on the list, as two organizations checked “other” in the area

in which they were to identify the type of organization. In an effort to reach communities

of color and faith based organizations in this sample, religious institutions on the list were

included.

Sub-sample of Anti-violence Nonprofit Organizations

The anti-violence sub-sample (N = 70) of the total includes the local membership

organizations of two of the statewide associations that deal with violence against women,

children, and families. These organizations are rape crisis centers and domestic violence

programs; some are freestanding programs while others are subsumed under existing

victim witness programs, or are housed within a hospital. This sub-sample consists of all

of the programs that are members of either the New York State (NYS) Coalition Against

Sexual Assault, or the NYS Coalition Against Domestic Violence that provided services

to women and children who are survivors of sexual and domestic violence during 1999,

the year of data collection.

Response Rates

The response rate for the sample as a whole was 15%. The response rate for the

two samples was also calculated separately for each sub-sample, since different follow-up

methods were used for each one. The response rate for the anti-poverty organizations

was 12% (135 out of 1172), while the response rate for the anti-violence organizations

was 40% (70 out of 174). The response rate for the anti-violence organizations make this

sample fairly representative of these specific types of nonprofits, as given the typically

low rates of returns for surveys of nonprofits, this rate is moderately good. However, the

very low response rate of the anti-poverty organizations means that this sub-sample

cannot be said to be even moderately representative of anti-poverty organizations. In

addition, due to the purposive nature of the sampling plan, neither sub-sample can be said

to be completely representative of nonprofits in general, or of human service sub-sector

nonprofits. However, the data collected from these nonprofits connected to statewide

advocacy organizations can be seen as relevant to the study of nonprofits in similar

circumstances.

It is important to note that while the anti-violence organizations are quite

representative of nonprofits of this type (the reasons for this are discussed further in this

chapter when the variable is presented), the category of anti-poverty organizations is

much less representative of all nonprofits working on poverty-related issues. In addition,

SENSES functioned more as a loose network, rather than a membership-based

organization. For example, many of the nonprofits on their list simply attended one of

their policy information sessions they provide across the state and signed up on their

mailing list.

A brief review of response rates in similar studies will provide some comparisons.

One survey of New York State nonprofits achieved a good response rate of 63 % (Harlan

and Saidel, 1994), while others have hovered between 40- 50% (Weed, 1997; Hoefer,

1993). Using a more up to date IRS list of nonprofits that regularly file 990s, Chaves et.

al. (2004) had a response rate of 70 percent. In a similar study of nonprofit advocacy, the

Council of Community Services (1998) had a response rate of 17% in a statewide sample

of nonprofits.

Survey Design: Instruments & Measurements

Survey Design: Reliability and Validity of Instruments

The survey instrument (Appendix C) was compiled by the researcher with

extensive input from nonprofit practitioners, graduate Fellows on Women and Public

Policy at the Center for Women in Government, and four nonprofit statewide advocacy

organizations, including the three whose membership formed the sampling frame.

Several drafts of the survey instrument were circulated to these nonprofit statewide

advocacy organizations, and representatives from these organizations provided feedback

on questions in the survey. A version of this survey was then field tested with these four

organizations.

In addition, the researcher reviewed surveys developed by other nonprofit

scholars. Many of the items used in this study are taken or adapted from a survey

conducted by Dr. Marie Weil and others in North Carolina (1997), and some are from a

survey conducted by the NYS Council of Community Services (1998).

While the issue of the reliability and validity of the measures used is discussed in

more detail for each measure or variable, a few general comments are in order to set the

context for evaluating these issues.

Many questions on the survey ask about standard organizational characteristics

that nonprofits often must provide to the IRS, to funders, or for other purposes. One

could presume that such types of information as the year of incorporation, or amount of

annual expenditures, etc., have a relatively high degree of both reliability and face

validity, as they are straightforward reporting of numbers, or immutable facts, about the

organization. However, due to the complexity of some organizational structures,

reliability can sometimes be an issue. For example, an organization with several distinct

programs, or sites, may interpret the questionnaire to only be asking about one program

area or site. Attempts were made in the way that some questions were worded to avoid

this problem.

Several variables attempt to assess organizational knowledge, which can be a

tricky venture, as such knowledge may vary depending on who was assigned to complete

the questionnaire, and how diligent that person was in capturing the full resources and

capacity of the organization. The researcher tried to address this by asking the person

who received the survey to assign its completion to people in the organization that would

be knowledgeable about the organizations policy advocacy activities. In addition, tests of

this organizational knowledge were possible using other variables. However, in spite of

these tactics, a robust reliability of these variables is not assured.

One set of measures was used for which reliability and validity remains quite

elusive are the measures concerning Aconnection to diverse communities@ that were

developed for this survey, and to the researcher‟s knowledge, have not been used

elsewhere. These responses were subjectively determined by the respondents, since no

definitions were given. Obviously this means that the reliability and validity of these

measures is poor.

Again, further information about the reliability and validity of each specific

measure are covered in more detail in the next section in the presentation of the variables

used in this study.

Survey Design: Measurements

This section will include a complete description of the variables used in this study.

The numbers listed in the parens after the variable names (s) list the corresponding

numbers in the Variables chart (see Appendix ___), if the reader wishes to refer to the

chart for more information, such as syntax used for transformations of the original values,

etc.

Predictor Variables

Anti-violence/anti-poverty (3).

The variable, anti-violence/anti-poverty, is a measure of an organization‟s primary

mission. Values of this variable were assigned based on which sample the nonprofit

organization came from. If it came from either of the statewide organization‟s mailing

lists that focus on violence against women, then it was deemed to be an antiviolence

organization. Nonprofits from the other statewide mailing list were deemed to be anti-

poverty. The validity of this ascribed measure was tested using the next variable, issues

of interest.

Issues of Interest (4).

In order to have data to assess whether or not the ascribed mission (of being

focussed on either anti-poverty, or anti-violence work) was valid, the researcher asked

organizations to check any of the following issue areas of focus that their organization

may be concerned with:

a) general issue related to women, children & families, and

b) poverty/economic development, and

c) violence against women and children, and

d) other.

This measure helped the researcher to determine that the ascribed value of being

anti-violence was more valid than the ascribed measure of being anti-poverty, as 97% of

those organizations deemed to be anti-violence agreed that their organization was

concerned with violence against women and children, and thus are a much more

homogeneous sub-sample than the anti-poverty organizations. The reliability of this

variable is also stronger for the anti-violence sub-sample, as most anti-violence human

service organizations in the US that work on issues of sexual assault and domestic

violence in any significant fashion offer similar services in a similar fashion, and most do

belong to a statewide advocacy organization (New York State Coalition Against

Domestic Violence, n.d.; New York State Coalition Against Sexual Assault, n.d).

While the missions of organizations can change over time, this tends to be a slow

process.

Organizational Identity (5 a, b, and c).

Organizations were asked to check any up to nine service identities (organizations

could check more than one). The following three were selected as relevant to the purposes

of this study: grassroots/community organizing, service provider, and advocacy

organization.

As no definitions of these identities were provided, the validity and reliability of

these measures is somewhat in question. The validity and reliability of the service

provider measure is most likely better than for the grassroots/community organizing and

advocacy organization, as the common understanding of service provision is more

straightforward, while the latter two are more subject to interpretation.

Information source (6 a and b).

Availability of policy-related information to nonprofits could have an effect on

their advocacy activities, and so the survey asked what kinds of information sources

nonprofits used “to keep abreast of issues.” The organizations were simply asked to

check from a list of possible sources as to whether or not they used these sources. Two

measures were employed in this study to assess this variable. The first was whether or not

nonprofits used information from statewide advocacy colleagues (statewide advocacy

connection), and the second was whether they used information from a coalition

(coalition connection). The researcher attempted to construct more narrow measures in

order to improve the validity and reliability of these variables. If the organization used a

statewide nonprofit or coalition for policy-related information, then a more significant

connection is presumed to exist, as opposed to simply knowing of the organization, for

example. While this is a narrow conceptualization of a connection, the researcher views

the face validity of the construct as an improvement over simply asking the organization

if a connection exists. For the purposes of this study, the variable statewide advocacy

connection was used in the multivariate phase of the analysis.

Knowledge of advocacy rules.

As current research has shown (Berry, 2003), the degree of knowledge that a

nonprofit has about what advocacy they may legally conduct is likely to be a factor as to

how much, or what kinds of advocacy they undertake. The survey asked respondents two

questions to indicate the degree of their knowledge about nonprofit lobbying restrictions

under the IRS code by circling one of three answers. The first question asked how

knowledgeable the organization was about IRS lobbying restrictions for 501(c)3

organizations, with possible choices being: 1) Very knowledgeable; 2) Somewhat

knowledgeable, and 3) Not knowledgeable. From this measure, the research created the

variable, knowledge of IRS rules. In this new variable, the first two categories listed

previously were combined for one value, and the last category of not knowledgeable was

used for the second value. The researcher combined the two values of very and

somewhat knowledgeable, as the results indicated that there was not much difference

between these two values (since the results indicated that while many stated they were

knowledgeable, in fact, they were not). Making this variable dichotomous thus may have

improved its face validity.

The second question asked whether or not the organization had taken the helection

(an option under the IRS code that can allow a nonprofit to do more lobbying), again with

three possible answers of: yes, no, or not sure. To create the variable, takes h election, the

researcher combined the latter two answers into one value, with the remaining value

being yes. Again, making this variable dichotomous may improve its validity and

reliability, as the focus of this research is to find and examine those organizations that are

policy players, and its is safer to assume that those who take the helection are such

organizations. Both of these measures for assessing an organization‟s knowledge of

advocacy rules were used in the multivariate phase of the analysis.

Income source.

Organizations were asked to list the percentage of funding they received from

various sources using six categories: government, donations, foundation grants, fees,

United Way and other. For the multivariate analyses, the two categories with the largest

averages were used, income from government and income from donations. In addition to

being the two most important sources of funding for this sample, these also represent

diametrically opposed types of resource dependencies. Nonprofits that depend primarily

on income from government are more likely to be indebted to those interests, which are

typically driven by state and federal mandates. Those nonprofits that have significant

income from donors are more beholden to those influences, which may be of a more local

or parochial nature. Thus, focusing on these two sources would likely reveal more

interesting differences in the sample.

As discussed previously, most supervisory, and certainly policy-related staff are

very likely to be knowledgeable about the sources of funding of an organization, as

funders play such a large role in both what organizations do, and how they run their

programs. The categories are distinct, and yet not too numerous, so as to help boost the

reliability of the numbers. While the percentage points may vary somewhat, the

researcher assumes that in general, these numbers have a relatively high face validity, and

fairly good (plus or minus a few percentage points) reliability.

Capacity Variables; Age, Total Staff, Total Expenditures, and Size

Age.

To determine the organizations‟ age as of 1998 (the year the data were collected),

organizations were asked to indicate the year of incorporation of the organization, and the

program‟s inception (since many anti-violence programs were indeed programs of larger,

multi-purpose organizations). As the researcher examined programs housed within larger

nonprofit organizations (such as a rape crisis service within a larger hospital or

multiservice agency), the survey requested the year of the program‟s inception as well.

The researcher used the program year to determine age in the case of anti-violence

organizations that indicated a program inception year differing from the larger parent

organization. This was done not only because most of these gave complete answers to

other questions that were connected to their program, not to their larger parent

organization, but to capture the unique nature of this kind of social movement

organization. The researcher used the organizational inception year for the anti-poverty

programs, as only a handful noted a program year of inception, and it seemed the

antipoverty organizations gave information for the entire organization.

Total staff and total expenditures.

The total staff simply totaled the number of both male and female staff members;

this information was provided by the organization in the survey chart on the survey

instrument. The total expenditures was also provided by the organization.

Of the two measures used above, the staff variable was likely to have less

reliability, as part time staff may be missed, and staff turnover can be great; this would be

especially true for mid size to very large organizations. Also, the directions did not

specify whether or not to count staff by full time equivalencies or not (the intention was

to not do so, but some may have interpreted it this way). The total expenditures is a

figure that must be tracked for any organization, and thus is a relatively valid and

accurate assessment of an organization‟s gross capacity.

Size.

In order to try and compensate for the missing information on the two

organizational capacity measures, number of staff and expenditures, a new size variable

was created using these two variables. Of the two capacity variables presented above,

staff and expenditures, the research selected the expenditure variable as the base variable

to use in creating the size variable, as more organization answered that question, and it

was deemed to be a more reliable measure. The staff variable was used to plug in

missing values for those organizations that did not list expenditures, but did list staff, as

they were highly correlated (r= .776; p<.01)

The size variable was created using total expenditures to create three categories of

small, medium and large organizations. The values for the size variable were created by

the researcher, using data from the nonprofit sector at large in addition to data from this

sample. While normally a researcher would want to set values that would evenly

distribute the number of organizations among the values, this approach does not reflect

the reality of the typical distribution in the nonprofit sector in which the numbers of small

organizations are very high, and the numbers of very large organizations are lower. Thus,

fewer numbers of organizations are captured in the higher values of the size variable.

The upper value for the first value of one was set just above the median value for

the total sample, which followed a typical categorization scheme used by leading

researchers in the field at the Urban Institute (Boris, ). The second category also was

drawn from the Urban Institute schema. The large category of over 1 million dollars in

expenditures captures a wide range of expenditures, from those with a million to one with

almost 72 million.

Since there were a few cases missing the expenditure data, the mean of the total

staff variable associated with the different size categories was used to assign a size value

for those cases. However, the SD and medians of total expenditures differed substantially

between antiviolence and anti-poverty organizations, as did the correlations between the

total staff and total expenditures differed for anti-violence organizations (R=.757,

p=.000), and anti-poverty organizations (R=.234, p=.016). Thus, separate tables were

used for assigning the values for the size variable for these two types of organizations. A

table was created for each type (anti-poverty and anti-violence) that showed the mean

number of staff for each size value. Organizations with surveys missing the expenditure

data were then assigned a size value based on the number of staff and the type of

organization (the researcher took half of the difference in the means to identify the

appropriate category).

The face validity of the staff and board variables are strong, with the exception

that whether or not a staff was part time or full time was not collected. However, since

this data ended up being grouped into three broad categories, this difference would have

less of a deleterious impact on reliability. While certainly the number of staff an agency

has can be miscounted, it is not likely that this error would be a large one.

In constructing the size variable, the researcher aimed to capture the gross

organizational capacity for policy-related advocacy by dividing organizations into three

categories. The researcher tried to boost the reliability of this variable by making the

categories small and broad. However, this effort may have been compromised by a loss

of validity, as the distinction between small or medium, or medium and large

organizations may have no real meaning in terms of the organization‟s true capacity for

advocacy, at least.

Diversity and Connection to Ethnic Communities

For the purposes of this study, six variables were used to measure an

organization‟s diversity, which for the purpose of this study, focus on ethnicity. The first

four variables, percent staff ethnicity, percent board ethnicity, organizational ethnicity,

and categories of organizational diversity, are measures that describe ethnic makeup of an

organization. The fifth measured the organization‟s connection to diverse ethnic

communities, and the sixth measured the organization‟s connection to African-American

and Hispanic communities only.

Percent staff and board ethnicity (13a and b), and organizational ethnicity (13c).

The first variable, percent staff ethnicity, was the percent of staff who were

ethnically diverse. This was accomplished by totaling the number of ethnically diverse

staff (including African-America, Asian, Hispanic and Native American populations), and

dividing it by the total number of staff.

The same data transformation as described above was followed as well for the

second variable, percent board ethnicity. A third variable, organizational ethnicity, was

created to help deal with the many missing values for either the board or staff variables.

The staff value was chosen to represent the percentage of organizational ethnicity as it

had fewer missing cases. As the board and staff variables were correlated (R=.59, p<.01),

this allowed the researcher to use the board variable value when the staff variable was

missing to create the third variable. The reliability and validity of the staff variable was

discussed previously. The reliability and validity of the number of board members and

their composition is likely to be relatively good, due to the clear face validity of the

measure, and as nonprofits are legally required to keep this information up to date, and it

is public information. Of course, some fluctuation is inevitable as boards can have as

much turnover as staff do in a nonprofit, so such information can be become out of date.

In addition, racial and ethnic identity is not always clear, or clearly self identified by

people. Both of these realities means that the reliability and validity of this measure,

while good, are by no means perfect.

The fourth variable, categories of organizational ethnicity, divided the percentage

values of the variable organizational diversity into five categories, 0=0%; 1=1-20%;

2=21-50%; 3=51-99%; 4=100%. Of course, a significant amount of validity is lost

when transforming a continuous variable into a categorical one, as one could argue that

the difference between category one and two is meaningless. However, one could

certainly argue that the difference between one and three, or one and five, does indeed

capture a meaningful difference. Transforming the fifth variable, categories of

organizational ethnicity, into five categories helps to address this loss of data.

Connection to Ethnic Communities (15 a and b).

Another measure, connection to diverse communities, was used to assess the

orientation of the organization toward issues of diversity. Organizations were asked to

indicate to what degree they were connected to the organizations and communities of four

different ethnic communities, African- American, Hispanic, Asian and Native American

using a four point likert scale. A value of one was assigned to the selection of not at all

connected, while a four was assigned to very connected. These values were summed for

each organization, creating a variable that ranged in value from 4 to 16. A second

variable was created that examined only the first two items of connection to African-

American and Hispanic communities and organizations (and the value of this variable

thus ran from 2-8). Both of these were treated as continuous variables in subsequent

analyses. As mentioned in the first part of this section, this ordinal variable was not field

tested, and thus no assessment of its reliability can be made.

Organizations with social workers (16) & consumers (17)

Two other predictor variables were constructed in order to test the hypotheses of

this study. The first, organizations with social workers, was constructed as a dichotomous

variable with a yes value meaning that organizations had at least one degreed social

worker (either at the bachelor or masters level) working as an advocate in the

organization, and the no value meant that no degreed social workers were advocates.

Another variable hypothesized to affect advocacy activities is the degree to which

consumers are involved in the organization. The survey asked about the number of

consumers serving on the board or staff. The researcher then created a dichotomous

variable, organizations with consumers, with two values: yes, consumers were on the staff

or board, or no, no consumers were on the board or staff.

While the reliability of these measures may be sufficient given the straightforward

nature of these measures, the validity of the constructs is certainly more debatable.

Having social workers as advocates does not necessarily mean that the organization will

adhere to embodying the social work professional ethics of engagement in social change

for social justice. Nor does having consumers present on the staff or board guarantee that

they will have an influence or voice in the advocacy process of an organization. In

addition, consumers may not be willing to self-identify, while many are not willing to do

so, and yet their sensibilities may have some influence. However, the researcher decided

to use these as indicators as they are more concrete indications of whether or not these

sensibilities and perspectives may be present in an organization‟s culture.

Dependent Variables

In this study the local nonprofit organization was the unit of analysis and the three

primary dependent variables are whether the nonprofit engaged in three types of

advocacy: case advocacy, public policy education, or legislative issue advocacy. Non

profits were asked to indicate on a survey whether or not they conducted these activities

at all during the previous year.

These three dependent variables were used to construct eight exhaustive

categories outlined below that present all the possible combinations of advocacy activities

an organization may conduct. The final dichotomous dependent variable used in this

study was constructed by creating one category, Empowerment Based advocacy

organizations, of those organizations that did all three advocacy activities (#3 in the

typology below), and then grouping all of the remaining categories into the other

category, Non-Empowerment-based advocacy organizations.

Case

Advocacy

Public

Policy

Education

Legislative

Issue Advocacy

Type of Organization

yes

1. Engaged Service Provider Organization

yes

2. Pragmatic Advocacy Organization

yes

3. Empowerment-based Advocacy

Organization

yes

4. Social Change Activist Organization

yes

5. Political Activist Organization

yes

6. Social Educator Organization

7. Disengaged Service Provider Organization

yes

8. Social Capital Organization

The reliability regarding advocacy data may be of some concern. For example,

two organizations mistakenly completed the survey twice, and the data concerning the

amount of their advocacy activities differed on the two surveys. However, whether or not

the organization conducted the types of advocacy remained the same. Thus the measure

of whether or not advocacy is conducted is likely to be reliable, considering that it is a

simple, gross measure with a value of yes or no.

To help boost the reliability and validity of these main dependent variables, a

page was inserted with the survey that provided a brief paragraph defining each term.

These definitions were reviewed by nonprofit practitioners, graduate Fellows on Women

and Public Policy at the Center for Women in Government, and NEI Consortium

members in an attempt to increase the clarity and face validity of the measures.

The reliability and validity of the variable case advocacy is likely to be quite

good, as this is a common practice in the human service field. Public policy education

was also quite clearly defined, and has the benefit of having a clear descriptor for a title,

which may increase its face validity. Due to the political nature of legislative issue

advocacy, a term that was defined as being equivalent to the legal definition of lobbying,

both the reliability and validity of this variable are more questionable. In fact, several

follow up questions were asked about lobbying, and many organizations said that while

they did do these lobbying activities, they did not engage in legislative issue advocacy.

The fear and lack of clarity surrounding the terms used, and the fear about nonprofits

engaging in these activities, means that this measure is likely to be less valid or reliable

than the other dependent variables.

Survey Administration & Procedures

Survey administration

Surveys were mailed out in hard copy to organizations, along with a signed cover

letter from the researcher with information about confidentiality. Organizations were

asked to fill the survey out by hand, and mail it back (SASEs were included).

Three weeks later, a post card was mailed to the sample to remind them to

complete and return the survey. Three weeks after that, a second survey was sent to the

organizations in the anti-violence sub-sample. Because cost was an issue, this was not

done with the larger anti-poverty sub-sample. Follow up calls were conducted to as many

non-respondents as possible in both sub-samples to increase response rates, and

additional surveys were faxed to contacts who faxed them back again. In several

instances, the researcher or research assistants filled out a survey over the phone, when

the appropriate staff person was located and willing to provide the information.

Human Subjects Review

The researcher reviewed the research procedures with the University at Albany

Office of Research, and received permission to proceed with the research project.

Standard survey procedures were followed in conducting the survey to ensure that

confidentiality and ethical considerations were assured. A cover letter explaining the

study and confidentiality protections was mailed to each participant. Each organization

was given an identification number used for coding and administrative purposes. The

code sheet with identifying information was destroyed after the data analysis, so that the

responses of the participants are anonymous and completely confidential. Additionally,

data were only presented in aggregate form. Surveys were handled only by the researcher

or by trained staff or graduate assistants.

Data Collection & Analysis

Data Collection

The data were collected, coded and handled by the researcher and two graduate

assistants, with regular meetings to discuss any interpretation issues and input procedures.

The researcher cleaned the data, and conducted the analyses. The data are currently in

SPSS, as well as JMP.

Data Analysis

The data analyses included three general steps. First, univariate counts, means,

and standard deviations were conducted for each variable, and also by the mission of each

organization (whether or not the organization was anti-poverty or anti-violence). These

latter bivariate tests allowed the first two research questions (that asked whether mission

makes a difference) to be addressed at the same time as the standard univariate

description of the sample and measured is presented.

Secondly, to answer the third research question regarding what kinds of

organizational characteristics were associated with certain types of advocacy activities,

bivariate analyses were conducted that explored significant relationships of those

variables that were correlated with any of the four dependent variables that measured an

organization‟s advocacy activities. Depending on the level of measurement, appropriate

tests were conducted to measure the theorized relationships between the independent and

dependent variables. For the bivariate tests of categorical variables with dichotomous

values, the Pearson chi-squares test was used. For variables with continuous values, a

one-way anova was used, as the independent variable was dichotomous.

The third phase involved the testing of hypotheses derived from the advocacy

empowerment theory presented in the previous section. Two steps were taken in the

multivariate analyses phase of the study.

First, due to the low number of cases and relatively large number of variables, a

nonparametric technique called classification trees was used to pare down the number of

variables for the subsequent logistic regression models. The rationale for use of this

method, methodology and details of this analytic procedure are presented along with the

results, so as to make this relatively new technique more comprehensible to the reader.

The following variables were selected for the final multivariate phase of the

analysis as predictor variables: Mission (anti-violence or anti-poverty), knowledge of

IRS rules, age, size, percent income from government, percent income from donations,

statewide advocacy connection, organizations with consumers, organizations with social

workers, percent organizational ethnicity, and connection to African-American and

Hispanic communities. All of these variables were used in the classification trees

analyses to try to predict the dependent variables: case advocacy, public policy education,

lobbying, and Empowerment-based advocacy organization (those that did all three kinds

of advocacy.

Secondly, using the variables identified by the classification trees analysis, four

separate logistic regressions were conducted with the dependent variable empowerment

based advocacy organization (EBA). The result of the logistic regression on EBAs will

allow the discovery of those factors that predict the probability of an organization being

an Empowerment Based Advocacy organization, which is the central aim of this study.

Conclusion: Limitations of Methodology

The methodology used in this study has several significant limitations. The most

important of these was the sampling process used. The use of a purposive sample means

that any results obtained are not representative of the nonprofit sector. Neither can the

sample be said to be completely representative of human service agencies, as a few of the

organizations include economic development, food pantries, or other types of general

advocacy organizations. Even if the sample was a random sample, neither of the

response rates of the two sub-samples were robust enough to warrant generalizability of

the results. However, these limitations of the sampling method used are less important,

because the main purpose of this study was more exploratory than explanatory in nature.

The testing of the conceptualization of variables, coupled with theoretical model building,

were a primary focus. As mentioned previously, this method ensured that a larger number

of advocacy organizations would be in the sample. However, as most theorists have

advanced, citizen-based, grassroots advocacy is indeed most likely to occur in the

nonprofit context, and so a focus on this type of organization was warranted.

As mentioned in the review of variables, several of the ordinal variables used

were not tested for reliability or validity. This obviously makes their reliability and

validity questionable.

Another important limit involved the theory and definitions used in the study. In

particular, the exclusion of community organizing as an advocacy activity is arguably a

significant limit, as the mobilizing function held to be so important in civil society is less

visible as a result (Saidel, 2002). Instead, the researcher decided to break up some of the

most common community organizing functions into three definitions used to reflect

reality of most human service organizations. While community organizing as a distinct,

major advocacy strategy was used more often in the 1960‟s, the great majority of human

service organizations do not conduct community organizing per se any longer. In

addition, the term connotes a kind of outsider political action that the researcher feared

would put off the respondents. Thus, the researcher believed that a focus on the more

widespread use of case advocacy, public policy education and lobbying was more useful.

A more in-depth discussion of the utility of using these definitions is one of the

main foci of the discussion chapter. The next chapter will present the results of this study.

CHAPTER III. RESULTS

The results presented in this chapter use a variety of quantitative analyses. This

chapter is organized by the four research questions used in this study. The first two

questions explore the importance of an organization‟s mission and its advocacy activities,

and the third examines those organizational characteristics associated with the

organization‟s advocacy activities. The fourth research question explores the theories

guiding this study by testing three hypotheses.

Restating of Research Questions and Chapter Organization

In the sections addressing the first two research questions, a description of the

sample is provided with all variables used in the study presented both in total and by the

two sub-samples with different missions: anti-violence organizations and anti-poverty

organizations. Further bivariate analyses are conducted to answer the third question, and

multivariate analyses address the last research question and the hypotheses that explore

the theoretical framework used in the study. Again, the research questions, which

delineate the four main sections on this chapter, are:

Section 1: Research Question 1 on organizational characteristics and mission

Do the organizational characteristics of the nonprofits in this sample vary by

mission (i.e. by anti-violence organizations and anti-poverty missions)? In this section,

univariate analyses and bivariate analyses by mission are presented for the variables in

the study to address this first question while providing descriptive data on the variables

used in this study.

Section 2: Research Question 2 on advocacy activities and mission

What kinds of advocacy are conducted by the nonprofits in this sample, and do

these activities vary by mission? Univariate and bivariate analyses by mission are

presented in this section on the advocacy variables.

Section 3: Research Question 3 on advocacy activities and other organizational

characteristics

In this section, further bivariate analyses answer a third research question: How

are advocacy activities associated with each other, and with other organizational

characteristics?

Section 4: How useful is an empowerment-based advocacy theoretical framework

in understanding the phenomena of nonprofit advocacy among human services organizations?

In this fourth section of this chapter, multivariate analyses using both

classification trees and logistic regression explore the study‟s conceptualization of

empowerment theory by addressing the three hypotheses described earlier. The three

hypotheses are:

7) An empowerment-based advocacy organization (EBA) is more likely to have

consumers on the board and staff (empowerment of consumers) than nonEBA

organizations.

8) An empowerment-based advocacy organization is more likely to have a

greater commitment to diversity (a more diverse staff & greater connections to

diverse communities) than non EBA organizations.

9) An empowerment-based advocacy organization is more likely to have social

workers on their staff conducting advocacy than non EBA organizations.

Section 1: Research Question 1 on organizational characteristics and mission

As described in the method section, the sample for this study is a purposive

sample of the organizations on the mailing lists of three nonprofit statewide advocacy

organizations. One of the sub-samples is created from the mailing lists of two

organizations focused on issues of violence against women. The other sub-sample is

from the mailing list of a loose network focused on anti-poverty issues. These two

subsamples are identified by the variable name: mission. In this study mission refers to

whether or not the organization, or in some instances, the program, is primarily an

antiviolence or an anti-poverty nonprofit and will serve as a grouping variable in the

following analyses. Mission, or service area, is likely to have a significant influence on

the kinds of advocacy conducted, as well as on other variables.

Research question 1.

The following text and tables provide univariate data on the variables of interest,

as well as by mission, in order to address the first research question of this chapter: Does

mission make a difference?

This section examines whether mission is associated with such organizational

demographic characteristics as: tax status of the organization, type of organization, main

mission of the organization, issues of interest, organizational identity, policy information

sources, knowledge of advocacy rules, income sources, age of the organization, total staff,

total expenditures, and size; as well as variables hypothesized to affect advocacy

activities, such as organizational ethnicity (focusing on the ethnicity of the staff and

board), connections to ethnic communities, whether the organization has social workers

as advocates, and involvement of consumers on the staff and board. Four dependent

variables include whether or not an organization conducts case advocacy, public policy

education, legislative issue advocacy (lobbying), or all three, making it an

empowermentbased advocacy organization.

Tax status

The great majority of nonprofit organizations in the sample, 185 (94%), indicated

on the survey that they are 501(c)(3) nonprofits. Four organizations (2%) are

incorporated as 501(c)(4)s, five (2.5%) are religious institutions providing services, and

three (1.5%) are listed as other. So few organizations were not 501c3 nonprofits that chi

square analyses could not be performed to see if there was any difference by mission.

Type of organization and mission

The sample consists of nonprofits culled from the mailing or membership lists of

three statewide nonprofit advocacy organizations, two of which are focused on

antiviolence work as their mission, and one that is concerned with anti-poverty work in

general. Of the 205 organizations that responded to the survey, 70 (34%) of the sample,

came from the two antiviolence statewide advocacy organizations; one that works on

domestic violence issues, and the other on sexual assault. These organizations address

issues related to violence against women as their central mission, and are referred to

hereafter as anti-violence organizations. The anti-violence organizations consist of three

types: rape crisis centers (18%), domestic violence programs (13%), and both rape crisis

and domestic violence services in one program or agency (3%). The other sub-sample

was the mailing list of a statewide nonprofit advocacy organization addressing issues

related to poverty, such as welfare programs, food programs, and to a lesser extent,

community development and housing programs. The respondents in this sub-sample, 135

(66%) of the total sample, are anti-poverty organizations that work on a wide variety of

social issues and services due to the broad nature of the topic area.

Issues of Interest

Another approach to assessing the mission of an organization was to ask them to

identify the issues of interest to the organization. Table 1 compares the organizations‟

mission, as determined by their membership or affiliation with a statewide organization,

to issues of interest that the organization identified as being important in the survey.

Table 1. Comparing Mission (Anti-violence or Anti-poverty) by Issues of Interest

Issues of interest

N and % of Organizations Interested in Issue Area

Anti-violence

N=70

N % Yes

Anti-poverty

N=131

N % Yes

Total

N=201

N % Yes

Pearson Chi

Square

df=1

Concerned w/ general issues

of women, children and

families

50 71 %

66 50%

116 58%

8.281**

Concerned w/poverty &

economic development

35 50%

89 68%

124 62%

6.212*

Concerned with violence

against women & children

68 97%

38 29%

106 53%

84.973**

Other

14 20%

81 62%

95 47%

32.030**

*p<.05; **p<.01 [Note: Organizations could select more than one issue area of interest.]

As would be expected, anti-violence organizations differ notably from antipoverty

organizations in terms of issue areas of interest, and all of these differences are

statistically significant at the .05 level using chi-square analyses. Compared to

antipoverty organizations, more anti-violence organizations indicated that the main focus

of their work was issues of violence against women and children (97% vs 29%), and

general issues of women, children and families (71% vs 50%). More anti-poverty

organizations, in contrast, identified poverty (68% vs 50%), and other issues of interest

(62% vs 20%) as being of interest. The anti-poverty organizations display less

consonance between their ascribed mission and identified issues of interest, as indicated

by the lower number that selected poverty issues as their area of interest, and a higher

percentage of “other issues” selected. The statistical significance and strength of the

above differences shows that mission is related to interest in relevant policy issues, and

that the ascribed construct of mission does fit the organizations.

Organizational Identity

Respondents were asked to indicate whether any of the following three phrases

described their organizations: Grassroots/community organizing; Service provider; and

Advocacy organization. The results are found in Table 2 below.

Table 2. Comparing Mission by Organizational identity

Identity

Anti-violence

n=70

Anti-poverty

n=130

Total n=200

Pearson

Chi

Square

df=1

YES

N %

YES

N %

YES

N %

df=1

Grassroots/community

organizing

29 41

57 44

86 43

.109

Service provider

66 94

105 81

171 86

**6.705

Advocacy organization

49 70

58 45

107 54

**11.785

*p<.05; **p<.01 (Note: organizations could select more than one identity)

Both anti-violence and anti-poverty organizations were about as likely to identify as

grassroots organizations (anti-violence, yes = 41%; anti-poverty, yes = 44%). While both

had a majority of organizations identifying as service provider organizations (antiviolence,

yes = 94%; anti-poverty, yes = 81%), the larger proportion of anti-violence organizations

selecting service provider as a descriptor than anti-poverty organizations was found to be a

statistically significant difference. Anti-violence organizations were more likely to identify

as an advocacy organization (70% vs 45%), and this difference was significant at the .01

level.

Information sources

Because connection to or membership in a statewide association that conducts

advocacy may play a role in whether or not a nonprofit organization conducts advocacy,

the survey asked whether the organization used information from statewide advocacy

colleagues and coalitions. Table 3 below displays the results of this variable.

Table 3. Comparing Mission by Information Source

Anti-violence

N=70

Yes

N %

Anti-poverty

N=134

Yes

N %

Total

N=204

Yes

N %

Pearson

Chi

Square

df=1

Used information from

statewide advocacy

colleagues

51 73%

108

81%

159

78%

1.602

Used information from a

coalition

67 96%

116 87%

183

90%

4.166*

*p< .05; **p< .01

As the results above show, the great majority of both anti-violence and

antipoverty organizations appear to depend on coalitions and statewide advocacy

colleagues for their policy information, with anti-violence organizations using

information from a coalition more often than anti-poverty organizations (96% vs. 87%,

respectively; p< .05). These two variables attempted to distinguish between belonging to

a coalition, which tend to be more formal associations or organizations (such as was the

case for the antiviolence organizations in this sample), and more diffuse, informal policy

networks.

Knowledge of Advocacy

Another factor that may play a role in what kinds of advocacy a nonprofit elects to

conduct is the degree of knowledge about nonprofit lobby laws, especially in terms of

what is allowed under federal law. The central rules that define nonprofit incorporation

are found in the 501(c) section of the IRS code, and this section lays out the parameters of

permissible activities for all nonprofits incorporating under the IRS code. Two survey

questions elicited responses designed to test organizational knowledge in this arena.

As the results below show, anti-violence organizations indicated less knowledge

of IRS lobbying laws with 41% stating they were not knowledgeable about the law, while

only 12% of the anti-poverty organizations were not knowledgeable (p<.01). However,

organizations that were knowledgeable may not have been, as 46% of those that were

very knowledgeable were not sure if their organization took the “h” election, which is a

key piece of information any nonprofit that lobbies should understand.

Table 4. Comparing Mission by Knowledge of Advocacy Rules

Knowledge of IRS

rules N=184

Anti-violence

Anti-poverty

Total

Pearson

Chi

Square

N %

df=1

Very/somewhat

knowledgeable

39 60

104 88

143 78

**20.62

Not

Knowledgeable

27 41

14 12

41 22

Takes “h” N=167

No or not sure

56 97

104 96

160 96

.122

Yes

2 3

5 5

7 4

*p<.05; **p<.01

Income Source

Source of income is another potentially important variable in terms of an

organization‟s ability to conduct advocacy. The chart below shows the average percent

of income by six categories of potential sources; one-way ANOVA analyses were used to

determine the F value.

Table 5. Mission by Income Source

% Income from Government

Mean

Median

Range

AV; N= 59

100

3.1

AP; N= 122

100

Total N=181

100

% Income from Donations

Mean

Median

Range

AV; N=59

10.7*

AP; N= 123

Total N=182

% Income from Foundation grants

Mean

Median

Range

AV; N=59

AP; N= 124

Total N=183

% Income from Fees

Mean

Median

Range

AV; N=59

AP; N= 122

100

Total N=181

100

% Income from United Way

Mean

Median

Range

AV; N= 59

AP; N= 123

Total N=182

% Income from Other

Mean

Median

Range

AV; N= 59

AP; N= 123

Total N= 182

*p<.05; **p<.01 AV= Anti-violence organizations; AP = anti-poverty organizations

Government provided the greatest amount of funds for the organizations in the

sample, providing a mean of 57% and a median of 70%. Both anti-violence and

antipoverty organizations showed that on average the majority of their support came from

government funding (63% and 54%, respectively), with ten organizations (6% of the total

sample) indicating that 100% of their funds came from government. Overall, only 12%,

or 18 organizations of the sample received no funding from government (5% for

antiviolence organizations, and 15% for anti-poverty organizations). Of the 31

organizations that received 10% or less of their funding from government, 84% were anti-

poverty, and 16% were anti-violence. On the other end of the spectrum, 27 organizations

(21% of the sample) received 90% or more of their funds from government (19% for anti-

violence organizations, and 22% for anti-poverty organizations). The second largest

category of support was income from donations and memberships, which was also the

only source of income where the difference between anti-violence and anti-poverty (7%

vs 19%) organizations was significant (p<.05). Neither organization type derived much

income from United Way, and both received roughly 10-11% of their funding from grants

and fees.

Advocacy Capacity: Age, Number of Staff, Total Expenditures and Size

A number of organizational characteristics have been found to be associated with

an organization‟s capacity for advocacy activities. Three of these characteristics, namely

age, number of staff, and expenditures, are reviewed below in Table 6. Again, one-way

ANOVAs were conducted to ascertain statistical significance.

Table 6. Mission by Age, Staff, and Total Expenditures

Age of organization (in

years)

Mean

Median

Range

Anti-violence; N= 68

117

4.345*

Anti-poverty; N= 131

197

Total N= 199

197

Total staff

Mean

Median

Range

Anti-violence; N= 59

107

1.455

Anti-poverty; N= 111

505

Total N= 170

505

Total Expenditures

Mean

Median

Range

Anti-violence; N= 56

$579,248

$375,000

$3,154,800

612,495

2.866

Anti-poverty; N= 122

$2,769,972

$400,000

$71,991,000

9,567,454

Total N= 178

$2,080,755

$400,000

$71,991,000

8,013,302

*p< .05; **p< .01

A large percent of the sample were younger organizations, with the median age

being 17 and 75% of the sample were no more than 26 years old. With the oldest coming

in at 198 years old, and the youngest at 1 year old, the sample shows a great deal of

variability. On average, the anti-violence programs were younger than the anti-poverty

organizations, with means of 19 years vs 28 years, respectively (p< .05). This difference

is not surprising given that the field of practice of violence against women is relatively

new, compared to the older field of anti-poverty work. The difference in mean ages

between the anti-violence and anti-poverty organizations was the only measure of the

three organizational capacity measures above found to be significant.

Information on expenditures was gathered in two places on the survey form. If

the program being surveyed was part of a larger nonprofit (as anticipated for a significant

number of anti-violence programs), then we also asked for both their age and program

revenue separate from the total revenue and age of the organization. Indeed, many more

anti-violence organizations indicated a program year, while the great majority of

antipoverty organizations indicated simply an organization year. This means that probably

more of the anti-violence organizations were actually programs housed within large

multi-service nonprofits rather than separate nonprofit organizations.

Salamon (1995) found that interesting age differences were found in different

types of organizations based on their generational cohort (for example, many of those

organizations formed in the 60s did much more advocacy). In order to capture the unique

nature of the cohort of anti-violence organizations (most formed in the late 70s through

the 80s as a result of the women‟s movement), the anti-violence age and expenditure data

was treated differently from the anti-poverty organizations. For the anti-violence

organizations, if both program and organization data were given for age and expenditures,

then the program level figures were used for the analysis; if only organization level data

were given for these two variables, then that data was used. For the anti-poverty

organizations, the organization year and expenditure data were used.

Extreme variability was evident among the two sub-samples in number of staff

and expenditures, as indicated by the large standard deviations in both measures. The

anti-poverty organizations had a larger degree of variability in expenditures with a

standard deviation of $9,567,454 compared to $612,494 of the anti-poverty organizations.

Interestingly, although the median number of staff in an anti-violence program was larger

that the median number of staff in an anti-poverty program (12 vs. 9), the median

expenditures for anti-poverty programs were higher ($400,000 vs $375,000). This may

be due to the capital intensive nature of some anti-poverty services, such as housing or

community development. Both sub-samples had quite a few very large multi-service

agencies, creating a number of outliers in regards to expenditures and number of staff.

As mentioned in Chapter 2, the researcher created a new size variable using the

total expenditures and total number of staff. This raised the number of cases with data on

organizational capacity to 193. As in the nonprofit sector at large, Table 7 below shows

that the majority of nonprofits in this sample were small in size (58%). Analyses

indicated no significant differences between anti-poverty and anti-violence organizations

in terms of size, although anti-poverty organizations did have more large organizations

(25% vs 17%).

Table 7. Mission by Size

Size

Anti-Violence

N %

Anti-Poverty

N %

Total

N %

Pearson Chi

Square df=2

Small

(under $499,000)

39 61

73 57

112 58%

1.44

Medium

($500,000-$999,999)

14 22

24 19

38 20%

Large

(Over one million)

11 17

32 25

43 22%

Totals

64 100%

129 100%

193 100%

*p< .05; **p< .01

Diversity and Connection to Ethnic Communities

Assessing the impact of an organization‟s diversity on its advocacy activities was

part of the fourth research question that explores the anti-discrimination aspects of

empowerment theory used in this study. In this study, assessing diversity will focus on

ethnicity as opposed to other kinds of diversity. As discussed earlier, one previous study

found that minority organizations (defined as organizations staffed and directed primarily

or exclusively by African Americans) were more likely to engage in advocacy activities

(Reisch, 1986a). As this research also focuses on advocacy, to further this line of inquiry,

this study limited exploration of diversity to examining ethnicities of persons of color.

Also, diversity data showed that other kinds of ethnic diversity was reported in the survey

very infrequently, and thus would pose a problem in statistical analyses. The first three

variables presented in the following table were created to capture the ethnic makeup of an

organization, through the percent of persons of color on staff and the board (defined in

this study as being people who identify as either African American, Hispanic, Asian or

Native American). The first variable, percent staff ethnicity, was the percent of staff who

were ethnically diverse. Similarly, the second variable, percent board ethnicity, was the

percent of board members who were ethnically diverse.

A third variable, organizational ethnicity, was created to help deal with the many

missing values for either the board or staff variables. The staff value was chosen to

represent the percentage of organizational ethnicity as it had fewer missing cases. As the

board and staff variables were correlated (R=.59, p<.01), this allowed the researcher to

use the board variable value when the staff variable was missing to create the third

variable. The fourth variable, categories of organizational diversity, divided these

percents into five categories. The fifth measured the organization‟s connection to diverse

ethnic communities, and the sixth measured the organization‟s connection to

AfricanAmerican and Hispanic communities only. Each of these is discussed below

along with the results.

Table 8. Comparing Mission by Organizational Ethnicity

% Board Ethnicity

Mean

Median

Range

Anti-violence; N= 40

18%

10%

0-100

.27

1.5

Anti-poverty; N= 109

24%

12%

0-100

.28

Total N= 149

22%

11%

0-100

.28

% Staff Ethnicity

Mean

Median

Range

Anti-violence; N= 58

27%

17%

0-100

.32

1.467

Anti-poverty; N= 109

33%

25%

0-100

.33

Total N= 138

30%

20%

0-100

.33

% Organizational Ethnicity

Mean

Median

Range

Anti-violence; N= 61

26%

35%

0-100

.31

2.116

Anti-poverty; N= 117

33%

43%

0-100

.34

Total N= 178

30%

41%

0-100

.33

*p< .05; **p< .01

While none of the differences in the above table were statistically significant,

some interesting patterns are worth noting. Anti-poverty organizations consistently

displayed higher percentages of organizational ethnicity, especially in the last variable,

percent organizational ethnicity. With the variability of the data, and especially the large

numbers of organizations with zero percent board or staff diversity, the distribution is

quite skewed. As indicated by the high standard deviations in the above table, the range

of values in the ethnic diversity of organizations varied enormously. Of the 150

organizations that answered questions regarding the ethnicity of board members, 44 had

all-white boards, and 11 had 80% - 100% ethnically diverse boards. Of the 167

organizations that replied to questions concerning the ethnicity of staff, 47 had all-white

staff, and 25 had 80% - 100% ethnically diverse staff. While the highest number of whites

serving on a board was 77, the highest number for African Americans and

Hispanics was 15 and 12 respectively. The highest number was 6 for both Asian

Americans and Native Americans serving on boards.

Examining the means and other statistics above does not completely reveal the

story that the data tell. As Figure 1 below shows, creating categories of the variable

provides a picture that shows how the distribution is skewed, and has significant numbers

of organizations at the low end of the scale of the variable organizational ethnicity

Figure 2. Comparing Mission by categories of organizational diversity

Categories of organizational ethnicity

Figure 2 above displays the categories were formed by grouping levels of

organizational diversity by percent into five categories. While chi square analysis found

no statistical difference by mission for categories of organizational ethnicity, the figure

indicates that the anti-violence organizations generally showed a higher percentage of

ethnic diversity in the lower categories (under 20%), while anti-poverty organizations had

100

40.0

30.0

20.0

10.0

0.0

Percent

Anti

violence

organization

Anti

poverty

organization

higher percentages in the larger categories (above 20%). Both anti-violence and

antipoverty organizations showed a significant percentage of organizations with no ethnic

diversity on the board or staff, 34% and 26% respectively. Four of the anti-violence

organizations (7%), and 12 of the anti-poverty organizations (10%), had 100%

organizational ethnic diversity.

Connection to diverse communities

As the ANOVA analyses below show, mission does make a difference in the

degree to which anti-poverty and anti-violence organizations are connected to diverse

ethnic communities for both the full and the limited measure, but not much of a

difference. For both measures, anti-poverty organizations had a statistically significantly

higher mean, suggesting a somewhat stronger connection to diverse communities.

Table 9. Comparing Mission by Connection to All Ethnic, and to African-

American & Hispanic Communities

Connection to ethnic

communities

Mean

Min.

Max.

Anti-violence; N=68

7.69

1.760

5.715*

Anti-poverty; N=128

8.61

1.865

Total N= 196

8.28

1.859

Connection to African-

American & Hispanic

communities

Mean

Min.

Max.

Anti-violence; N=70

4.79

1.735

7.519**

Anti-poverty; N=128

5.53

1.865

Total N= 198

5.27

1.851

*p< .05; **p< .01

Even though the above table does show a significant difference between

antipoverty and anti-violence organizations‟ connection to African American and

Hispanic communities, the figure below shows some complexity in the data. The

variable is displayed below in a bar chart, showing some variation in the distribution of

the values. While in general, anti-poverty organizations were more represented in the

higher percentage categories, there was one higher percentage category (value 7), where

antiviolence organizations were more represented.

Figure 3. Mission by Connection to African-American and Hispanic

Communities

2 3 4 5 6 7 8

Connection to African American and Hispanic

communities

The mixed results in Figure 3 above are difficult to interpret, as no clear pattern

emerges from the data. Of the eight values of the variable depicted in the next chart,

there were more anti-violence organizations in the values of 2, 3, 4 and 7, while there

were more anti-poverty organizations in the value categories of 5, 6, and 8. So, while

anti-violence organizations tended to be clustered at the lower end of the distribution,

they also fell into the higher end value of 7. Again, the bivariate analyses did not result in

statistically significant differences, so no clear conclusion can be drawn other than the

observation that both sub-samples did have varying degree of ethnic diversity among the

organizations in them.

Nonprofit Organizations with Social Workers and Consumers

While the next two measurements are somewhat different, they are presented

together here as they both attempt to measure the possible effect that having different

types of people in certain roles within the organization, and both are dichotomous.

Having degreed social workers conducting the advocacy activities would denote a more

professionalized approach to advocacy, while having consumers on the staff or board

would indicate a different, but not necessarily incompatible, orientation.

To address the hypothesis that greater involvement of social workers would

increase the number of advocacy activities an organization conducted, the survey

requested degree information on those working in the three advocacy activities.

Organizations were asked if advocates held bachelor or master level social work degrees

or other professional degrees.

To measure the construct of consumer empowerment, a dichotomous variable was

created that indicates whether or not any current or former consumers served on the board

or staff.

The table below shows the number and percentage of organizations with degreed

social workers employed in each area of advocacy, as well as whether social workers

were employed as any kind of advocate in the organization, and whether or not any

consumers were either on the board or staff. The results show that mission did not appear

to play a role in either of these predictor variables.

Table 10. Comparing Mission by Organizations with social workers and consumers

Anti-Violence

Anti-Poverty

Total

Pearson

Chi

Square

df=1

YES

N %

TOTAL

N %

YES

N %

TOTAL

N %

YES

N %

TOTAL

N %

Social workers

conducting any

advocacy activity

9 13

70 100

Missing=0

14 10

134 100

Missing=1

23 11

204 100

.286

Organizations with

consumers on the

board or staff

26 51

51 100

Missing= 17

61 64

95 100

Missing =40

87 60

146 100

2.412

*p< .05; **p< .01

As the chart above shows, only 11% of the organizations in this sample had

degreed social workers as advocates, and there was no statistically significant difference

in this low rate by mission. More social workers were employed in case advocacy for

both anti-violence and anti-poverty organizations as opposed to the other two advocacy

activities. Just 13% of anti-violence organizations hired social workers for any of the

three advocacy activities, and only 10% of anti-poverty organizations hired social

workers for advocacy work (this difference was not statistically significant). Both

subsamples of organizations had quite large percentages of organizations with consumers

on the board or staff (51% and 64%, respectively), but again this difference was not

statistically significant. It is interesting to note that the mean number of consumers

involved for anti-poverty organizations, 3.5, was significantly greater than the mean for

anti-violence organizations, 1.9 (p<.05). While many organizations had no consumers on

the board or staff, the highest number of consumers within an organization in the sample

was 21.

Research Question 2 on advocacy activities and mission

As discussed in previous sections, the central purpose of this research project is to

examine the organizational characteristics associated with certain advocacy activities.

This section asks the second research question: What types of advocacy do nonprofit

organizations conduct, what patterns of advocacy activities are evident, and does mission

(i.e. anti-violence and anti-poverty) make a difference?

Advocacy Variables

Nonprofit organizations were asked to indicate on the survey whether their

organization engaged in advocacy, a concept that was broken down into three broad

categories: case advocacy, public policy education, and legislative issue advocacy. A

chart of results for the total sample by each possible configuration of patterns of advocacy

is presented below. The main dependent variable of interest for the final research

question, empowerment-based advocacy, was created using the three core dependent

variables (i.e., it is the third type of organization in the table below). As shown in the

table below, the empowerment-based category was the most numerous pattern of

advocacy evident in this sample. Organizations indicating that they conducted all three

activities were said to be empowerment-based advocacy organizations, while those with

all other patterns of advocacy were seen as not being empowerment-based.

Table 11. Patterns of Advocacy

Case

Advocacy

Public

Policy

Education

Legislative

Issue

Advocacy

N %

Type of Organization

yes

17 8.6

1. Engaged Service Provider Organization

yes

38 19.2

2. Pragmatic Advocacy Organization

yes

97 49

3. Empowerment-based Advocacy

Organization

yes

15 7.6

4. Social Change Activist Organization

yes

8 4

5. Political Activist Organization

yes

6 3

6. Social Educator Organization

6 3

7. Disengaged Service Provider Organization

yes

11 5.6

8. Social Capital Organization

Since many of the cells in this categorization schema were too low for chi-square

or other analysis, the remainder of the study focuses on the three core dependent

variables: case advocacy, public policy education, and legislative issue advocacy, and a

fourth dependent variable, empowerment based advocacy organization. This last variable

was created by grouping all the cases with complete information on all three dependent

variables into two values. All of the 97 organizations in category three (in cases where

three of the aforementioned variables have a value of yes), were assigned a value of being

an empowerment-based advocacy organization, while all of the other categories were

grouped into the value of being a non-empowerment-based advocacy organization.

Descriptive data is presented in the next section for these three main advocacy activities

and the composite variable.

Table 12 below presents the results of examining the three advocacy activities:

case advocacy, public policy education, and legislative issue advocacy, and the dependent

variable of interest in this study, being an empowerment based advocacy organization, by

mission.

Table 12. Mission by Advocacy Activities

Anti-violence

Anti-poverty

Total

Advocacy

activity

Yes

Pearson Chi

Square df=1

Case

Advocacy

69 99%

(n=70)

100 74%

(n=135)

169 83%

(n=205)

**19.1

Public

Policy

Education

49 72%

(n=68)

82 62%

(n=133)

131 65%

(n=201)

2.146

Legislative

Issue

Advocacy

60 87%

(n=69)

101 76%

(n=133)

161 80%

(n=202)

3.408

Empowerment-

based

42 68%

(n=67)

55 42%

(n=131)

97 49%

(n=198)

**7.602

*p< .05 **p< .01

Case advocacy was the most frequent type of advocacy conducted among the

organizations surveyed. Out of 170 organizations, 82.7% stated that they

conducted case advocacy during 1997-98. As indicated in the table above, 98.6%

of nonprofits engaged in anti-violence work were engaged in conducting case

advocacy, compared to 74.6% of anti-poverty nonprofit organizations. The

difference between the number of antiviolence and anti-poverty organizations

conducting case advocacy was the only one of these three measures that was

statistically significant using chi square analysis.

Sixty-five percent of organizations in the total sample conducted public policy

education, making it the least frequently conducted advocacy activity among the three

categories of advocacy used in this study. Legislative issue advocacy was the second

most common type of advocacy activity conducted by the organizations in the sample,

with 80% indicating that they engaged in some legislative issue advocacy.

The last dependent variable created is the empowerment-based advocacy variable

that has two values, with yes representing those organizations that conducted all three

types of advocacy, and no including all other advocacy patterns. While neither the

antiviolence nor the anti-poverty organizations had a large percentage of organizations

that were categorized as empowerment-based advocacy organizations, substantially more

anti-violence than anti-poverty organizations were empowerment-based, and this

difference was significant at the .05 confidence level.

In summary, to answer briefly both Research Question 1 and 2, the data shows

that mission does matter. Of all bivariate analyses with the variables of interest (a total of

26), to this study, the mission variable of anti-poverty/anti-violence had the highest

number (12) of significant relationships. Clearly, the focus and mission of a nonprofit

organization has as much bearing on its nature as any other characteristic. In the next

section, the third research question will be explored with further bivariate analyses.

Research Question 3 on advocacy activities and other variables

A central aim of this study is understanding the characteristics of organizations

that conduct case advocacy, public policy education, legislative issue advocacy, and

empowerment-based advocacy (those that conduct all three forms of advocacy), and how

such characteristics are associated with the organizations‟ advocacy activities. This

section addresses the above question with bivariate analyses that explore how advocacy

activities are associated with each other and with other variables of interest. Only

relationships that are significant at p=.05 or lower are presented or discussed.

Associations among Advocacy Variables

First, some advocacy activities were associated with other advocacy activities.

Chi square analyses indicated that both case advocacy and public policy education were

associated with legislative issue advocacy (p< .05 and .01, respectively), but not with

each other. Among organizations that conducted case advocacy, 83% also engaged in

legislative issue advocacy, while 66% of the organizations not conducting case advocacy

did legislative issue advocacy.

Interestingly, when using another measure for lobbying (generally not used in this

study), acknowledged legislative issue advocacy (in which only explicitly acknowledged

legislative issue advocacy is counted as such), the relationship between case advocacy

and legislative issue advocacy disappeared. This indicates, as stated before, that a good

portion of lobbying by human service organizations is done unwittingly, and that thus the

connection between case advocacy and legislative issue advocacy is not one that would

be clearly recognized and affirmed by an organization that did both. This finding is

important only in so far as how empowerment is defined. Is lobbying that is done either

unwittingly, or in some sort of hidden fashion indicative of an empowered organization?

This question, and an examination of the limits and assumptions underlying the variables

used to answer it, will be discussed in the final chapter of this report.

A greater percentage of organizations that conducted public policy education also

conducted legislative issue advocacy than organizations that did not conduct public

policy education (87% vs 66%, respectively). Since the last variable for advocacy,

empowerment-based advocacy organization, is made up of the three advocacy measures,

analyses were not conducted with this variable and the three advocacy measures.

Associations among Advocacy Variables and other organizational characteristics

The next two tables present the results of chi-square analyses of the identity

variables of being a grassroots, service provider, or advocacy organization (organizations

could select more than one descriptor) with the dependent variables. Only those

relationships with a p value of <.05 or less are presented. As the following tables show,

exploration of the association of advocacy variables with the identity variables revealed

several significant relationships. First, those associated with case advocacy are presented.

Table 13. Identity of Organizations and Case Advocacy

Advocacy variable

Identity Variables

Pearson Chi

Square df=2

Case Advocacy Total

N=200

Service – Yes

N=171

Service – No

N=29

26.14***

Yes, N=164

88%

48%

No, N= 36

12%

52%

Case Advocacy Total

N=200

Advocacy – Yes

N=106

Advocacy- No

N=94

13.82***

Yes, N=164

92%

71%

No, N=36

29%

* p<.05; ** p<.01; *** p<.001

Two identity variables, being a service provider and/or an advocacy organization,

were associated with an organization conducting case advocacy. The table shows that of

those that identified as service provider organizations, 88% conducted case advocacy,

compared to only 48% of the organizations that did not identify as service provider

organizations. It makes sense that the service provider identity is indicative of those

organizations that conduct case advocacy, as case advocacy is a basic service that human

service organizations should provide (NASW, n.d.).

However, the finding that of those organizations with an identity of being an

advocacy organization, 92% conducted case advocacy, compared to 71% of organizations

that did not have an advocacy identity, is of interest. Most scholars of nonprofit advocacy

do not consider advocacy for individuals to be a bona fide advocacy activity, as they do

not view case advocacy to be focused on systems change. So, this finding that human

service agencies that deliver case management services, including case advocacy, do hold

an identity as an advocacy organization runs somewhat counter to the prevailing notions

about such organizations. As the theory in this study postulates, for those 49% of

organizations in this sample that do all three advocacy activities, their advocacy identity

may be based on a wider range of advocacy activities that operate on multiple

sociopolitical levels, from the individual to up to government.

Moving on to other associations with identity, in Table 14 below, the results of

associations between the identity variables and the other two dependent variables,

legislative issue advocacy (or lobbying) and public policy education are presented.

Table 14. Identity by Legislative Issue Advocacy and Public Policy Education

Advocacy variables

Identity Variables

Pearson Chi

Square df=2

Legislative Issue Advocacy

Total N=197

Grassroots – Yes

N=84

Grassroots – No

N=113

4.7*

Yes, N= 157

87%

74%

No, N= 40

13%

26%

Public Policy Education

Total N=196

Advocacy – Yes

N=104

Advocacy- No

N=92

13.13**

Yes, N= 104

75%

50%

No, N= 92

25%

50%

Public Policy Education

Total N=197

Grassroots – Yes

N=84

Grassroots – No

N=112

10.57**

Yes, N= 124

76%

54%

No, N= 72

24%

46%

* p<.05; ** p<.01; *** p<.001

Providing public policy education was associated with being a grassroots

organization, with 76% of organizations conducting public policy education also

identifying as being a grassroots organization (compared to 54% that conducted public

policy education but did not identify as a grassroots organization). A similar pattern was

discovered with public policy education and being an advocacy organization. There was

also a statistically significant association between being empowerment-based and a

service provider, although this association was more on the negative side in that 75% of

those that did not identify as service providers also did not have an empowerment-based

advocacy pattern, while only 51% of those that identified as being a service provider also

had an empowerment-based pattern.

Moving to the main dependent variable of interest, being an empowerment-based

advocacy organization, those identity variables associated with this variable are presented

below (table 15.). As the table shows, being an empowerment-based advocacy

organization was positively related to all three identity variables.

Table 15. Identity by Empowerment-based advocacy

Empowerment-based

Advocacy

Organization

Identity Variables

Pearson Chi

Square df=2

Grassroots – Yes; N=82

Grassroots – No; N=111

10.94**

Yes, N= 91

61%

37%

No, N= 102

39%

63%

Service – Yes; N=164

Service – No; N=29

7.25*

Yes, N= 91

51%

24%

No, N= 102

49%

76%

Advocacy – Yes; N=103

Advocacy- No N=90

17.41***

Yes, N= 91

61%

31%

No, N= 102

39%

69%

p<.05; ** p<.01; *** p<.001

One factor that may influence whether or not an organization conducts advocacy

is access to information on issues of concern. All statistically significant associations

of advocacy variables with information source are presented below in Table 16. The

information source variable asked whether or not the organization used information

from a coalition the organization belonged to, or a colleague from a nonprofit

statewide advocacy organization.

Table 16. Associations of advocacy with information source

A) Conducted

legislative issue

advocacy

Uses information from statewide advocacy colleagues

YES NO

N % N %

Pearson Chi

Square

df=2

YES

130 85

29 64

8.648**

24 15

16 36

N=201 Total

154 100

45 100

Conducted

legislative issue

advocacy

Uses informat

YES

N %

ion from coalitions

N %

YES

149

12 57

7.753**

9 43

N=201 Total

180

100

21 100

Conducted case

advocacy

N=204

Uses informat

YES

ion from coalitions

N %

YES

155 85

14 67

4.310*

28 15

7 33

Total

183 100

21 100

p<.05; ** p<.01; *** p<.001

As the above table indicates, legislative issue advocacy was associated with both

sources of information. A greater percentage of organizations that used information from

statewide advocacy colleagues to keep abreast of issues conducted legislative issue

advocacy (84%) than did not use such information (64%), and this pattern was true as

well for using information from coalitions of which the organization was a member.

Another factor that might influence whether or not an organization conducts

advocacy is knowledge level of the rules that govern nonprofit advocacy and lobbying, as

more knowledgeable organizations may be more likely to conduct political advocacy. A

significant relationship was found between the variable knowledge of IRS rules on

lobbying and whether or not an organization conducted legislative issue advocacy. Table

17 below presents the results of the association between such knowledge and legislative

issue advocacy (lobbying).

Table 17. Association between legislative issue advocacy and knowledge of IRS rules

Conducts

legislative issue

advocacy

n=184

Knowledge of IRS rules on lobbying

Pearson Chi

Square

df=2

Very

knowledgeable

N %

Somewhat

knowledgeable

N %

Not

knowledgeable

N %

YES

34 94.4

91 85.0

30 73.2

6.66*

2 5.6

16 15.0

11 26.8

Total

36 100

107 100

41 100

p<.05

These findings show that while a majority of organizations did legislative issue

advocacy in each category of knowledge level, the majorities in the very knowledgeable

and somewhat knowledgeable categories (94% and 85%, respectively) were larger than

the majority in the not knowledgeable category (73%), and those differences were

significant. When the categories of very and somewhat knowledgeable were merged and

compared to not knowledgeable (in order to create a dichotomous variable for later

analyses), the significance of the relationship remained (p=.036), and still, a greater

percentage of organizations with some or a high degree of knowledge conducted

legislative issue advocacy compared to those with no knowledge (87% vs. 73%).

One of the variables hypothesized earlier in this study to be associated with

advocacy was the presence of consumers on the board or staff of the organization. Table

18 below lays out the significant findings from comparing organizations with consumers

and advocacy variables.

Table 18. Associations of advocacy activities with consumers in organization

A) Conducts

public policy

education

n=142

Organizations with consumers on staff or board

of directors

YES NO

N % N %

Pearson

Chi Square

df=2

YES

63 74

29 51

8.078**

22 26

28 49

Total

85 100

57 100

Empowermentbased

organization

n=140

Organizations w/cons

YES

N %

umers on staff or board

N %

YES

18 32

9.236**

38 68

Total

100

56 100

* p=.05; ** p=.01; *** p=.001

As the table above indicates, having consumers on the staff or board was

positively associated with both conducting public policy education and being an

empowerment-based advocacy organization. Among organizations with consumers, 74%

conduct public policy education, compared to 51% that conduct public policy education

and do not have consumers. Of organizations with consumers on board or staff, 58%

were empowerment-based advocacy organizations, while only 32% of organizations

without consumers on the board or staff were empowerment-based advocacy

organizations. Clearly having consumers on the board or staff of an organization is

positively associated with several advocacy activities the organization conducts.

As shown in Table 19 below, the receipt of government funding, as well as

receiving funding from donations, was positively associated with conducting case

advocacy and being an empowerment based organization. Surprisingly, the other two

advocacy variables did not have a significant relationship to either source of funding.

Table 19. Percent Income from Government and Donations by Type of Advocacy

Empowerment-based Advocacy (EBA)

by % Income from Donation

Mean

Median

p-value

Conducts EBA; N= 81

9.5

.007

7.4

Doesn‟t conduct LIA; N= 94

Total N=175

14.5

Research Question 4: Does the theory hold up? Testing three hypotheses to explore the

study‟s conceptualization of empowerment theory

As stated previously, two multivariate approaches were used to test the hypotheses

of this study, classification trees and logistic regression. While this researcher did

conduct, analyze and write up the results presented in this section, the services of a

statistics consultant, Charles E. Heckler, Ph.D., were used to instruct this researcher,

guide the analysis and provide editing assistance.

The first method used to address research question 4, classification trees, is a

relatively new, nonparametric technique. This statistical technique was used to identify

predictor variables for creating the logistic regression models for each of the dependent

variables.

After the results of the classification tree analyses are presented, the results of the

logistic regression analysis follow. Before presenting the results of the tree analysis that

is the central aim of this study (using empowerment-based advocacy organizations as the

dependent variable), an example of a tree analysis of the predictors for case advocacy is

presented as an introduction to the reader about this multivariate form of analysis.

Classification trees

As classification trees are not often used in the social sciences, this section begins

with an introduction to how the technique works in JMP. JMP is the statistical software

that was used in this study to produce the classification trees. The tree produced for case

advocacy is presented below and used as an example for understanding the terminology

and results of such analyses. A description of the procedures and decision rules used in

this study follows. Finally, the results from the tree for empowerment-based advocacy

organization are presented.

Classification trees are used by researchers in many disciplines for a variety of

reasons. First, classification trees provide a clear visual representation of the data.

Second, they are often used in exploratory analyses (Clark & Pregibon,1998), as they

identify variables associated with a dependent variable and rank them in order of

importance. Finally, this technique is especially well-suited to data that are not normally

distributed and have missing values (Lewis, 2000), which is the case with this sample,

and it is useful in developing models for logistic regression (Kuhnert, Do, and McClure,

2000). Using classification trees to build a model is especially helpful in this study due to

the small size of the sample, as the number of variables, including interaction terms, that

can be used in a logistic regression analysis with a small sample is limited.

Classification tree methods provide a sequence of simple decision rules based on

the independent variables to classify the dependent variable. These decision rules are

each defined by splitting the given predictor into two parts (i.e., the building of a “tree”

via the “branches” of the splits). Classification trees use several statistical methods to

discover which splits best represent the distribution of the data. The resulting tree

graphically depicts which variables have the greatest impact on the dependent variable.

They are used extensively in medical research, especially diagnosis. For example, in one

study of pediatric surgery, the results of a classification trees analysis produced a decision

rule that instructs a surgeon to use a scalpel of a certain size for throat surgery based on

the length of the child‟s neck (Lewis, 2000).

Understanding classification trees and their use for building models

As Clark and Pregibon explain (1998), classification trees use binary recursive

partitioning to split the data into a tree with binary decision rules that are visually

displayed in a pictorial format. Binary refers to the fact that this technique splits, or

partitions, the data into two groups, and this process is repeated on the groups to build a

tree with several branches (hence the term recursive).

The tree program examines all possible predictors (and splits in continuous

independent variables) and picks the predictor that has the strongest influence on the

dependent variable. More specifically, the program calculates the probabilities of the

impact of each of the predictors, and then selects the predictor which groups the

observations using the dependent variable into two sets with the greatest difference in

probabilities for the dependent variable. In the case of a continuous variable, JMP also

calculates all the probabilities of possible splits of the predictor, and then splits that

variable at the point that will result in the greatest difference of probabilities for the

dependent variable. For example, in the example presented below of a classification tree

predicting case advocacy, JMP found that the predictor percent of ethnic diversity was

best split at >=.05, as the probabilities of being empowerment-based differed the most;

the probability of being an empowerment-based advocacy organization is 81% for those

organizations at or above 5% ethnic diversity, and the probability being only 56% for

those below that percent.

A few words about terminology are in order. The top node of the tree, or the root,

represents the all of the observations of the dependent variable. Next, there are splits on

specific predictors, and the splits are defined within a node. Nodes with no further splits

are called terminal nodes or leaves. The successive splits resulting from the analysis

produce decision rules based on the predictor variables in the order of their importance in

predicting the response. This process is done until the splits result in homogeneous leafs,

or splitting further is not possible or desirable based on the stopping and pruning rules

adopted by the analyst (Lewis, 2000).

In Figure 3 for the case advocacy tree, the top node is case advocacy, the

dependent variable the tree is predicting. In the first box, which represents case

advocacy, one can see the count, or (n), for that box, along with the G^2 and logworth

value for the first split below (these latter two numbers are statistics that guide the

splitting process, and are explained in more detail shortly). JMP tests all the predictors to

find which one is most important in predicting case advocacy. In this case, JMP selected

the first predictor to be the variable anti-violence/anti-poverty (the node called APAV in

the figure), and then selected the second most important predictor to be organizational

ethnicity (the node called PercDiv in the figure). Organizational ethnicity was split into

two categories, under 5% ethnically diverse, and over 5% ethnically diverse.

Inside the boxes representing the two nodes in the first split of the tree (just below

the top node of case advocacy), one can find the values of the variable at the top of each

box, just after the variable label. The node on the left represents those cases that are

antiviolence organizations(APAV=1), and the one on the right is anti-poverty (APAV=0).

In all of the nodes for the variables, the value (labeled as “level”) refers to those

organizations that do conduct case advocacy (1 = yes), and those that do not (0 = no). In

the column to the right of the levels are the probabilities, shown as percentages, for each

level. For example, in the right node of the first split, one can see that anti-poverty

organizations (APAV=0) have a 74% probability of conducting case advocacy. As is

evident by the bars in the left node of the split, being an anti-violence organization

(APAV=1) almost perfectly predicts whether or not it conducts case advocacy, as 99% of

those organizations conduct case advocacy. Before proceeding to discuss the tree results

further, an explanation of G^2 and the logworth for G^2 is required, as these tests are

used in determining whether or not to proceed with another split for a node.

The symbol G^2 represents the Likelihood Ratio (LR) Chi-Square statistic (SAS

Institute, Inc, 2005, p. 601-2). The LR Chi-Square test, analogous to the Person

ChiSquare test, essentially measures how large the differences in probabilities (of the

dependent variable) is between two groups of data (again, the groups are created by the

splits the tree analysis generates). The logworth statistic, which functions as a kind of

pvalue for the LR Chi-Square test, is a key tool to use to prevent overfitting, and is a

feature unique to JMP. Logworth for a given split is –log10(p), where p is the p-value

associated with a test that the data within a node is homogeneous, and has no association

with any of the available predictor variables. The logworth indicates whether a split and

all previous splits are significant statistically. Values of 1.3 or more can be considered

significant at the 0.05 significance level. At a practical level, any split with logworth <

1.3 is not be statistically significant. Because the splitting is applied recursively, a

complex procedure based on multiple comparisons is necessary in order for the logworth

values to be accurate (Sall, 2002).

In the tree for case advocacy below (see Figure 3), note that JMP places the

logworth values in the right side of the box representing each node that is split. In the

two splits of the case advocacy tree, neither of the logworth values for the G^2 tests were

under 1.3 (these can be found in the first box on the right side, and in the box for

AVAP=0). Thus, these splits are deemed to be statistically significant. The process for

producing the tree is that the researcher clicks on each node and thus directs JMP to split

the data and create new branches (splits) of the tree until an insignificant logworth value

is encountered in a node. Then the researcher prunes back any insignificant splits, so that

only the statistically significant splits remain.

Another tool to prevent overfitting is crossvalidation. The goal of crossvalidation

is to assess how well a model will predict cases that were not used to develop the model.

JMP randomly assigns cases in the dataset into a number of groups. In this study, the

number of groups was set equal to 10 to keep the group sizes small. Although there is no

well-defined criterion for the number of cases used for crossvalidation purposes, the

number 10 is commonly used for samples of this size (Charles E. Heckler, personal

communication). For each group, the model is fit using cases not in that group, and the

cases in the group are predicted using that model. The result (was the prediction correct

for each case in that group?) is stored, and the procedure continues for the other groups.

The aggregate result is converted into an R^2 statistic (the larger, the better),

called the “folded R^2”, and displayed alongside the ordinary R^2 for the model that

used all the cases. The folded R^2 will always be less than the ordinary R^2. If the

folded R^2 decreases when a split is added, this is another sign of overfitting. The box

below the tree for case advocacy below shows the split history, indicating that the folded

R^2 (the lower line) for the case advocacy tree began to level off after two splits,

resulting in a folded R^2 for the model of .1495. None of the splits beyond thus point

had significant logworths.

Figure 3. Classification tree predicting organizations that conduct case advocacy

Case advocacy

Crossvalidation

k-fold

-2LogLike

RSquare

10 Folded

162.032423

0.1495

Overall

156.764657

0.1772

Split History

All Rows

205

ount

190.51

496

G^2

6.3579241

LogWorth

vel

0.1756

0.8244

Prob

APAV(1)

Count

10.482636

G^2

Level

0.0143

0.9857

Prob

PAV(0)

135

Count

154.5

1579

G^2

1.7712327

LogWorth

evel

0.2593

0.7407

Prob

PercDiv>=0.046667

Count

96.820788

G^2

Level

0.1919

0.8081

Prob

PercDiv<0.046667

Count

49.461234

G^2

Level

0.4444

0.5556

Prob

100

The two variables identified by the tree as being the best predictors for the

dependent variable, case advocacy, are mission (AVAP- either anti-violence or

antipoverty), and organizational diversity (PercDiv). The node for APAV = 1 (the value

for anti-violence organizations) shows that being an anti-violence organization has a 99%

probability of predicting that it is an empowerment based organization. As the tree

indicates, the second variable, organizational diversity, is the best predictor only for the

branch in the tree representing the anti-poverty organizations. So, for anti-poverty

organizations only, organizations with 5% or more ethnically diverse boards and staff are

more likely to conduct case advocacy. One can see that the nature of the classification

process results in a graphic depiction not only of the best predictors, but also the

interactions of those predictors.

Summary of decision rules used in this study for trees analyses

The following decision making rules and processes for the classification trees

were used in this study. For crossvalidation, the number of groups was set to be equal to

10 to keep the group sizes small. Second, any splits with a logworth < 1.3 would not be

statistically significant and were not used. Further, if the folded R^2 began to decrease or

level off, as in the case above after two splits, splitting was stopped. While JMP offers a

number of powerful approaches in using cases with missing data, in this study, the

missing values rule was set to random, as this option is more conservative and less

biased. This means that cases with a missing value were assigned randomly to groups

during the procedure.

101

Another conservative rule was observed to keep the number of predictors small.

A simple tree is easier to interpret. If there are too many splits, the complexity could

make the tree hard to interpret. If the number of cases in a node is small (e.g. 10), the

practical significance of splitting further is doubtful for survey data, as such data tends to

be noisy, so thus any splits with fewer than 10 cases in a node were not used (Charles

Heckler, personal communication). All of the above measures are conservative rules

designed to prevent overfitting of the data, and to ensure that the results of the

classification tree analyses would be small, parsimonious models.

Tree for Empowerment-based advocacy organization

The tree created below in Figure 4 used the dependent variable

empowermentbased advocacy. As the results show, the model includes three variables,

income from government (split at >=.26 and <.26), anti-violence (1) or anti-poverty (0),

and organizations with consumers (0=no, 1=yes). The first box at the top contains all

cases used for the analysis, and the percents (as probabilities) of the dependent variable

are shown inside the box (in this tree, 49% of the cases are empowerment-based

advocacy organizations (EBA); EBA organizations are indicated in the diagram as the

hatched bars, while the non-EBA organizations are shown in red).

In the first split, the node on the left side shows that having 26% or more income

from government is the strongest predictor that an organization will be an

empowermentbased advocacy organization, and on the right side, having less than 26% is

the strongest predictor that an organization will not be empowerment-based. Moving to

the second level of the tree, on the right node, for those cases having less than 26%

income from government, the strongest predictor of whether or not an organization is

empowermentbased is whether an organization has consumers on the staff or board, as

102

39% of organizations with consumers were empowerment-based compared to 16% of

those that did not have consumers on the staff or board. On the left side, for those

organizations with 26% or more of their income from government, being an anti-violence

organization is the strongest predictor of whether an organization is empowerment based

(69% of antiviolence vs 50% of anti-poverty). The third level of nodes for anti-violence

organizations with 26% or more income from government shows that again,

organizations with consumers have the greatest predictive power for being an

empowerment-based organization. The model has a folded R² of 0.094.

Figure 4. Classification tree predicting organizations that conduct empowerment-based

advocacy

Legend: Conducts empowerment-based advocacy (1):

Does not conduct empowerment-based advocacy (0):

103

Empowerment-based advocacy organization

Again, classification trees help to identify those variables that have the most

important effects on the dependent variable. The above results helped to identify three

104

variables as being the most important of those associated with the dependent variable of

being an empowerment-based organization. These three, income from government,

organizations with consumers, and anti-violence or anti-poverty, are thus used in the next

phase of the multivariate analysis.

Logistic Regression Analyses

The predictor variables identified through the classification trees analysis were

used to build a logistic regression model for predicting whether or not an organization

conducted empowerment-based advocacy (EBA1). All three of these variables, income

from government (IncGov), organizations with consumers (OrgsCon), and anti-violence

or anti-poverty (AVAP), plus all interaction terms, and one quadratic term for the one

continuous variable, income from government, were entered into a logistic regression.

A backwards, mixed stepwise regression identified variables for removal. While a

forward stepwise procedure may produce a more parsimonious model, the backwards

stepwise process was selected in order to retain as many terms as possible. A backwards

mixed process was used, starting with the predictors identified by the tree analysis and all

interaction terms. Backwards mixed selection process was selected as the best regression

procedure to use to determine the predictors given the parameters, such as the size of the

sample (Charles Heckler, personal communication). Since all of the predictors were

selected by the tree analysis, this allowed the researcher to use a backwards mixed

105

process (a requirement of a backwards process). A backwards mixed process produces

models with “greater prediction accuracy of future observations,” and is preferable to a

forward selection process, as a forward process is biased in an optimistic direction

(Charles Heckler, personal communication). A backward mixed procedure allows a

variable that has been dropped to reenter the model, which does not happen in a forward

selection procedure, and this reduces the bias of the final model (Charles Heckler,

personal communication).

In Table 20 below, the results for associations of variables used in the logistic

regression model, means or counts, and standard deviations for the two continuous

variables, are provided.

Table 20. Associations, means, and standard deviations for empowerment-based

advocacy organization and predictor variables

Variable 1

3 4

Means

Or counts

106

1. Empowerment-

based advocacy

Yes=97

organization

No=101

N/A

198

2. Anti-poverty/

anti-violence

5.575*

AP=131

N/A

198

AV=67

3. Organizations w/

Consumers

10.914*

4.102*

Yes=84

N/A

146

150

No=56

4. Percent income

Government

5.118*

2.762

.053 -

.568

.339

174

180

135

107

*= p<.05

Note: The table values on items 1-3 represent results from Pearson chi-square analyses, while #4 used

one-way anovas to obtain F statistics.

The table above shows, as would be expected, that all three of the main effects

variables are associated with the dependent variable, and as previously noted, the

organizations with consumers variable is correlated with the anti-violence/anti-poverty

variable. If predictors have strong associations among themselves, this is likely to

adversely affect the interpretation of the regression coefficients; here, none of the

predictors were associated strongly enough to cause a problem.

Since a number of cases were missing values for one or more of the four

independent variables, they could not be used in the analyses. A total of 130 cases of the

original 198 cases which had a value for being an empowerment-based advocacy

organization and the variables in the model were used in this logistic regression analysis.

This problem of missing data is discussed in the limitations section of the next chapter.

Again, using a backwards mixed stepwise procedure, all of the interaction terms

were found to be unimportant. Income from government squared was significant. The

final fitted model thus had four variables consisting of three main effects and one

quadratic term. Let p denote the probability that EBA1=1. Then the model is:

log APAV OrgsCon IncGov IncGov2 + e

108

Since SPPS offers the standard Hosmer and Lemeshow Lack of Fit test for logistic

regression, and JMP has a different lack of fit test, the final logistic regression model was

also run in SPSS, using the binary logistic procedure in SPSS, version 14.0; results are

below in Table 21.

Table 21. Summary of Logistic Regression Analysis Predicting Empowerment Based

Advocacy

________________________________________________________________________

Variable Regression

Odds Ratios

Odds Wald 95th C.I. for

__________ Coefficient

Orgs

Ratio_ __Statistic_____Lower - Upper

With 1.567

Consumers

[Values: 0=no (Ref.);

1=yes]

0.444

5.179

13.823* 2.176 - 12.324

Anti-violence/ 0.971

Anti-poverty

[Values:

0=Anti-poverty (Ref.);

1=Anti-violence]

0.446

2.464

4.151* 1.035 - 868

Income from 6.291

Government

[Values 0-100%]

2.643

370.634

5.074* 2.156 - 63705.823

Income from -4.858

2.5

.011

3.274 .000 - 1.455

government²

* p < .05

109

The model proved to fit the data fairly well, as the area under the ROC curve was

0.757 (1.0 is perfect), with sensitivity at 0.683, and specificity at 0.671 (cutoffs for both

were .5). In addition, the Hosmer and Lemeshow Lack of Fit test (using SPSS‟ defaults)

also indicated that there was no lack of fit (Chi-square 7.518, p=.482). The Wald

statistics listed in the table below indicate that three of the four variables used were found

to be statistically significant. While the nature of logistic regression does not facilitate

the use of a R² as does ordinary least squares regression, several statistics have been

developed that are somewhat equivalent. One of the most commonly used statistics,

Nagelkerle R², produced a modest value of .279 for this model. The model fit statistics

(e.g. area under the curve), the Wald statistics and the Hosmer and Lemeshow lack of fit

tests indicated that the model used met current standards for valid logistic regression

analyses. There is no evidence of lack of fit, the model has a statistically and practically

significant fit to the data, and thus the model can be reliably interpreted.

The results of the logistic regression model presented in Table 20 show that the

odds of an organization being empowerment based is 5.2 times greater for those with

consumers on the board and staff (given the reference values noted above, i.e., for

antipoverty organizations with no government funding) than for organizations with no

consumers. The odds of an anti-violence organization being empowerment-based (given

the reference values of no government funding and no consumers on staff or board), are

2.5 times greater than for anti-poverty organizations. The odds ratio for income from

government is so extraordinarily high compared to the other variables in the model

110

because it is a continuous variable with some very large values for a number of cases, and

is not bounded from 0-1 as are the other variables. As many find comparing probabilities

to be more intuitive, further comparisons are made below using predicted probabilities.

Comparing probabilities

Using the predicted probabilities is often regarded as more accessible than using

odds ratios (which is a number made up of four separate ratios, and thus more difficult to

comprehend). Two charts of predicted probabilities are provided in this section that

demonstrate the effect of the variables in the model.

In Table 22 below, the probabilities are presented for the variables in the model,

using all possible combinations of the dichotomous variables, and the values of the

continuous variable of 0%, 50%, 70% and 100%. As Figure 5 shows later in this section,

these percentages capture the most salient values of the variables income from

government and income from government².

Table 22. Comparing probabilities for an organization being an empowerment-based

advocacy organization by anti-violence/anti-poverty and other main effects variables.

________________________________________________________________________

Predictors Probability of being empowerment-based

Anti-violence Anti-poverty

Organizations

With consumers

Income from Government=0% 40% 20%

111

Income from Government=50% 82% 63%

Income from Government=70% 83% 65%

Income from Government=100% 73% 51%

Organizations

Without consumers

Income from Government=0% 12% 5%

Income from Government=50% 49% 26%

Income from Government=70% 51% 28%

Income from Government=100% 37% 18%

________________________________________________________________________

As previously discussed, anti-violence organizations were far more likely to be

empowerment-based advocacy organizations, in each of the above comparisons by around

20% compared to anti-poverty organizations, except in the case of organizations with no

income from government and no consumers in the organization on staff or board.

This increased probability of an anti-poverty organization being empowerment-based vs.

anti-violence organizations is consistent with the results of the trees analysis. Having

consumers on the staff or board also had the effect of increasing the probability that an

organization was an empowerment-based advocacy organization. This variable did so

most dramatically for those with no government income. The probability for

organizations with consumers was about four times higher for both anti-violence and anti-

poverty organizations compared to those without consumers.

112

Income from government also played a key role. While having income from

government increased the probability that an organization would be empowerment-based,

that probability decreased with income levels greater than 70%, actually beginning to

curve down as the percent of income approached 100%. The values of 50% and 70%

were selected for this depiction in the above chart as 50% of income from government

captures the zenith of the influence of this variable, while 70% marks the possible decline

of its influence. The implications of this finding have relevance for future research in this

area, as some important studies have used the dichotomous measure of either under or

over 50% of income from government.

These findings will be discussed further in the next chapter, but before moving on

to that chapter, one more chart presenting the probabilities in linear form will further

enhance our understanding of the data in this model.

Figure 5. Predicted probabilities of being empowerment-based by the four possible

configurations of Anti-violence or Antipoverty, and having consumers on the board or

staff

113

♦ A000= Anti-poverty organizations with no consumers

♦ A100= Anti-violence organizations with no consumers

♦ A001= Anti-poverty organizations with consumers

♦ A101= Anti-violence organizations with consumers

In Figure 5 above, the predicted probabilities of being an empowerment-based

organization by percent of government income is presented using the two other

dichotomous variables in the logistic regression model. The legend shows the following

four distributions representing different values of the two variables on the chart as

114

follows (starting with the lowest, solid line on the chart and proceeding to the top, dashed

line with the highest values of income from government).

Again, as Figure 5 clearly shows, the impact of having consumers on the staff or

board significantly raised the probabilities for both anti-violence and anti-poverty

organizations, showing that among these organizations, this variable was more influential

than the effect of an organization‟s mission. The distribution displayed on the chart above

also shows that the effect of income from government was universal across all value

configurations of the other variables. The quadratic variable of income from government

captured the hockey stick shape of the relationship between income from government and

being an empowerment-based advocacy organization. The usefulness of this variable in

the model is clear, as the quadratic form of the continuous variable captured the

curvilinear nature of this data.

Income from government increased the likelihood of an organization being an

empowerment-based advocacy organization as it grew from 0 to about 50%, at which

point the effect started to level off and at 70% income from government, started to drop.

The drop is of particular interest due to the fact that government funding requirements

include strict restrictions regarding the use of such funds for lobbying activities. Two

possible causes of this drop, discussed further in the next chapter, are the potential impact

of legal restrictions regarding the political use of government funds, as well as political

considerations of the power of the purse where nonprofits may be loath to bite the hand

that feeds them. While the decline shown by the graphic may not be statistically

significant (as it may be a mathematical artifact of the parabola used in a squared value),

115

the good number of organizations at the higher end of the distribution means that this

result is based on sufficient data. This drop in probabilities of about 10% across the board

for all cases has implications for both policy and empowerment-based practice that will

be discussed in the next two chapters (see Section 5 of Chapter 4, and Section 2 of

Chapter 5).

Summary of multivariate results

Both the trees analysis and the logistic regression fit the data and produced

feasible models. This tells us that these independent variables are indeed significantly

related to the pattern of advocacy an organization conducts. Furthermore, the fact that

both types of analyses proved fruitful indicates that the model itself has some predictive

value.

Both the classification tree and the logistic regression model used to predict

empowerment-based advocacy organization demonstrated that the variable income from

government played the leading role in predicting if an organization was an

empowerment-based advocacy organization among the nonprofits in this sample. The

tree split the variable income from government at 26%, which as the graph of

probabilities shows in Figure 5, is roughly the halfway mark between the low and high

points of each of the distribution lines. Both analyses demonstrate that the effect of

income from government is definitively positive from about 25 to 50%, at which point

the effect levels off. At about 70% of income from government, the effect starts to

116

decrease, to almost the same point as income from government at about 35-40%. This

finding that too much government funding can be detrimental to nonprofit advocacy has

been hinted at in findings from previous studies that analyzed qualitative data, primarily

from interviews of executive directors of nonprofits from a moderately good sized sample

(Berry, 2003; Donaldson, 2007). To this author‟s knowledge, this finding is the first

assessment based on quantitative data. The implications of this finding are discussed in

more detail in the next chapter.

In summary, mission, income from government and the involvement of consumers

are organizational characteristics that play a formative and interactive role in whether or

not an organization is empowerment-based in its advocacy activities. These findings are

discussed in the next chapter.

CHAPTER IV. DISCUSSION

Introduction and Chapter Organization

In this chapter, a discussion of the four main research questions is presented. The

relationships between the mission variable and the rest of the variables in the study are

discussed for the first two research questions that address the impact of mission on these

organizational characteristics. The third research question is then addressed, and a

pathway model that may be used for developing future hypotheses about the route human

117

service organizations may take to becoming empowerment-based is presented. Finally, in

order to address the central aim of this study that seeks to understand the organizational

characteristics associated with empowerment-based advocacy conducted in the nonprofit

sector, the findings generated from the two multivariate methods that test three

hypotheses are discussed.

This chapter is organized into six sections. The bivariate results for the first three

research questions are discussed in the first three sections, and in the fourth section, the

multivariate results that test the three hypotheses are discussed under research question 4.

After the findings from the four research questions are discussed, a discussion of

serendipitous findings is presented in section five. The sixth section, the conclusion

section, summarizes the discussion of the results. The limitations of this study, as well as

an exploration of the implications for theory, and social welfare policy and practice, are

presented in the final, concluding chapter of this dissertation report, chapter five.

Section 1. Findings for Research Question 1.

Again, the descriptive research question asked is: Did the organizational

characteristics of the nonprofits in this sample vary by mission (i.e., anti-violence

organizations or anti-poverty organizations missions)? In this section bivariate results

that pursue this question with all of the study‟s variables except the advocacy variables

are presented. The bivariate results with the advocacy variables (the dependent variables),

are presented in the following section that addresses the second research question. The

118

discussion of these bivariate results is presented below in the variable order used in the

results chapter (chapter 3).

Study Variables

Identity Variables

As noted in the methods chapter, two sets of variables were used in this study to

ascertain the organization‟s general identity. These two sets of variables were: 1) issues

of interest, and 2) identity variables that explored an organization‟s orientation toward

advocacy (they could select any combination of the following three categories: advocacy,

service provider or grassroots/community organizing).

The first set of identity variables explored what issues were of interest to the

organization, and again, these were used to check the validity of the mission variable, as

the mission variable (with labels of being anti-violence or anti-poverty) was an ascribed

measure of an organization‟s primary mission (based on the database from whence the

organizations came). The results of the bivariate analyses of these issue identity variables

indicated that the label given to the organizations in the mission variable were indeed

positively correlated in the manner expected. With 97% of anti-violence organizations

indicating that they were concerned with issues of violence against women and children,

the analysis determined that the attributed label of anti-violence was an almost perfect fit

for those nonprofits. In addition, the anti-violence organizations did indeed differ from

the anti-poverty organizations in this issue area focus. However, this does not mean that

the anti-violence organizations were narrowly focused on only issues related to violence

119

against women, as 71% of the anti-violence organizations noted that they were also

concerned with general issues of concern to women and children, as compared to only

50% of the anti-poverty organizations. This strong interest among the anti-violence

organizations in women‟s issues in general implies that these organizations may retain

much of the social movement fervor that such a broad focus indicates. Ferree and Martin

(1995) describe how the women‟s movement spawned many alternative service provider

organizations to deal with the unmet needs of battered women. These organizations

pioneered new services and focused not only on helping individual women, but on

creating systemic social, cultural, and policy changes through advocacy and

empowerment. These analyses confirm that the mission variable ascribed by the

researcher is indeed a very good indicator of the organization‟s mission for the

antiviolence organizations.

However, this identity variable was less descriptive for those deemed antipoverty,

as only 68% of these noted that they were concerned with poverty and economic

development issues. Thus, these anti-poverty nonprofits need to be viewed as

representing a broader constituency of nonprofits than those that are focused solely on

ameliorating or ending poverty. For the anti-poverty organizations, the descriptor of

antipoverty suffices if one assumes a great deal of diversity in this category.

In spite of the aforementioned differences, these results serve to confirm that the

mission variable, overall, fits as intended for both sub-samples (i.e., anti-violence and

anti-poverty), and is indeed a good indicator of the general missions of the organizations.

120

Having served their purpose to help assess the validity of the mission variable (being

either anti-poverty or anti-violence organizations), this set of identity variables is not

included in any subsequent analyses.

The results of analyses with the second set of identity variables (the three

identities organizations could select included: advocacy, grassroots/community

organizing, and service provider) describe the manner in which the organizations in this

sample have, or do not have, some sort of political or advocacy identity. A large majority

of the total sample identified as service providers (86%). This finding demonstrates that

the great majority of these organizations are not primarily advocacy organizations, in the

sense that the term is typically used in political science and nonprofit studies

(Minkoff,1999; Jenkins, 1987). Thus, as intended, the sample appears to be made up

primarily of community-based nonprofits that form local service systems.

These three identity variables revealed some interesting differences by the

antiviolence and the anti-poverty missions. Both of the organizational identities of service

provider and advocacy were statistically associated with the mission variable. The

antiviolence organizations were more likely to identify as service provider organizations

than the anti-poverty organizations (94 vs. 81%, respectively), as well as selecting an

advocacy identity (70 vs. 45%, respectively).

One likely reason for the marked difference by mission with the advocacy identity

variable is that anti-violence programs, like most victim services, almost always include

client advocacy in the legal system. In addition, the feminist movement‟s focus on

women‟s empowerment has been quite successful in shaping current domestic violence

121

and sexual assault policies (Ferree & Martin, 1995; Gelb & Palley, 1996). The strong

focus on legal advocacy, and thus the offering of case advocacy services, has become

standard practice in the anti-violence and victim services field. In addition, there has

always been a focus on macro level prevention activities in this field that involves

community education and systems change in both the human services and criminal justice

arenas (Ferree & Martin, 1995). Many of the job titles for staff conducting case advocacy

that were listed by the anti-violence organizations in a few open ended survey questions

included the word “advocate”. One person from an anti-violence organization filling out

the survey even scrawled in the margins of the survey: “Everything we do is advocacy!”

Such an activist orientation to service delivery is indicative of strong feminist values

(Weil, 1996).

In this same vein of thought, another interpretation of this finding is that as the

anti-violence organizations and programs are still relatively young, they may be more

closely identified with the social movement that spawned them than the older antipoverty

organizations. Previous research has shown that the domestic violence shelters and

sexual assault programs that make up these anti-violence organizations remain strongly

committed to the social change purposes and goals of the women‟s movement (Feree &

Martin, 1995; Gelb & Palley, 1996). While many of these organizations have become

relatively stable institutions, and to some degree, have become more professionalized, for

the most part, those who work in and steer these organizations do tend to identify

primarily as advocates.

Many scholars have observed that generally nonprofit organizations tend toward

122

wanting to be viewed as professional, especially as they age. In doing so, the more

grassroots and protest politics that embody community education and organizing as a

human service interventions and policies aimed toward ameliorating poverty, violence

against women, and other politically unpopular causes has appeared to decline, even

among organizations that had their root in the War on Poverty era. Guinier (1999) has

observed that such tactics have decreased in general in the civil rights movement since

the 1960s, and she claims that a loss of democratic empowerment among the poor is a

result. Peterson and Walker (2000) found that among interest groups representing social

services, both government funding cuts and a decline in membership caused these

organizations to shrink their operations from 1980 to 1985. The decline of government

funding for social services has continued to decrease since that time, and it seems likely

that the interest groups that represent social work constituencies have continued to

struggle to survive. Verba, Schlozman and Brady (1995) found that lower income

individuals were under-represented in various kinds of protest activities

Increasing inequality among the most vulnerable populations that social workers

serve is an important factor in assessing empowerment and involvement in advocacy.

Verba et al. (1995) found that for those on the lower socio-economic strata of U.S.

society, the organizations that offered the opportunity to develop political skills were not

the same organizations that also invited their members to participate in some political

activity. For example, they found that while 66% of youth organizations offered

opportunities for practicing civic skills, only 6% asked participants to be politically active

(p.378). Their model shows very strong support for the factors they identified as being

123

key to citizen involvement, that more educated citizens with some degree of economic

resources who are politically engaged in some manner by an institution, and are asked to

participate, tend to be politically active. They conclude from their analyses that “the full

effect of the participatory process works against the recruitment of those most in need.”

(p. 487). In their study of membership organizations, Peterson and Walker (2000) find

that insider strategies that do not tend to confront the status quo as much as work with it

are overwhelmingly used by citizen groups, and they conclude that the prospects in the

U.S. for political mobilization against economic inequality are dismal. Political

mobilization of those at the bottom of the social order is exceedingly difficult because

there are few patrons able or willing to risk the danger to their own political well-being

that might arise from heavy political conflict over redistributive social programs. Elected

officials, of course, are free to take the lead in promoting legislation designed to aid

disadvantaged groups without prompting from any outside force, but they know that once

conflict begins over their proposals, there will be few organizations in place that can

effectively mobilize expressions of support, supply information and ideas, or raise the

financial resources needed to combat the program‟s critics. (p.196)

Perhaps it is a result of this state of affairs that neither the mission of being an

anti-poverty or an anti-violence organization was associated with the identity of being a

grassroots/community organizing nonprofit. To this researcher‟s knowledge, there are no

surveys of nonprofit human service organizations with representative samples that

document the prevalence of anything approximating a grassroots/community organizing

124

identity. Just 43% of the organizations in this sample viewed grassroots action and

community organizing and mobilization as part of their identity, which indicates that a

minority of the organizations in this sample conducts community organizing. One would

hope to see a majority of these organizations conduct community organizing, as such

activism would seem to be more likely among human service organizations that serve

vulnerable populations that yet are still in need of social and economic justice, and thus

see a greater proportion identify as grassroots community organizing nonprofits.

Nonetheless, if we assume that identity equates to action, the organizations in this sample

were more likely to conduct community organizing than the typical human service

nonprofit.

The three identity variables also revealed some aspects of the organizations‟

potential or inclination to engage in several advocacy activities. They are included in

subsequent bivariate relationships under research question 3 and are used to build a

model that describes a potential pathway for organizations becoming empowermentbased.

However, as these are not integral to the theory guiding the hypotheses in research

question 4, they were not included in the multivariate phase of later analyses. The

discussion in the next section of the findings regarding the capacity variables used in this

study further explores the concerns and questions discussed above.

Capacity variables

125

In this section, bivariate findings of the capacity variables by mission are

reviewed. These analyses included: information source (operationalized as two variables,

used information from statewide colleagues, and used information from a coalition),

knowledge of advocacy rules (operationalized as two variables, knowledge of IRS rules

surrounding advocacy, and takes “h” election), age of the organization (in years), source

of income (including the following categories: donations, government, fees, foundation

grants, united way, and miscellaneous), and size (including budget and total number of

staff).

Information sources

The first two capacity variables explored what information sources nonprofits

used to stay abreast of issues and results indicate that the great majority of both

antiviolence and anti-poverty organizations are part of policy networks that provide them

with policy-related information. The only one of these two that had a statistically

significant difference by mission was the variable, used information from a coalition,

with 96% of the anti-violence organizations using such information, and 87% of the

antipoverty nonprofits doing so. This finding is not surprising, given that the anti-

violence organizations were selected for the sample because they are members of

statewide coalitions. Hence, it is probably by definition that the anti-violence

organizations were most likely to receive information from coalitions. It does not seem on

the face of it that this difference by mission is a terribly meaningful one, as it seems that

the main finding of interest is that the great majority of both indeed used such policy

126

networks for their policy information. Certainly, regardless of whether such policy

networks are formal structures such as coalitions, or informal, loose ones, as the other

variable attempts to capture (used information from statewide advocacy colleagues),

either of these organizational connections could be presumed to add to the advocacy

capacity of an organization, if only by the provision of policy information.

Nonetheless, this finding that the anti-violence organizations appeared to use

resources offered by two statewide coalitions (in that they were significantly more likely

to use such a coalition) implies that they may have had more capacity to reach and inform

their constituent organizations than the anti-poverty organizations. This finding could

also be affected by the reality that the anti-violence coalitions, with far fewer

organizational members, benefited from the ease and lower costs of staying connected

that a smaller membership number might offer. Also, more funding for policy work may

have been available to the anti-violence nonprofits, given the constant state aid to the

umbrella state-level organizations that these local member organizations have enjoyed.

While some of the anti-poverty organizations may have access to or membership

in similar networks, they may not all have had the opportunity to join such formal

coalitions. Thus, it may be that this difference could be an important indicator of other

significant factors related to mission, namely that the anti-violence organizations had a

greater capacity for advocacy than the anti-poverty organizations due to having a formal

coalition with funds that supported the infrastructure for outreach activities to their

members.

127

Knowledge

Related to the notion of information enabling or facilitating advocacy, two other

capacity variables assessed the institutional knowledge a nonprofit may have about the

rules of engaging in advocacy: Takes h election, and knowledge of IRS rules. Since so few

nonprofits in this sample took the “h” election (seven in total), chi square analyses could

not be performed on this variable. A statistically significant difference by mission was

found on the second variable, with 88% of anti-poverty organizations stating that they

were very or somewhat knowledgeable of IRS rules, compared to 60% of the antiviolence

organizations. There is no obvious cause for this difference by mission without examining

other associations. In addition, this difference may not mean much, because 46% of the

organizations that stated they were very knowledgeable did not know if their organization

took the h-election. This is an important piece of knowledge for any nonprofit

organization to have that is active in legislative advocacy. However, this finding may

simply be due to a lack of knowledge on the part of the person who completed this

survey.

Perhaps the anti-poverty organizations tended to be more professionalized, or

again, were less connected to a recent social movement. Or, their knowledge of the IRS

rules had a dampening effect, and they avoided conducting “lobbying” as they did not

want to engage in political controversy, or be seen as biting any hands that may be

feeding them. Another possible reason for this finding is that the anti-poverty

organizations were older, and thus on average had more time to acquire organizational

128

knowledge and experience. However, increased age was not associated with greater

knowledge. The association of age and mission is discussed in the next section.

Age

Age was another capacity variable that had a statistically significant difference by

mission. The mean age of the anti-violence organizations was 19 years, while the

antipoverty organization‟s mean was 28 years. This difference in age could possibly

mark a variation of the two organizational missions‟ connections to the social and

political movements that gave birth to them. As Salomon (1995) found, organizations

spawned during the civil rights era had a strong political change identity. On the other

hand, the younger anti-violence organizations are certainly closer in time to the more

recent women‟s movement that spawned them, which could make the effect of that

movement on these organizations more pronounced. As other scholars have suggested,

the activists that founded these alternative women‟s organizations may still be leading

them, or their stamp on them may not yet be erased by age and professionalization

(Ferree & Martin,

1995; Perlmutter, 1994).

Income source

Another clue to this puzzle is capacity, which is not just measured by the amount

of resources, as in the following variable, but also by the source of income and what such

an association with a funding source may foster. A finding of interest in this study is the

129

variable percent income from government was not statistically different by mission, while

percent income from donations was positively associated with being an anti-poverty

organization. A likely reason for this difference is that the anti-poverty organizations

tended to be older, and had more time to develop a strong reputation and nurture

relationships with donors. In addition, anti-violence issues tend to be more controversial

and challenging to our society, compared to feeding or assisting needy children, and so do

not tend to attract donors as easily. As a higher age was also positively associated with a

greater percent of income from donations, this interpretation seems likely.

No other source of income was found to be associated with mission. Nonetheless,

with the average of 57% income from government, this one source was clearly important

as it was far and away the largest contributor of funding to all of the organizations in this

sample. While anti-poverty organizations had a higher amount of income from

government than the anti-violence organizations (63% vs. 54%, respectively) this

difference was not statistically significant.

Size

The size variable measured the organization‟s capacity in the traditional sense of

the word, by its budget size, or in a few cases where budget figures were not provided, by

the total size of the staff, since this variable was highly correlated with expenditures.

While there tended to be a few larger organizations in the anti-poverty sub-sample, size

was not found to differ by mission on a statistically significant basis. Both anti-violence

130

and anti-poverty organizations displayed a great deal of variability in size, as is often the

case in most samples of nonprofit organizations (Child & Gronjberg, 2007; Boris, 1999).

The fact that almost 60% of the sample was deemed to have a small size (with revenues

under $499,999) does indicate that this sample may have captured more grassroots,

smaller local nonprofit organizations than is typically the case with the typical samples

used in these kinds of studies which tend to use either national (IRS) or state lists of

formally incorporated entities. The distribution of size in this sample did provide a good

degree of variability which facilitated the assessment of the impact of this key capacity

variable. Yet, from these analyses, mission does not appear to make any difference in

terms of an organization‟s size.

So far the discussion of the results shows some important differences by mission

in the following variables selected to measure various aspects of organizational capacity:

information source, knowledge of policy rules, age, and source of income. Anti-poverty

organizations were statistically more likely to have income from donations, to be

knowledgeable of the IRS rules on advocacy, and to be older, while anti-violence

organizations were more likely to use information from a coalition to keep abreast of

issues. Yet the fact that no statistically significant difference existed by mission in

regards to the size of the organization means there was no difference in the most

traditional sense of capacity, the amount of funding an organization has.

Exploring the first research question of how mission may affect or represent

significant organizational differences continues in the next section that examines how

mission is related to other predictor variables used in this study.

131

Diversity and Other Predictor Variables

Eight variables are discussed in this section. These include six variables

constructed to measure aspects of organizational diversity (specifically, to measure

ethnicities of color), and two that measure the involvement of social workers and

consumers in the organization.

Diversity variables

Six predictors attempted to measure organizational diversity by assessing the

degree to which the organization either represented, or was connected to, communities of

color. The variables measuring the construct of organizational diversity included: percent

ethnic (people of color) on the staff and board, organizational ethnicity (percent people of

color overall in the organization), categories of percent ethnic diversity, connection to

ethnic communities, and connection to African-American and Hispanic ethnic

communities. Only the two latter diversity variables were found to have statistically

significant differences by mission, and will be discussed below.

Anti-poverty organizations were found to have a slightly stronger connection to

diverse communities. In the case of the variable, connection to diverse communities, the

values for this variable consisted of a sum of the answers regarding the four communities

of color examined in this study, but the magnitude of the difference was only .92 (on a

five point Likert scale, with a total possible value of 16). For the variable connection to

132

African American and Hispanic communities (which only measured the organization‟s

connection to African American and Hispanic communities), the difference was .74

(again, on a five point Likert scale, with a total possible value of 8). Since the magnitude

of these differences was quite small, a reasonable assessment is that while these

organizational diversity measures found the anti-poverty organizations to be slightly more

connected to communities of color, this was not a noteworthy finding. This finding is

likely to simply be the result of the anti-poverty organizations on average being older,

which could allow them more time to conduct outreach and establish connections to

diverse communities. It is also very likely due to the fact that many anti-poverty

programs are located in urban centers (as poverty disproportionately affects many

communities of color), and thus the effort to connect with these communities is more

pronounced among the organizations with an anti-poverty mission. Also of interest here

is the finding that both the anti-violence and the anti-poverty organizations could be seen

as being, at least to some degree, representative of those communities, since many do

indeed have people of color on the staff and board. However, 34% of anti-violence and

26% of anti-poverty organizations had no persons of color on their staff or board.

The results of the analyses of the diversity variables by mission shows that any

differences between the anti-poverty or anti-violence organizations in terms of diversity

were either not notable in this sample, or not adequately captured with the variables used

to measure the construct of organizational diversity.

The seventh predictor variable assessed whether or not the organization had social

workers on staff as advocates, and the eighth assessed whether or not present or former

133

consumers were on the staff or board of the organization. Bivariate analyses by mission

indicated that neither of these predictor variables showed any significant differences.

More than 50% of both anti-violence and anti-poverty organizations had some consumer

involvement by having present or former consumers on the staff or board, and both had a

very low level of social workers hired as advocates.

In short, mission was not associated in a noteworthy way with any of the main

predictor variables of interest used later to test the theory used in this study.

Summary of discussion on Research Question 1

Do the organizational characteristics of the nonprofits in this sample vary by

mission? The answer is: To a large degree, yes. Among the 25 variables used in this

study, the only other variable with as many statistically significant associations as the

mission variable was the identity variable, advocacy organization.

The two identity variables of advocacy organization and service provider indicate

that anti-violence and anti-poverty organizations do, in general, differ in both their

missions and their orientation to social change. The statistically significant finding that

anti-violence organizations are more likely to identify both as advocacy organizations and

service providers is not that unexpected, given the very high rate of case advocacy being

conducted. Indeed, the fact that their orientation toward both the identity of being a

service provider and advocacy organization may signify that the anti-violence

organizations, many of which have a feminist orientation, view advocacy as being at the

134

core of all of their services. The anti-violence organizations were also more likely to be

younger, which could partly explain this social movement orientation, as there is a greater

likelihood that the influence and impact of the activists that created these alternative

organizations still lingers in the political orientation of these organizations. However, that

mission was not found to have a stronger association with identifying as being

grassroots/community organizing may indicate that this political orientation is generally

less accepted among the nonprofits in this sample.

There were no significant differences by mission in terms of the critical capacity

variable, size. If the differences by mission in terms of the other capacity variables were

indeed influential in terms of tendency toward conducting advocacy for one mission type

or the other, then these differences may have tended to balance each other out on both

sides. In other words, both organizational missions displayed different kinds of strengths

in the area of capacity. Anti-poverty organizations on average had a higher age and more

knowledge of advocacy rules working in their favor, while anti-violence organizations

used information from coalitions more often, and had a clear identity as an advocacy

organization. Both garnered a substantial percentage of income from government. While

the anti-poverty organizations received more income from donations, this was unlikely to

bolster their advocacy capacity, and may even be a factor that could have curbed any

impetus to advocate.

None of the variables used as predictors in the multivariate analyses under

research question 4 were found to be significantly associated with the mission variable,

except two of the measures of organizational ethnicity, but again, this difference was

135

quite small in magnitude. Among the next eight predictor variables discussed, only the

two that measured an organization‟s connections to diverse communities were found to

differ by mission. Again, while the difference was statistically significant, the magnitude

of the difference was quite small. Both anti-violence and anti-poverty organizations had

a fair degree of involvement of consumers, and both had a low number of social workers

on staff as advocates.

In this sample of nonprofit organizations, mission was associated with quite a few

variables. Having an anti-violence mission increased the likelihood of an organization

identifying as an advocacy organization and service provider, and using information from

a coalition, while being an anti-poverty organization increased the likelihood that the

nonprofit was older, received income from donations, was connected to diverse

communities, and had more knowledge of the IRS advocacy rules. The data suggests that

the anti-poverty organizations may have been less likely to identify as advocates as these

organizations included more of the larger, older, human service organizations which had

more income from donations and thus would be more likely to be professionalized in

their service orientation. This may mean that many anti-poverty organizations in this

sample had more of a tendency to preserve the status quo rather than disrupt it. Such a

tendency may not necessarily be a result of their outlook on policy issues, but because

receiving larger donations from wealthy donors (who would more likely, on average, to

be more conservative politically), and having an older, perhaps more staid reputation to

protect, had the effect of discouraging overt action against current policies.

136

Researchers have found differing levels of advocacy activity based on whether

nonprofits are human service or another type of organization, such as environmental

groups (Berry, 2003; Child & Gronbjerg, 2007). The findings discussed under this first

question indicate that as previous research has shown, there were indeed significant

organizational differences within sub-fields of the larger domain of human service

organizations, as indicated in this sample by the many differences according to an

organization‟s primary mission. Thus, these findings show the worth of conducting more

detailed examinations of human service nonprofits based on a bit finer tuned definition of

mission or issue focus of the organization.

This discussion of the impact of mission continues on to the next research

question that examines whether or not the types of advocacy conducted differ by mission

among the organizations in this sample.

Section 2. Findings for Research Question 2.

Research Question 2:

What kinds of advocacy are conducted by the nonprofits in this sample, and do

these activities vary by mission?

In this section, the bivariate analyses by mission of the dependent variables used

in later analyses in this study and the four advocacy variables (the three core advocacy

variables: case advocacy, public policy education, legislative issue advocacy, and the

137

empowerment-based advocacy pattern in which an organization conducts all three of

these), is discussed.

The main finding of interest relevant to this question is that of the eight possible

patterns of advocacy (created using the three core advocacy variables), the largest number

of organizations followed the pattern of most interest in this study, the empowerment-

based advocacy organization. This finding provides support for the contention that in

addition to meeting basic human needs, many of the nonprofit human service

organizations in the sample are indeed contributing to the work of maintaining

democratic institutions, as many political scientists and civil society scholars posit (Dahl,

1998). The fact that these three advocacy activities are taking place means that the

potential for a client of the organization to get individual level assistance towards

personal empowerment, policy information on issues related to the need they themselves

are experiencing, and the potential opportunity to impact the policy process though

legislative means, are all indicators that three of Dahl‟s (1998) essential democratic

political institutions, freedom of expression, alternative sources of information, and

associational autonomy are to some degree being realized in this sample of human service

sector. This finding also lends support to the theoretical framework guiding this study,

namely that all three types of advocacy are indeed related, and are often found to co-exist

in the nonprofit human service sector. Further discussion of the prevalence of each of the

three types of advocacy: case advocacy, public policy education, and legislative issue

advocacy, by mission is presented below.

138

It is heartening that so many of these human service organizations did conduct

case advocacy. This kind of client-based advocacy is indicative of social services that

meet the NASW ethical code, as social workers aim to ensure that a client basic human

rights and needs met, and that services take heed of the voice of consumers and are

responsive to individual needs. At 99%, case advocacy was the most frequently

conducted advocacy activity for the anti-violence organizations, a prevalence

significantly higher than the 74% for anti-poverty organizations. Of the three core

advocacy variables, only case advocacy was found to have a statistically significant

difference by mission. This finding is most likely due to the program standards of the

anti-violence programs that almost universally used case advocacy, compared to the very

diverse anti-poverty sub-sample that included a variety of service organizations, such as

housing and economic development, that do not by definition engage in service delivery.

Legislative issue advocacy was the most frequently conducted advocacy activity

for the anti-poverty organizations (76%), although not by much, as again, 74% of these

organizations conducted case advocacy. For the anti-violence organizations, legislative

issue advocacy was also the second most frequent advocacy activity (87%).

While a majority of both anti-violence and anti-poverty organizations conducted

public policy education (72% and 62% respectively), this type of advocacy was not

conducted as much as the other two types. As discussed in chapter 1, civil society

theorists extol the virtues of this sector in terms of its critical role in educating and

encouraging deliberation of pressing political problems (Dahl, 1998). From

empowerment based and democratic theoretical perspectives, the ideal pattern of

139

advocacy activities would see public policy education being conducted at least as often as

legislative issue advocacy, if not more so. This data shows a pattern of nonprofit

advocacy that is not a perfect fit with the ideals represented by empowerment and civil

society theories. This could be a result of the fact that public policy education activities,

which can be quite costly, are not typically funded by government, while case advocacy

is typically part of many service delivery systems, and can often be supported with

government funds. Rees (2001) found that in a sample of effective nonprofit advocacy

organizations, more insider, direct forms of advocacy such as legislative issue advocacy

were seen as being more efficacious than advocacy directed at the general public.

Although this study was not designed to facilitate an assessment of the prevalence

of advocacy (as this sample cannot be said to be representative of the population of

nonprofit human service providers), the fact that roughly half of the nonprofits in this

sample conducted all three advocacy activities is noteworthy. Using a definition of

advocacy that covers both legislative issue advocacy and public policy education, Child

and Gronjberg (2007) in their rigorous survey using a representative sample of Indiana

nonprofits found that the percent of human service nonprofits conducting advocacy was

only 15%. Berry (2003), found that 74% of the nonprofits in his nationally representative

sample did lobby, but only 31% did so frequently.

That the advocacy levels were relatively high among the nonprofits in this sample

may mean that the sample was biased in the direction of being more prone to such

activity. However, the rate of legislative issue advocacy was very similar to Berry‟s

(2003) national research finding. This potential bias of the sample was noted as being

140

likely earlier in the methods chapter, because these nonprofits were selected from the

mailing lists of statewide advocacy organizations. So, by definition they would be more

prone to advocate compared to the general population of human service nonprofits.

Regardless, these findings show that this sample indeed provides much data to use in

exploratory analyses aimed at refining the constructs and theories that seek to explain

nonprofit advocacy.

Summary for Research Question 2

Several important findings are evident. First, while anti-violence organizations

conducted more of each type of advocacy than the anti-poverty organizations, the only

advocacy variables with statistically significant differences by mission were case

advocacy and being an empowerment-based organization. Second, while many of the

human service nonprofits in this sample do certainly play a role in furthering democratic

and political engagement processes, a significant number do not. An important finding

related to the crucial need for an enlightened citizenry in a healthy democracy, the finding

that less public policy education is conducted compared to the other two activities may

mean there is a hole in the web of advocacy these nonprofits weave in their local

communities. This means that across both fields of practice, something of a „donut hole‟

existed in the continuum of advocacy from local to state socio-political levels. The gap

suggests that perhaps this crucial democratic function is not being fulfilled on the local,

grassroots level as often as one taking this view would hope. Public policy education is

141

not only a very effective advocacy strategy, it is needed to support and create an informed

public that can deliberate and weigh in on critical public policy issues facing society.

While some may be encouraged by the finding that more than half of the nonprofits in

this sample conducted public policy education, the data shows that public policy

education needs to be more fully supported and expanded among human service

providers, if the goal of client empowerment on macro levels of society through social

services is to be more fully achieved. This position is argued in more depth in the

conclusion section of this chapter.

While both the anti-violence and the anti-poverty organizations conducted a good

amount of all three types of advocacy, the anti-violence organizations were clearly more

likely to be empowerment-based. Again, there are several reasons that the anti-violence

organizations in this sample were more likely to be empowerment-based organizations

and conduct all three advocacy activities. They were likely to have greater connections to

a strong policy network by virtue of belonging to a coalition and they were younger and

thus were more likely to retain the ties to the more recent social movement that spawned

them. However, as a counterweight to the ability to advocate, they had less knowledge

than did the anti-poverty organizations of IRS rules.

Mission and empowerment-based advocacy organizations

In summary, a few observations may be noted from examining the results from

both research questions (1 & 2). While all of the variables and their relationship to

142

whether or not an organization conducts certain types of advocacy or not are analyzed

more completely in the next two research questions, a preliminary review of the factors

surrounding an organizations mission and advocacy activities is part of finishing the

discussion for the first two questions that focused on the impact of mission, as well as

setting the stage for the next phase of this study.

First, as noted before, the variations with both mission and age may combine in

different ways to have either a positive or negative influence on the organizational

tendency to advocate. The younger anti-violence organizations may have had less of the

experiences and connections that facilitate advocacy, but, as they are more closely allied

to the women‟s movement that created them, they could have had a more activist

orientation to their service delivery. So, perhaps the reason anti-violence organizations

are more likely to be empowerment based is because these tend to be younger,

freestanding nonprofits. Among the older anti-poverty organizations, their greater age

could have been an advantage in terms of the accumulated organizational knowledge and

credibility that established organizations are more likely to have, although the more years

that have passed since the civil right‟s movement could reduce the greater capacities that

come with fervent movement activists. And age may simply be a factor for organizations

with both missions, in that those who did indeed have the ability to advocate were both

more likely to survive. As mentioned before, the opposite may also be true, as the

increased donations that come with age could have a conservative influence on the

organization‟s inclination to advocate for social change (as may have been the case for

many of the anti-poverty organizations in this sample). Alternatively, non-government

143

support could create the freedom nonprofits need to try to change public policy. Clearly,

the interplay between these various capacity variables can result in many different

conceivable outcomes for organizations.

Again, while the older anti-poverty organizations had a greater length of time to

observe other organizations engaging in legislative issue advocacy, and perhaps get

encouragement and information in policy networks that such activities were legal, such

experience and access was not was sufficient to cause anti-poverty organizations to

engage in more lobbying than the anti-violence organizations. Surprisingly, there also

was no statistically significant relationship between an organization‟s age and its

knowledge of the IRS rules. However, this may have been the result of the wrong person

filling out the survey, as such knowledge is perhaps of a more technical and legal nature

than the average frontline advocate might have.

In addition, the anti-poverty organizations had more income from donations than

the anti-violence organizations, and this factor is likely to work against organizational

inclinations toward advocacy activities, as wealthy donors (many of whom may sit on the

board of directors of these organizations) tend to have more interests in the direction of

preserving the status quo rather than upsetting it. This suggests that the capacity provided

by strong policy networks is more important than the capacity provided by age or size,

given that both missions were almost equally well supported by income from

government, and that an older average by age did not appear to make the anti-poverty

organizations more likely to advocate.

144

The evidence as examined so far remains unclear as to whether on average

antiviolence or anti-poverty organizations had greater advocacy capacities. As discussed

earlier in the section discussing research question 1, the impact of the capacity variables

that differed by mission may cancel each other out to some degree in terms of the final

result, as both sub-samples had greater capacities in some aspects. Clearly, mission does

appear to make a difference, with the anti-violence mission having a greater number of

organizations engaging in empowerment-based advocacy. But a conclusive rationale as

to why mission appears to be a very important variable, with its many statistically

significant associations, remains elusive. The fact that mission was not associated with

conducting either public policy education or legislative issues advocacy makes this

question a rather crucial one.

Again, the data here suggest that for organizations with different missions, there

may be more than one set of capacities that make a difference in the provision of an

organizational capacity to advocate. Perhaps mission is simply a rough indicator that

there exists more than one set of circumstances that lead an organization to conduct all

three core advocacy activities. This question about the interplay of all of these variables

and their role in whether an organization is empowerment-based or not will be explored

in greater depth in the next section. A more in-depth discussion of how all of the

variables are associated with the advocacy variables is continued in the discussion of the

next research question that explores other significant bivariate relationships among the

variables of the study and the advocacy variables.

145

Section 3. Findings for Research Question 3.

Research question #3: How are advocacy activities associated with each other, and with

other organizational characteristics?

Several of the more intriguing findings discovered in the process of answering

research question 3 were regarding the relationship among the three primary dependent

variables, case advocacy, public policy education, and legislative issue advocacy, and

what these might say about the set of conditions under which an organization is more

likely to be empowerment-based (i.e., does all three types of advocacy).

The finding that public policy education was associated with legislative issue

advocacy is not surprising, as to be effective lobbyists, most organizations would need to

engage in at least some public policy education activities. This finding of an association

also supports the theory used in this study that connects these two variables as being

integral to the concept of an organization being empowerment-based.

While much of the literature on human service nonprofits and advocacy might

lead one to expect that service providers would not engage in lobbying (Verba, et. al.,

1995; Peterson & Walker, 2000), but that was not the case for this sample as many did

engage in lobbying and case advocacy was also positively associated with legislative

issue advocacy. As noted in the methods chapter, the measure of legislative issue

advocacy used counted organizations that both intentionally and unintentionally

conducted legislative issue advocacy, or lobbying. Whether intentional or not, the

146

majority of the human service organizations in this sample were engaged in legislative

policy change processes.

That the relationship between public policy education and case advocacy was not

statistically significant tells us this important civic engagement advocacy activity was not

as likely to be found among the nonprofits in this sample conducting case advocacy, i.e.

advocacy directed to and for individual clients. This implies that public policy education

activities may often be disconnected from social work service systems and the clients

they serve in many of these organizations. As Ezell (1991) found, advocacy actions by

social workers may be limited to simply informing clients of their rights, as opposed to

preparing and encouraging them to be active citizens in policy change processes. This

finding is not only of concern to the social work profession, which is committed to an

empowerment process that is robust on all socio-economic levels, but also to any who

care about having a robust democracy (Dahl, 2008).

The second part of the research question explores other variables‟ relationships

with the advocacy variables. The identity variables displayed an interesting pattern of

relationships among the organizational characteristics. Organizations that conducted case

advocacy were more likely to identify both as service providers and advocacy

organizations. Public policy education was associated with having both an advocacy and

a grassroots/community organizing identity, which speaks to another kind of capacity for

nonprofit advocacy: the passion that comes with a social movement orientation.

The capacity variables that measured the „hard‟ resources that one generally

associates with the capacity of an organization, such as size, total staff and total

147

expenditures, did not differ by any of the advocacy variables. This finding was similar to

Berry‟s (2004) research that also did not find any association between advocacy and total

number of staff, or total funding. The one capacity related variable he did find to be

associated was whether or not at least one staff person was designated to have

responsibility for government relations or advocacy functions in the agency; no variables

that measured that construct were used in this study.

In terms of source of income being a factor, it is interesting to note here that for

the organizations in this sample, income from government was negatively correlated with

income from donations (Pearson‟s R = -.606; p= .000), which implies that organizations

tended toward one or the other of these two funding sources. While this would seem to

be obvious, as if one increased another source would have to decrease, there were enough

other possible sources of income that this one statistically significant association remains

of interest. This data also suggests that nonprofits with a social service orientation that

provide case advocacy rely more on government funding. Perhaps this is due to the fact

that funding for case advocacy is focused on the poor and those with very pressing needs,

populations which have not been found to be extremely desirable for wealthy donors.

While income from government was not positively associated with legislative issue

advocacy or public policy education, income from donations was negatively associated

with conducting case advocacy (Spearman‟s Rho=-.168; p=,024).

Again, several scholars have hypothesized that some limitations may come along

with funding from donors, as perhaps a more conservative orientation, or one that simply

tries to avoid political controversy, is attached, and acts as a damper to nonprofit

148

advocacy. On the other hand, donations provide a key capacity unhampered by

government restrictions. Yet, both case advocacy and being an empowerment-based

advocacy organization were strongly associated with funding from government. While

the circumstances of when and how these sources possibly help or hinder nonprofit

advocacy remain elusive, the source of income certainly seems to play a key role here.

While such “hard” resources as staff and budget size were not associated with any

advocacy patterns among the organizations in this sample, there were some variables that

measured other types of capacities relevant to advocacy. Legislative issue advocacy was

positively associated with using information from statewide advocacy colleagues. Again

the resources of statewide policy networks appear to be an important contributor to an

organization conducting legislative issue advocacy, but were not associated with being an

empowerment-based advocacy organization.

Organizations that conducted public policy education were more likely to have

consumers on the staff or board (74% of organizations with consumers conducted pubic

policy education, compared to 51% of organizations that also did, but had no consumers

formally involved in their organization). This finding may reflect another type of resource

to build an organization‟s capacity to advocate: the power of passion and commitment of

volunteers and people directly affected by the issue at hand. Edwards and McCarthy

(2004) referred to such capacities as moral resources. As posited in the theoretical

framework used for this study, it may also be an indicator of an empowerment approach

to service provision and social change that purposefully involves and empowers those

149

affected by policies and supports their involvement in changing them through educating

them about public policies.

One interpretation of this pattern of connections among the dependent variables

and the predictors is that there are two typical sets of factors that may contribute to the

evolution of an empowerment-based organization, or perhaps two categories of

empowerment-based organizations. To lay the groundwork for this argument, and to

further explore research question 3, a model was developed using the bivariate results that

graphically displays the relationships among twelve key variables with a statistically

significant association with any of the three core advocacy variables. Using an inductive

approach to interpreting the data, the model presented below in Figure 5 illustrates two

main constellations of organizational factors that are associated with a nonprofit

organization being an empowerment-based organization. This model may offer some

utility not only in identifying the variables that predict empowerment-based advocacy in

the nonprofit sector, but also in recognizing that there may be more than one set of factors

to explain what leads an organization to be empowerment-based.

The list of predictor variables used to create figure 5 included 12 selected from all

of the variables used in the study, plus the three core dependent variables, case advocacy,

public policy education, and legislative issue advocacy. These included: the three identity

variables: service, grassroots/community organizing, and advocacy, and, key capacity and

predictor variables, including: income from government, organizations with consumers,

knowledge of IRS rules, use of information from statewide advocacy colleagues, age,

150

size, percent ethnic diversity, connection to diverse communities, and organizations with

social workers.

The mission variable was not used in this model as its influence was covered

extensively under research questions 1 and 2, and this researcher wanted to further

explore the other organizational characteristics most promising from the bivariate

analysis to see what picture might emerge, while trying to keep the number of predictors

as low as possible (note: all of these variables were used in later multivariate analyses, so

their influence was not neglected in other phases of this study). Other variables that were

not used were either not germane, or were omitted because another variable was deemed

to better capture the intended construct.

Only income from government was selected for use from all of the variables

describing sources of income, because overall it was the most important source of

funding for the organizations in this study, and because as discussed earlier, income from

donations was likely to simply measure the reverse effect of what income from

government measured. Of the predictors used in the next multivariate stage, only

organizations with consumers was used, as it too was a variable with a significant

associated with public policy education, a variable of key interest in this study.

In this model, those variables shown below the dotted line result from the

bivariate analyses, while above the dotted line are the three variables that were used in the

multivariate model (note: the multivariate results are discussed in more depth in the next

section; this brief preview is provided to introduce the context for understanding the

bivariate results). The depiction of relationships in figure 5 allow the opportunity to

151

explore what the pattern of relationships might reveal about the conditions that

accompany each core advocacy activity, case advocacy, public policy education, and

legislative issue advocacy, and may lead to understanding the underlying factors that

influence whether or not an organization is empowerment-based. Only statistically

significant bivariate relationships are depicted in figure 5 below. In Figure 5, significant

correlations represented by the two-way arrows were determined by Spearman Rho

analyses. With 12 key predictors, a p-value of .004 was determined to be significant using

the Bonferoni correction, and the corrected correlations are depicted with a solid line.

Bivariate associations with p-values of more than .004, but equal to or less than .05 are

shown with a dashed line.

Figure 5. Theoretical model showing two main constellations of variables associated

with empowerment-based advocacy organizations

152

The bottom half of figure 5 examines the pattern of bivariate relationships among

the three core variables that make up the main dependent variable of being an

153

empowerment-based advocacy organization. The heavy dashed line in bold leading from

legislative issue advocacy to empowerment-based advocacy simply denotes that when an

organization does all three, it is empowerment-based, and it also moves the reader‟s eye

up to the top three variables that make up the multivariate model that is used to predict

empowerment-based advocacy. Two-way arrows indicate the significant relationships

among the variables in the figure above. The solid lines indicate relationships that are

statistically significant after using the Bonferroni correction for the number of variables

used, which determined a significance level of p<.004 for overall 0.05 level significancc.

The dashed lines in the figure represent those relationships with a p value of more than

.004, but less than .05.

This diagram reveals two clusters of variables that are statistically associated with

each other: Cluster 1 are the variables on the lower left side of the diagram connected to

case advocacy. Cluster 2 are the variables on the lower right side around public policy

education. One interpretation of this picture is that these two clusters of variables with

some significant associations among them denote two different sets of conditions that

describe subsets of empowerment-based organizations. These two constellations of

variables may capture the phenomena that there may be different predictors that lead to

an organizations being empowerment-based for different subsets of organizations.

Cluster 1 on the left may reveal the factors most relevant for service provider

organizations that conduct case advocacy that is directly associated with legislative issue

advocacy, and only indirectly associated with public policy education via a developed

identity of being an advocacy organization. The predictors in cluster 2 on the right may

154

describe social movement organizations focused more on public policy education due to

their grassroots identity, with a stronger direct relationship to legislative issue advocacy,

than with cluster 1.

Again, for cluster 1 on the lower left side of the diagram, conducting case

advocacy is statistically associated with the identities of being both a service provider and

an advocacy organization. Case advocacy also has a statistically significant relationship

with income from government and legislative issue advocacy when only the bivariate

analyses alone are used (they are not statistically significant using the Bonferroni

correction). The fact that income from government is associated with case advocacy, but

not with providing public policy education or legislative issue advocacy, appears to

support this notion of a different set of factors for different subsets of human service

providers that are empowerment-based. Human service providers usually depend on

government funds to deliver government-approved services, and they will often advocate

in support of continued capacity to fulfill their missions (Rees, 2001). It makes sense that

this group of nonprofits may see themselves as human service providers primarily, and

are less concerned with fulfilling educational, social capital building, and social change

goals. Perhaps initially these service provider organizations are only pragmatic or

engaged nonprofits (again, referring back to the typology presented in Chapter 1 that

describes those organization that only conduct case advocacy, or case advocacy and

legislative issue advocacy), and they become empowerment-based as a result of gaining

more government funding that creates greater organizational capacities, and/or as the

155

organization learns that public policy education is integral to effective legislative issue

advocacy.

The second constellation of variables in the model, clustered on the lower right

side of the diagram around public policy education, describes another set of factors

associated with conducting public policy education. Perhaps the association of the two

identity variables, grassroots/community organizing and advocacy organizations, signals

that organizations that hold both identities may fall under the definition of a social

movement organization that sees social change as part of their mission (McCarthy &

Zald,1973). Also on the right side of the figure, public policy education is associated with

being an organization with consumers. This is another indicator that many of these

organizations may be social movement organizations that focus on systems change

activities. These organizations appear to either be fueled in part by the passion of

consumers most closely affected by the relevant public policies, or, driven by their

ideology to use an empowerment-based approach that seeks to give voice to consumer

perspectives.

As the diagram in Figure 5 depicts, while both case advocacy and public policy

education are associated with an organization identifying as an advocacy organization,

yet only the grassroots identity was associated with organizations conducting public

policy education. It is most interesting to note that conducting legislative issue advocacy

was not directly associated with having an advocacy identity (but was indirectly

associated through its association with public policy education and case advocacy).

While theorists and national civil society leaders alike claim that advocacy (typically

meaning policy advocacy) is a central function of the nonprofit sector, a number of the

156

nonprofits in this sample that conducted legislative policy activity did not conceive of

their organization as having an advocacy identity, even though they did lobby. This could

be the result of organizations avoiding this label, as these nonprofit organizations may

lack either the knowledge or the nerve it takes to be a political actor. Or again, it could

support the notion that some nonprofits are advocates simply due to pragmatic concerns

regarding obtaining funding for their particular mission. Yet others bring a more

ideological fervor devoted to wider policy and social change to their advocacy activities,

and a more consistent commitment to social change. The fact that the two identities of

being an advocacy organization and grassroots/community organizing were associated

with each other, but the service and advocacy identities were not, provides some support

for these interpretations. If legislative advocacy were indeed statistically associated with

having an advocacy identity, then perhaps that would indicate that more service providers

viewed lobbying more as an integral part of their mission, as opposed to simply a means

of ensuring organizational survival. The data in this study do not provide a clear way to

distinguish whether or not these organizations hold social change and social justice goals,

but an advocacy identity may indicate a mission that includes policy participation for

social change.

For both clusters of organizations, two variables appear to play a role as to

whether or not these organizations elect to conduct legislative issue advocacy. As Figure

5 shows, greater knowledge about the IRS rules for nonprofit advocacy has a positive

effect on a nonprofit‟s willingness to engage in policy advocacy. Indeed, there are many

stories of nonprofits that have been „punished‟ by those in power for various advocacy

efforts that displayed even minor breaches of the playing rules, as well as threatening to

withdraw future government funding (Bass et al, 2007). Berry (2003) found similar

results in his national study of nonprofits. As the diagram posits, the association between

legislative issue advocacy and connection to statewide advocacy colleagues also appears

157

to play an important role here. This association supports the obvious conclusion that the

resources of policy information that statewide networks can provide are essential for the

engagement of local organizations in legislative advocacy. While local nonprofits may be

very aware of what the needs are on the ground, they have little time or energy to figure

out who to target and how to affect legislative change. Perhaps these colleagues also

assure nonprofits that a certain amount of lobbying is acceptable, or simply offer the

comfort that comes with the bit of cover provided by coalition policy work. In addition,

being a social movement organization may reduce the dampening effect such rules can

have on an organization, as the commitment and fervor of being an advocate outweighs

such fears. The actions of many activists, coupled by the fact that the IRS has been

compelled on occasion to intervene to ensure that the rules governing nonprofit political

advocacy are followed, indicates that on occasion activists for political change do not

care what the rules were, or are. This may be for the better or worse, as it is hard to

imagine the civil rights movement, for example, having any success without breaking the

rules and doing what it did to fight for social justice. This latter interpretation is further

supported by the data in that both being an organization with consumers and having a

grassroots/community organizing identity was negatively associated with greater

knowledge of IRS advocacy rules, as it gives the impression that for social movement

organizations knowledge and playing by the rules is not as important as passion and

vision that most often comes with a grassroots identity.

Summary for Research Question 3

158

With the nonprofit organizations in this sample, it appears that while case

advocacy can depend on support from government, public policy education capacity may

typically rely on the kinds of resources akin to those that fuel social movements. The

differences between a more tradition human service nonprofit and one that involves

consumers and is more grassroots in orientation may denote two different sets of

motivations and capacities, or pathways, that leas an organization to conduct all three

advocacy activities to becoming an empowerment-based advocacy organization.

Legislative issue advocacy is associated with a greater knowledge of the rules that govern

nonprofit advocacy, as well support from statewide policy networks. In the final step of

this study, all of these variables are put to various tests to see which ones emerge as

having the greatest impact on how an organization evolves to being empowerment-based.

Findings for Research Question 4.

Research Question 4 was: How useful is an empowerment-based advocacy

theoretical framework in understanding the phenomena of nonprofit advocacy among

human services organizations? The multivariate phase of the analysis addressed this

question by exploring the relationship of key variables hypothesized to be associated with

nonprofit advocacy activities. Essentially, this question leads us to ask how the theory

squares with the findings, and what other refinements to empowerment advocacy theory

159

or hypotheses may be generated from these findings. Three hypotheses were conceived

in order to test and expand the concepts of the empowerment theory used in this study.

The three hypotheses explored were:

1) An organization with an advocacy pattern of an empowerment-based

advocacy organization (EBA) is more likely to have consumers on the board

and staff (empowerment of consumers) than non-EBA organizations.

2) An organization with an advocacy pattern of an empowerment-based

advocacy organization is more likely to have a greater commitment to

diversity (a more diverse staff and greater connections to diverse

communities) than non-EBA organizations.

3) An organization with an advocacy pattern of an empowerment-based

advocacy organization (EBA) is more likely to have social workers on their

staff conducting advocacy than non-EBA organizations.

To test these hypotheses, all of the predictor variables of interest were used in a

classification tree analysis to predict the dependent variable, whether or not an

organization is empowerment-based. The model generated by the tree analysis contained

three variables: income from government, being an anti-violence or anti-poverty

organization, and having organizations with consumers on the staff or board. Because

tree models may not optimally handle continuous predictors (income from government),

these variables were then used to build a logistic regression model with main effects, all

possible two-way interactions, and a quadratic term for income from government. All

160

three of the main effect variables in this logistic regression model were statistically

significant. In addition to being the most important contributor in both the trees analyses

and the regression model, the logistic regression indicated that income from government

had an interesting curvilinear effect. There was no significant lack of fit the for this

logistic regression model. The Nagelkerke R² of the logistic regression model was only a

moderate value of .279.

The logistic regression and the classification tree analyses yield similar

interpretations, and provide partial confirmation of the importance of empowerment

theory to understanding nonprofit advocacy in some human service sectors. Both models

found the involvement of consumers in the organization to be a significant predictor of

empowerment based advocacy, and this finding indeed provides support for a central

tenet of social work empowerment theory: that an empowerment-based advocacy

approach, which concerns advocacy on multiple levels of society, involving individuals,

communities and the social policy apparatus of the state, is indeed associated with

consumer involvement. While there is no causality implied by these findings, there is a

confirmation that these two phenomena are linked in the practice world of anti-violence

and anti-poverty organizations.

Findings regarding Hypothesis 1

Again, the first hypothesis sought to determine whether or not an

empowermentbased advocacy organization (EBA) was more likely to have consumers on

161

the board and staff (empowerment of consumers) than non-EBA organizations. The data

did support this hypothesis, as this variable was selected by the trees analysis to be one of

the three most important variables associated with being an empowerment-based

organization.

The results also indicated the importance of this variable in predicting whether or not an

organization conducted empowerment-based advocacy. The odds ratios in Table 20 (in

the results chapter) show that the variable organizations with consumers was the second

most important variable (OR=4.8) after income from government in terms of predicting

whether or not an organization was empowerment-based, with the odds of being

antiviolence or anti-poverty being about half as large (OR=2.6).

For anti-violence organizations, having consumers on the board or staff of an

organization increased the probability of being empowerment-based for all three levels of

income from government shown in the table by percents ranging from 28-36%. For

antipoverty organizations, the percent of increase ranged from 15% for those with 0%

income from government to 37% for those with 54% income from government. It is

interesting to note that the percent increase for both anti-violence and anti-poverty

organizations was lowest for those organizations with no income from government, again

showing the relative importance of this critical variable.

Again, while no causality can be inferred from the confirmation of this

hypothesis, this finding nonetheless shows the importance of the variable organizations

with consumers on staff or board in understanding the full picture of the organizational

characteristics of nonprofits that elect to engage in a wide range of advocacy activities.

162

Also, while these patterns may differ across the two missions, the positive impact of

involved consumers is still evident. This finding provides some support for a basic

assumption underlying the empowerment theory framework used in this study that

consumer involvement is an indicator of an organization that is guided by empowerment

principles.

Findings regarding Hypothesis 2

Hypotheses 2: “An organization with an advocacy pattern of an

EmpowermentBased Advocacy Organization is more likely to have a greater commitment

to diversity (a more ethnically diverse staff and greater connections to ethnically diverse

communities) than non-EBA organizations. As the trees analysis did not select either of

the two diversity variables used in the analysis as being significant as to whether or not an

organization was empowerment based, there was no support found in this data for this

hypothesis.

While the percent of people of color on the staff or board, and the organization‟s

perceived degree of connections to communities of color seem to be two rather direct

ways of measuring this construct, certainly there are other ways to try to define and

capture an organization‟s diversity or commitment to diversity. The lack of support for

this hypothesis due to this issue of defining diversity will be discussed in more detail in

the next chapter in the section on limitations.

163

Findings regarding Hypothesis 3

Hypothesis 3: An organization with an advocacy pattern of an

EmpowermentBased Advocacy Organization (EBA) is more likely to have social workers

on their staff conducting advocacy than non-EBA organizations. The data in this study

did not support this hypothesis, as the trees analysis did not select this variable as being

significant as to whether or not an organization was empowerment based. This finding

indicates that the presence of advocates with social work degrees is not an integral aspect

of empowerment-based advocacy as it is practiced in these organizations. This may be

due to the fact that a significant number of organizations in this sample identified as

having a grassroots identity, and thus these organizations may have a less

professionalized approach to practice.

While there was no support found in the data for this hypothesis, this may have

been a result of so few organizations employing social workers as advocates. Another

limitation could have been the way this variable was conceptualized. Other measures

could have been used to see if a social work ethos or professional values were imbued in

the organization‟s practices. For example, a variable that assessed whether or not social

workers were present elsewhere in the organization, such as on the board of directors may

have resulted in confirmation of this hypothesis.

Section 5. Serendipitous Finding:

164

In general, the findings from this study lend support to general conclusions found

elsewhere that funding, especially from government, appears to play a key role, and

perhaps the most important role, in terms of what advocacy activities nonprofit

organizations engage in. Resource theory has been used to explain these types of

findings, with the simple notion that without the organizational capacity to do so,

nonprofits are not as likely to engage in the sustained efforts that advocacy requires.

While the results of this study reinforce these conclusions from previous studies that

found nonprofits in receipt of government funding to be more likely to conduct advocacy

activities, these findings also may reveal one important caveat. These quantitative results

affirm what has previously only been found in qualitative studies in the sector; that

perhaps sometimes too much government funding can be deleterious for nonprofit

advocacy (Berry, 2003). This serendipitous finding will be discussed further in the next

section.

The Complexities of Government Support, or, Resource Theory Modified?

The one unexpected finding from this research, discussed in this section, is that

too much government funding may become a damper to nonprofit advocacy at higher

levels. The data from this study confirm previous studies that regardless of the raw

capacity in dollars and staff, an organization with income from government appears to

benefit from the capacity such funding brings (Berry, 2003). This capacity could include

the insights, relationships and access created by a firmer connection to government

165

described by other scholars (Saidel, 2000; Berry, 2003), or the power that staff and money

facilitate (Salamon, 1995; Berry, 2003). Government often brings nonprofits together to

educate them about regulatory requirements and similar issues, and sometimes to ask for

their input into policy decisions. Being a part of these networks is likely to enhance both

a nonprofit‟s stature as a policy player, as well as its knowledge of policy and its impacts.

The positive association between an organization‟s use of statewide policy networks and

their propensity to conduct legislative issue advocacy (figure 5) also advances this notion,

as statewide associations typically provide additional capacity, encouragement and

opportunities for local nonprofits to engage with the policy making process.

Again, the findings show that the amount, or percentage, of funds on either end of

the spectrum from high to low government funding, makes an important difference as to

whether or not an organization conducts empowerment-based advocacy. Both the tree

analysis and the logistic regression identified income from government as playing the

most critical role in predicting whether or not an organization conducted

empowermentbased advocacy. The tree analysis split this continuous variable from 0-

100% at the 26% point, indicting that those organizations with 26% or more of their

income from government were more likely to be empowerment-based. The fact that

income from government was selected by the tree analysis as the most important branch,

and the fact that the odds ratio for this variable was 2.1 at the maximum (the distribution

for the odds ratios for income from government is curved, reaching a maximum at ca.

0.7% income from government), shows that it had a substantial effect on the dependent

variable of being empowerment-based. Clearly, having a significant amount of funding

166

from government, at about the 30% mark, increases the probability of an organization

being empowerment-based, at least for most of the organizations in this sample.

However for some organizations, very high levels of government funding

appeared to suppress empowerment-based advocacy. The predicted probabilities in the

logistic regression analysis showed a curvilinear shape, with the effect of income from

government leveling off at about 54%. If this data is indeed representative of the typical

impact of government funding on human service nonprofits, this could indicate an

important finding: that the impact of funding from government depends on its percentage.

The results of this study showed that in the middle ranges (30-70%) funding from

government is associated with higher probabilities of an organization being

empowerment-based, however, too little or too much government funding is associated

with a reduced probability of an organization having an empowerment-based advocacy

pattern. The implication of this critical finding, and what policy changes should be

considered to deal with this potential barrier to empowerment-based practice, are

discussed in the next chapter.

A competing test of empowerment theory could view organizational identity as

the dependent variable indicating empowerment rather than the advocacy activity itself.

This would assume that a social movement orientation is the best indicator of an

organization being empowerment-based. To explore this different application of the

theory, the advocacy identity of being grass roots/community organizer (as a proxy for

being a social movement organization) was used as a dependent variable in a tree analysis

(using the same standards and process of the other tree analyses conducted in this study,

167

and the same variables). The results indicated that the only significant predictor of

whether or not an organization was grassroots or not was whether or not an organization

conducted public policy education. Again, among those organizations that conducted

public policy education, 51% identified as being grassroots, while only 29% of those that

did not conduct public policy education identified as being grassroots. One could deduce

from this relationship that public policy education was the most crucial advocacy activity

for empowerment-based advocacy organizations, even while it was the least conducted

advocacy activity. As mentioned previously, the fact that public policy education was not

associated with case advocacy is also of concern to those who wish human service

organizations to take a more empowerment-based approach to their service delivery.

Clearly the function of conducting public policy education and having a

grassroots/community organizing identity are closely linked. That mission was selected as

a key variable by the trees analysis and was also a significant variable in the logistic

regression model may be in part due to these sub-populations of nonprofits having

different attachments to the social movements that created them. The interplay and

associations among these variables shows that they may have promise for exploring how

identity, mission and connection may be important variables to explore in future research,

especially as they may indeed be markers of how much an organization is part of a social

movement.

Conclusion

168

One of the key findings of this study is that public policy education, long

associated with the nonprofit sector by many civil society theorists (Salamon, 1995;

Boris, 1999; Dahl, 1998; Reid, 2001), was the least commonly conducted form of

advocacy. This finding is similar to those from other recent research efforts. In recent

research exploring the social work interventions of four self-identified

empowermentbased human service agencies, activities related to policy or political

advocacy, including any reference to activities similar to public policy education, were

notably absent (Everett, Homstead and Drisko, 2007). While leadership and partnership

were activities and stages of the empowerment process that were consciously encouraged

with clients, these tended to involve clients in intra-organizational activities such as

advisory groups. An advisory group, whether semi-autonomous or directly under the

aegis of an organization, could certainly serve as a vehicle for policy advocacy, yet there

was no mention of such action in this study. Abramovitz (2005) found in a study of 105

human service agencies in New York City that advocacy activities had increased in 78%

of the agencies as a result of welfare reform. However, most of this activity was targeted

to policy makers and staff members, and not toward clients or the broader community. It

is striking this recent research found that humans service nonprofits displayed a dearth of

public policy education toward consumers and the general public in one of the most

political cities of the U.S.; such results provide some confirmation of the concerns raised

from the data of this study.

It appears that social work advocacy practices may still lean heavily toward those

that focus on empowerment on the individual level, versus societal level empowerment

169

activities that lead to political action aimed at the pursuit of social justice. Or, if

organizational activities that support clients in policy advocacy are conducted, they have

yet to be documented in such studies. It may also be the case that these efforts are

somewhat underground, or clients are referred outside the agency, as nonprofit leaders

fear or do not understand the IRS rules that allow such activities within their

organizations.

Organizations that identify as being grassroots/community organizers may be

responsible for a good deal of the civic function of public policy education produced by

human service nonprofits. The results show that the advocacy capacity that comes with a

social movement orientation should not be underestimated. Resource theory holds that the

resources of an organization are very likely to determine whether or not an organization

conducts advocacy. Thus one would have expected to find variables that measure basic

capacities of an organization, such as total staff or total amount of expenditures,

positively associated with advocacy activities. This was not the case among these

organizations, although perhaps if this researcher had used the variable Berry (2003) did

in his study that asked whether or not an organization had at least one person who had

some advocacy-related responsibilities, the role of “hard” resources such as staff time, as

well as funding, would have surfaced in this sample as well. Nonetheless these results

helped reveal that more non-traditional “soft” measure assessments of organizational

capacity, such as the capacity inherent in a passionate commitment to social change

among the board and staff of an organization (measured here by a grassroots identity, as

170

well as perhaps by the inclusion of consumers on the staff or board), or their connection

to policy networks, remain integral to understanding the determinants of human service

nonprofit advocacy at the local level.

Does this different, “soft” conception of capacity mean that obtaining policy

related information, and the encouragement and assurance of a coalition urging action, is

more important than whether or not an organization knows the rules of the game?

Perhaps this is true for the grassroots organizations that identify as such. Many observers

of the nonprofit sector have noted that the wealthy interests that one often sees on boards

of directors (donors, or corporate executives) are not typically those who espouse a liberal

politics that one could assume would emerge out of human service organizations

concerned with their clients‟ welfare. Therefore, the effect of receiving a larger share of

organizational income from donations may be more influential than the greater

knowledge that these organizations have regarding the rules of advocacy. Perhaps an

important factor that determines whether or not a nonprofit advocates is what kinds of

people serve on its board of directors. Measuring the socio-economic status of the board

of directors and their political orientation may be another important predictor for future

studies.

In summary, the findings indicate that three variables, mission, having consumers

on the board or staff, and income from government, were the most important

determinants among the variables analyzed in this study of being an empowerment-based

organization. Other variables that show some promise to inform future studies may

include those that were associated with these and the dependent variables, including

171

connection to coalitions or other statewide advocacy organizations, knowledge of IRS

advocacy rules, and identity.

In the final chapter of this dissertation report, the limitations of this study will be

discussed, as well as the implications of these findings for policy, practice, research and

theory development.

CHAPTER V. CONCLUSION

Introduction and Chapter Organization

This research project was conceived as an exploratory and descriptive study of

nonprofit advocacy among a sample of local human service providers that are likely to

conduct advocacy activities. This introductory section of the chapter begins below with a

discussion of the main limitation affecting the results of this study, and then describes

how this determines the manner in which the rest of the chapter is organized.

While the purposive sample used for this study was not representative of human

service nonprofits in general, the sample consisted of a diverse group of nonprofit human

service providers that had a connection to a statewide advocacy organization or network,

making it a rich sample to use in researching nonprofit advocacy. In addition, with a

large number of organizations that identified as being grassroots/community organizing

nonprofits, the sample afforded a view of organizations that are more closely tied to

social movements than the typical human service organizations that are less likely to have

a strong tendency toward social action. The relatively large number of nonprofits in this

172

study was another strength. In spite of the non-representativeness of the sample, the

diversity of organizational characteristics among the nonprofits, and the presumed

tendency to advocate due to the sources of the purposive sample, did yield data that

proved to be fruitful and informative.

Since this was a purposive sample, any firm assertions about causality or findings

that would assume some degree of accuracy in predicting the prevalence of advocacy

among nonprofit human service providers generally are not possible. However, the

findings do provide some confirmation of basic tenets of empowerment theory, and offer

some insight as to what organizational characteristics are associated with empowerment

practices. This study tested some of the underlying assumptions of empowerment theory,

(e.g. that the involvement of consumers will be associated with an empowerment-based

advocacy pattern), as well as identified and refined the variables that attempt to measure

those organizational characteristics in the field that tend to be associated with nonprofit

advocacy. Refining the variables that are used in surveys of nonprofits will help scholars

measure advocacy in more accurate ways, as well as inform researchers how to

specifically identify and locate empowerment-based advocacy organizations for further

study. The multivariate results contribute further evidence that points to the relative

importance of these variables.

Since the limitations of the sampling design mean that any findings from this

study do not warrant the kind of confidence on which to base any strong

recommendations for practice or policy change, it makes sense to discuss the limitations

of this study primarily in the context of examining any implications for theory and

173

research. Hence, the limitations of this study will be discussed mostly in the next section.

Certainly, any proposed course of action to change practice or policy must be based on

more solid ground than this study alone offers. Nonetheless, implications for practice and

policy are discussed in the second section, because, if future research continues to

undercover similar findings, such implications may be worth further consideration.

The third and final section of this chapter briefly summarizes the main findings

from this study and assesses their importance and relevance.

Limitations of study and implications for theory and research

The first variable conceptualization to be discussed in terms of its limitations as

well as theoretical and practice implications is the mission variable.

The results discussed in the previous chapters indicate that mission does indeed appear to

be an important variable that is associated with many other key variables, including the

dependent variable of interest in this study. Clearly the antiviolence organizations, all of

which were members of statewide coalitions focused on issues of violence against women

and children, shared a strong, focused mission. The anti-poverty organizations were not

only more diffuse in their focus, but were also part of a looser network that encompassed

many different missions. Thus, it may be the case that some of the difference found

between the anti-violence and anti-poverty organizations may in part be a result of the

fact that this variable was a name ascribed by this researcher that did not fit the anti-

poverty organizations as closely as the ascribed mission did for the antiviolence

174

organizations. Yet, in spite of these limitations regarding the accuracy of the fit of the

mission variable, the many significant differences found in this data still reinforce the

larger point that a nonprofit‟s mission, or field of practice, appears to play a role in

whether and how a nonprofit conducts advocacy.

Aside from mission, the next most important variable to assess is the construct of

what constitutes an empowerment-based advocacy organization. The first step of this

assessment reviews the definitions of the three core dependent variables of case

advocacy, public policy education, and legislative issue advocacy that are part of the

construct of being an empowerment based advocacy organization.

Both case advocacy and public policy education are activities that are intuitively

rather accessible terms, even if they may be a bit vague to those unfamiliar with either of

them. A definition roughly a paragraph in length was provided to the organizations

answering this survey, and hopefully this increased the accuracy of the responses.

However, given the problems this researcher and others have encountered using the term

lobbying, caution is still clearly needed when defining such activities. Berry (2005)

showed conclusively in his large national study of nonprofit organizations that nonprofit

organizations tend to avoid using the term lobbying to describe any advocacy activity.

Even with this study, which used a more diffuse and approachable term such as

“legislative issue advocacy”, nonprofits in this sample still shied away from noting that

they conducted lobbying, as evident from a significant number of nonprofits noting

elsewhere in the survey that they did engage in lobbying. Again, this lack of

acknowledgement could be from fear, confusion, ignorance, or political naiveté, but

175

nonetheless it shows that how this variable as defined continues to be a critical step in

such research. This suggests that the strategy used in this study, and others, of asking

more than one question about specific, concrete actions to determine if a nonprofit is

lobbying is one that future researchers should consider, rather than only one question that

asks about lobbying in general.

Case advocacy is rarely studied along with other forms of advocacy, as was done

in this research project. Advocacy as defined by most nonprofit scholars is typically

focused on the political actions that target actors in the various branches of government,

or the public, about laws and other public policy matters (Berry, 2003). Political science

scholars would likely contend that case advocacy does not belong in the lexicon of terms

that describe political action, and is a social service. Yet, social work empowerment

theory does indeed see that case advocacy is ideally linked to cause advocacy, and

therefore should be considered one of the necessary core civic engagement functions of

human service nonprofits, in addition to being a social service. At the very least, social

work scholars would claim that such action is critically important when done with, or on

behalf of, those that are traumatized by their circumstances, or in other ways are either

temporarily or persistently politically disempowered. This is especially important in this

country at this time, given the conservative tendencies of the last few decades that deny

the very reality of those in need.

The finding that case advocacy was frequently conducted among the nonprofits in

this sample shows that many humans service providers do recognize that this function is

needed in local communities. From the empowerment perspective, case advocacy is

176

connected to political action because ideally it promotes two change processes: one

process that focuses on larger systems change, and other simultaneous processes on the

individual level that assist people in meeting important needs (Gutierrez and Lewis,1994;

Reisch, 1986; Rose, 1990). When connected to public policy education and legislative

issue advocacy efforts, case advocacy is often the first and perhaps most crucial step

toward individual political empowerment. In both systems and individual change

processes, it is essential to build the capacities of citizens to see their individual problems

as being connected to public policy. Scholars have described the political awakening that

accompanies the growing awareness that one‟s individual problems are intertwined with

the fabric of our society (Sachs & Newdom,1999; Sosin & Caulum, 1983; Sunley; 1983;

Reisch, 1986). Perhaps the finding in this study that case advocacy was significantly used

less by antipoverty nonprofits in the sample is an indicator of, as well as a possible reason

why, redistributive policies are not found at the top of the political agenda.

The other two forms of advocacy studied here, public policy education and

legislative issue advocacy, are both necessary functions for ensuring and strengthening

civic engagement and our democracy (Dahl, 1998). However, some may contest the

notion that these functions should be provided by human service nonprofits, especially

those receiving government funds, as opposed to other entities in our society, such as

political parties, associations, or other membership-based, mutual benefit organizations.

Yet scholars and activists who care about ensuring that the voices of poor, working and

middle classes are heard, and that the needs of charities and the social workers who

within them are heard as loudly as any other interests, strongly assert that more

177

empowerment-based advocacy aimed at ameliorating poverty is needed (Abamovitz,

2005; Lamb, 1992; Specht & Cortney, 1994; Hagen & Davis, 1994; Sunley, 1997;

Fabricant & Burghardt, 1998). This is especially imperative given the shrinking of

political spaces for civic engagement on these issues (Verba, Sclozman & Brady, 1995).

Scholars also have noted that social and political forces have decreased the membership

of, as well as the number of, unions and membership-based nonprofits (as opposed to

those with self-perpetuating boards). The membership of fraternal and other mutual

benefit organizations that traditionally had served as organizational vehicles for civic

engagement in the first half of the 20th century has declined without any equivalent

replacements (Skocpol, 1999; Putnam, 2000). In many areas and policy arenas, the

successful federated structure of the older organizations has not been supplanted by

similar organizations that are organized on local, state and federal levels; (Skocpol, 1999;

Beatty, 2001).

As Imig notes, “advocates can succeed only when grievances are channeled into

political action through organizational networks….In other words, mobilization efforts

need to work through the existing infrastructure of communities (2001, p.203).” This

imperative for local human service nonprofits to be a mobilizing force for civic

engagement, as articulated by Imig and others mentioned above, provides the basis for

examining empowerment-based advocacy in the form that this study uses. The fact that

this definition did successfully capture and describe the activities of many of the

nonprofits in this sample in and of itself provides some degree of confidence in the

manner in which this variable was defined.

178

Another contention that could be raised to critique the construct of the dependent

variable, being an empowerment-based advocacy organization, is whether or not

empowerment-based advocacy activities, however these are defined, must all occur

within the confines of one agency in order for empowerment to occur (as is assumed by

this study). The interpretation of empowerment-based advocacy theories used in this

study is that among human service providers serving vulnerable populations, a program

that seamlessly incorporated all three levels of advocacy within itself is more likely to

produce political empowerment of the people the organization serves. Yet one could

question whether it really is important that all organizations conduct a particular

advocacy activity, as long as all of the consumers and members of an organization are

able to benefit from such advocacy from an accessible organization. While it is an

intensive and more complex undertaking to research inter-organizational collaborations,

this critique has merit, and is one that leads to the conclusion that empowerment-based

advocacy should ideally be examined on the local level in a way that better captures inter-

organizational collaborations, both formal and informal, including policy networks,

coalitions and other forms of collaborations across organizations. The finding in this

study that such policy networks on the state level are statistically associated with

legislative issue advocacy is an indicator that such an approach is warranted. Studies

using methods such as GIS mapping and network analyses are likely to yield more

thorough and refined research results, as such techniques could examine how a

consumer‟s proximity to multiple networks of action and policy practice may affect her

participation and empowerment in the political process.

179

Another issue that must be noted regarding the definition of empowerment-based

advocacy is that some advocacy activities are not captured using only these three

definitions. Most importantly, the function of community organizing and mobilization is

not fully captured in this schema. Future scholars of nonprofit advocacy, including this

one, should consider adding this category to ensure that a more complete range of

advocacy activities is accounted for in this schema. Mobilization is a core element of

democratic political process (Saidel, 2002; Dahl, 1998).

Next, a discussion follows of the limitations of the capacity variables used in this

study, such as source of income, size, and other factors related to an organization‟s

advocacy capacity, as well as the implications of such limitations for future research. In

this section that deals with the implications for research, special attention is paid to noting

when and how variable definition can be altered to facilitate a nonprofit‟s ease on

answering. But before this discussion, first it is important to note why reducing the

number of questions a nonprofits is asked and streamlining the variables is so important.

Nonprofit organizations, which are typically overburdened and underfinanced, are

notoriously difficult to survey, as the rate of return of surveys is typically quite low

(Berry & Portney, 2003). In their study of Indiana nonprofits using multiple state,

national and local sources of lists, Grønbjerg &Paarlberg (2002) found that response rates

varied by the source of the database, with nonprofit organizations that contracted with the

state providing the best response rate. To raise the response rate, it is imperative to

simplify the work a nonprofit has to do to answer survey questions (Berry & Portney,

2003). Given that one of the more important ones, income from government, proved to be

180

so key in this and other studies of nonprofit advocacy, perhaps simplifying the

construction of this variable into a few categories rather than having to specify an exact

percent would facilitate the ease of responding. This is underscored by the fact that

source of income was one of the variables used in this study that had a high number of

missing values.

To facilitate their answer without loss of the gradations of effect the variable

income from government has, the results of this study seem to show that asking a

nonprofit to select one of five categories for income from government: 0%; 25% or

under; 50% or under, 75% and under, and 76-100% would be useful. To help determine

whether or not income from donations either encourages or discourages advocacy, the

inclinations of the board of directors toward or against advocacy should also be assessed.

Researchers should consider asking one or two questions as to whether or not most

donors support an organization‟s advocacy activities directly, or whether donations are in

response to fundraising efforts that highlight the organization‟s social change efforts.

Otherwise, it seems that for the most part, the sources of income variables are fairly

straightforward and easy to understand.

On the surface, other capacity variables used in this study - age, number of staff

and total expenditures - were defined in a relatively straightforward fashion. Perhaps if

the age variable had been used to determine era cohorts, as Salamon (1995) and Childs

and Gronbjerg (2007) did, some significant differences may have emerged due to the

different orientation toward advocacy that organizations that began during the Civil

Rights and Great Society eras have sometimes demonstrated.

181

The identity that an organization selected, grassroots/community organizing,

advocacy or service provider, provided additional insight as to when and why an

organization might conduct certain types of advocacy. Having a grassroots and

community organizing identity could be seen as being a marker of a social movement

organization that views public policy education as a key social change strategy. It is

important to note again that being a grassroots organization was associated with public

policy education activities, which is a finding in alignment with the precepts of

empowerment theory. Another interesting difference again is that the nonprofits in this

sample that identified as advocates were also more likely to conduct case advocacy and

public policy education, but not legislative issue advocacy. Clearly, for many nonprofits

in this sample an advocacy identity fits with both micro-level casework and with more

community-based education efforts, and not with the traditional definition of advocacy

groups that have advocacy and social change as primary goals (Andrews & Edwards,

2004). However, the answer to this question of what identity an organization has is more

subjective in nature than other capacity variables, and thus could simply reflect the

notions of the person answering the questionnaire rather than any organizational

phenomena.

In terms of the predictor variables used in this study, certainly many other ways

exist that could be utilized to define and capture an organization‟s diversity or

commitment to diversity. While percent of racially diverse staff and board and the

organization‟s perceived degree of connections to ethnic communities seem to be two

rather direct ways of measuring this construct, another approach could have tried to

182

measure attitudes. While examination of this construct did not result in any significant

findings, this may have been the result of a weak conceptualization of the measure used

for this variable. In addition, the local context and racial make-up of the surrounding

communities were not taken into consideration at all. Surely comparing the diversity of

organizations located in rural counties of upstate New York to nonprofits in the New York

City area is not an exercise in equivalency, so some attempt to assess and make sense of

the surrounding environment‟s level of diversity would be in order.

Exploration of the use of the predictor using social workers as advocates didn‟t

net any significant results. This may be the result of the professionalization of the field

(Specht & Courtney, 1994). Or, the overarching political climate of the past few decades

which have often been hostile to any advocacy regarding the human services, may be a

factor that restricted or discouraged the involvement of professionals (Abromovitz &

Morgen, 2006; Bass et. al., 2007). Again, this may be the result of too few organizations

that had hired social workers as advocates, or that anti-violence and anti-poverty

organizations are both on average lower paid areas of social work that do not attract

degreed social workers.

However, the variable describing the involvement of consumers did yield

significant results. The confirmation of the first hypothesis that posited a connection

between an empowerment-based advocacy pattern and the involvement of consumers

provides support for the empowerment theory used in this study. Gutierrez (1995) notes

that sometimes the notion of empowerment is often diluted in practice. Simply handing a

183

client a flier on some needed advocacy action is not the same as ensuring their views are

heard and their power felt by openly embracing their involvement in the organization, and

sharing power to define and determine advocacy efforts. Nonetheless, this finding shows

that many nonprofit human service providers do take to heart the more radical

implications of empowerment theory, which is to give voice to, and not talk for, those you

serve.

Nonetheless, it is just as important to assess the organization‟s inclination to value

the perspectives of consumers and their impact and influence in ways other than simply

adding up the number of identified consumers in the organization. Donaldson devised and

used a scale to measure empowerment that delved into several facets of empowerment

(2007). Using both types of measures (counting the number of consumers and measuring

their influence on the organization) in future studies can help identify when and how any

tokenism of consumers could occur. Whether or not the presence of consumers on the

board or staff and the positive association of such presence with conducting

empowerment-based advocacy is a cause or consequence of an organizational orientation

toward empowerment is not determined here. But the simple association again does show

that the linking of these phenomena provides some proof that the theoretical constructs

used in this study have some consistency and meaning, although certainly much more

study of nonprofits over time is needed for any assessment of causality.

The results of this study also point to other hypotheses that should be explored,

especially for those that would help illuminate the critical advocacy function of public

184

policy education that was also the least frequently conducted type of advocacy. One

possible hypothesis that may be generated from the model in Figure 5 is that

organizations that identify as being grassroots are more likely to conduct public policy

education, as they see this function as being critical to empowering local citizens. And

from the opposite direction of causality, perhaps other hypotheses would include

propositions that those organizations providing services are less likely to both conduct

public policy education and see themselves as being a grassroots organization, while

those that conduct public policy education are more likely to select the identity of

grassroots/community organizing. Regardless of how the construct of identity may be

used in future research, these findings underscore the notion that organizational identity

may play a causal role in whether or not an organization conducts advocacy.

Indeed, the term grassroots may be interpreted by human service nonprofits as an

identity that describes a more democratic, authentic and representative organization.

Many staff and leaders of nonprofits may view grassroots/community organizing

nonprofits as the engines that drive social change, even if they do not fit the typical

definition of grassroots used by scholars of being organizations with all or mostly

volunteers, with little money and few paid professionals. The linking of these two

variables, having a grassroots identity and conducting public policy education, is

supported by the trees analysis mentioned in the discussion chapter that, using the same

methods and variables in the other trees analysis, selected public policy education as the

primary predictor for whether or not an organization selected the identity of being

grassroots/community organizing. Dahl‟s (1998) democratic theory explains why such an

185

association makes sense, as his theory deduces that healthy democracies have strong civil

society sectors that engage, educate and enable citizens to participate in civic life (Dahl,

1998; Reid, 2001; Saidel, 2001). This finding further confirms that such theories are

likely to have some explanatory power in this area of inquiry.

In addition, the notion of linking the advocacy activity of public policy education

to the nature of being “grassroots” provides a possible route to solving some thorny

public policy problems. Recent grassroots lobbying campaigns conducted by several

shady “astroturf” organizations, which are often funded covertly by monied interests

(such as the nonprofit organization run by the convicted lobbyist, Jack Abramoff) are

clearly more propaganda campaigns than educational efforts, and are the antithesis of the

robust civic engagement processes described by Dahl (1998) as being so essential to a

healthy democracy. Political pundits warn that the sharp increase in citizen cynicism and

inability to trust any political information coming out of the beltway or elsewhere is

likely to result in the loudest, wealthiest voices winning again (Verba, Brady and

Schlozman, 1995; Rich, 2009). Robust and high quality public policy education could

counter the effects of such cynicism, especially if conducted by a local nonprofit that was

at the forefront of bringing people together, perhaps even in concert with government

when appropriate.

Could public policy do more to encourage the functions of providing high quality

information (often known quaintly as “facts” in the eras before postmodernism), as well

as the offering of opportunities to the general public to constructively deliberate issues

(unlike the recent chaos this past summer orchestrated by political elites nationally to

186

deliberately misinform citizens and disrupt local forums on health care reform) by

nonprofits? How can the important, but perhaps waning function of public policy

education be enhanced and expanded among human service nonprofits? These questions

will be pursued in the next section.

Limitations and implications for social work practice and policy

What do the findings, particularly the findings regarding the patterns of advocacy

activities found in this study imply for social work practice and policy? The most

important observation is that empowerment-based advocacy, at least as it is defined in

this study, is indeed taking place within two major sub-fields of the nonprofit social

service system. While this finding is heartening to some degree for the social work

profession, which encourages and advances empowerment-based advocacy, it also shows

that this approach to practice is by no means universal. The prevalence of

empowermentbased advocacy, especially the provision of public policy education, needs

to be strengthened among human service nonprofits.

While it is true that the majority of organizations in this sample conducted all

three, those that conducted case advocacy directed at individuals and legislative issue

advocacy only (the pragmatic service providers) were highest in number after

empowerment based advocacy organizations. That such a significant number of

organizations displayed a hole in what would ideally be a continuum of advocacy

activities from the level of the individual to socio-political levels is indeed cause for

187

concern. Increasing the number of nonprofits that conduct public policy could improve

the credibility and influence of nonprofits, if they are seen as having quality data about

community needs. This study‟s data support this claim, that bolstering the function of

public policy education and the role of consumers in human service organizations will

increase the frequency of empowerment- based advocacy. Recommendations for policy

change to foster the increase in empowerment-based advocacy, particularly of public

policy education, will be discussed further at the end of this section. First, observations

about these findings that have practice implications will be reviewed.

Again, both the bivariate and multivariate results appear to support the central

presumption of the theoretical framework used in this study: that consumer involvement

is associated with the advocacy practices defined as being empowerment-based (Rose,

1990; Gutierrez, 1995). Government funding also adds to the probability of being

empowerment-based, except at very high levels of funding. Whether the association of

being empowerment-based and having consumers in an organization is due to some

government funding streams requiring such consumer involvement, as certain fields of

service can tend to have such as mental health, or, whether government adds the capacity

needed to support the involvement of consumers alongside its support of an increased

advocacy capacity, is unknown. Nonetheless, one important implication for practice is

that social workers need to recognize, appreciate and effectively use all available

resources for advocacy work, including the vision and energy of consumers, and the

passion for social change inherent in grassroots social movement organizations.

188

If indeed the involvement of consumers plays a causal, or formative, role in the

type of advocacy pattern an organization displays, executive directors of social welfare

agencies would do well to enhance and affirm the importance of consumer involvement.

More attention needs to be paid by both practitioners and researchers to promising

practices of consumer involvement, and models of various means to do so need to be

explicated, developed and evaluated, especially in case advocacy practice. This is a

potential task well suited for statewide advocacy organizations, as these umbrella

nonprofits often play a key role in not only providing technical assistance to the local

organizational members, but they also provide the means for these local organizations to

be involved in state public policy processes.

The data in this study indicate that the above recommendation would appear to be

both critical, and also a very challenging one, for anti-poverty organizations. As

described earlier in this chapter, there are many barriers to effectively involving poor

people in an organization‟s advocacy efforts, but yet it is obviously needed. More radical

groups such as the Kensington Human Rights League (KHRL) and the Social Welfare

Action Alliance (SWAA) offer avenues for potential alliances on both state and local

levels for public policy education work. Yet such coalitions would require some hard

work, as compromises regarding both the policy goals and tactics for achieving them

would likely be required. However, public policy education processes are less

controversial than legislative issue advocacy, as the former by definition allows for a

broader more diffuse approach to engaging the issue by simply providing information in

various educational formats. In Rochester, SWAA holds very successful Reality Tours

189

that expose people to the realities of poverty, and local human service providers do

cosponsor these tours, as these organizations recognize the importance of educating the

broader community about the extent and effects of poverty. More creative cooperative

outreach and education tactics between a broader array of actors and activists in public

policy education efforts could make a difference, especially if linked up where possible to

legislative issue advocacy campaigns.

Also, it appears that having strong connections to statewide coalitions and

networks may play a role here, as the bivariate analyses showed that the more active

antiviolence organizations had a strong connection to a coalition. With the dissolution of

the Statewide Emergency Network for Social and Economic Security (SENSES), there is

not one statewide advocacy organization in New York State with a primary and clearly

articulated focus on poverty as an issue; this fact alone is very telling about the current

state of nonprofit advocacy for the poor. While there are some active single-issue groups,

such as HANNYS and the Homeless Action Network, these organizations tend to focus

primarily on their issue of concern, and they are also severely under-resourced

organizations. The Fiscal Policy Institute has stepped out of its role as a policy think tank

to do some work promoting revenue coalitions, but this effort has only one person

working on it full time.

Under these circumstances, what are local human service providers supposed to

do? Local organizations can press the statewide advocacy organizations themselves to

consider collaborating together on developing a coordinated campaign to address broad

issues related to poverty. These efforts should assist local providers in participating in

190

such campaigns in significant ways, not simply by lending their organization‟s name to a

coalition list of co-sponsors to a legislative campaign. If a few central policy goals could

be specified by some key players, and the efforts of all local providers could be

coordinated locally to do public policy education, more policy change in the direction

needed is a real possibility on both the state and federal levels. One simple place to start

would be for all organizations to help the legal services community to work to remove the

federal restrictions on legal aid offices to do legal case advocacy and class action suits for

the poor. This policy change led by conservative forces was pushed as they knew the

potential power of legal aid to affect poverty policy, and tying the hands of these public

law practitioners has indeed had its intended effect. This policy win (and it seems given

the current administration that this goal should be relatively easy to attain) could give

momentum to a renewed movement for economic justice that could address many issues

crying out for policy reform, especially concerning the TANF program. Given current and

future political and fiscal realities, much more advocacy will be needed in the future to

create & sustain robust and effective human service systems. So, in short, while local

human service organizations cannot address poverty policy alone, they can collaborate

with statewide advocacy organizations and others to educate the broader population,

recruit and prepare the next generation of activists and policy actors with strong public

policy education efforts, and offer various avenues to their various communities of

interest for policy participation that also involve consumers.

Finally, a major implication for practice is that social workers could benefit from

more education about the important role that nonprofits play in the democratic process,

191

especially regarding public policy education, and how empowerment-based advocacy

processes on all levels (micro, mezzo and macro) within and without human service

organizations can result in needed policy changes. The new accreditation standards from

the Council of Social Work education require educating social workers so that they are

aware of various practice contexts, and certainly understanding the organizational and

political contexts that affect practice are part of such a goal. Social workers need to have

a clear understanding of the key role that nonprofit organization play as promoters and

sustainers of democracy. Such understanding is essential for their engagement as policy

actors, and for engaging in policy practice on pressing issues, including the implications

for policy laid out in the rest of this section.

An implication for policy practice is that an overt involvement of consumers can

be a sign that nonprofits are not simply advocating only for organizational maintenance,

but are striving to hear, and represent in an authentic way, the communities they serve.

More findings such as these could be used to counter a criticism thrown frequently by

those who seek to decrease government support for social services: that most nonprofit

advocacy is only done to keep the nonprofit solvent, regardless of whether or not their

services are needed. In their history of federal regulation and legislation overseeing the

voice of nonprofits, Bass et. al. (2007) provide detail on how this conservative framing of

the issue that challenged the right of nonprofit organizations to lobby was used to try to

pass the Istook amendments (that would have denied the right of any nonprofit that

received government funds to lobby the government).

192

This and other attacks on the nonprofit sector at large was driven by a

conservative goal to defund the left. Grover Norquist, who was one of President Bush‟s

leading and very powerful advisors, would often feature in his public appearances a huge

picture of a pig eating at a trough full of money; the money depicted was from the federal

government, and the pig was “the Nonprofit Sector.” The Istook campaign was a

coordinated and intentional effort to muzzle the nonprofit sector, based on the notion that

the sole purpose of nonprofit advocacy was to fatten the organizations themselves,

regardless of what good they may be doing. The fact that the majority of social workers

are employed by the nonprofit sector now, and that so many social work services are

delivered by the sector means that the profession needs to play a leading role in ensuring

the rights of the nonprofit sector to speak are not just protected, but enhanced.

After much effort, the national organizations representing the nonprofit sector as a

whole defeated the Istook amendments that had passed in the House but were stalled in

the Senate, but it was a long, draining, fight that took much effort, all during a time of

major cutbacks in government funding in many arenas (Berry, 2003; Bass et. al., 2007).

In assessing the reasons for the conservatives losing this argument, at least for the time

being, Berry (2003) argues that conservatives underestimated both the power of the

mainstream nonprofits that form the core of social support to local communities, as well

as people‟s attachment to them. This brief review of recent history is a reminder that

policies shaping the role of nonprofits in the public policy process, as well as the amount

of government funding that flows to the nonprofit sector for human services, are very

contentious and politically charged issues (Berry, 2003; Bass et. al., 2007). A clear

193

implication for social work education and practice of the findings in this study that show

public policy education is the least likely advocacy activity to be conducted in this sample

is that social workers must be informed about the democratic principles that undergird the

nonprofit sector‟s constitutional right to have a voice in the policy process.

As mentioned previously, the difference by mission concerning case advocacy and

empowerment-based advocacy raises an important question regarding the political power

of the poor, given that the anti-poverty organizations in this sample did less of both.

Interestingly, anti-poverty nonprofits did have a higher average number of consumers on

the staff or board (the different means were 4 vs. 2; F=4; p=.047). It may be the case that

anti-poverty organizations had staff or board members with less education, or less time or

resources to be involved in advocacy, than the anti-violence organizations, which could

be possible factors affecting an organization‟s capacity to advocate. Anti-poverty

organizations received less funding from government on average than the anti-violence

organizations, conducted less case advocacy, and coupled with other factors limiting

capacity, such as the contentiousness of anti-poverty politics, may face more significant

barriers to involvement in the public policy making process than the anti-violence

organizations.

As Abramovitz & Morgen (2006) note, extensive tax cuts and cuts to social

programs have been enacted over the past thirty years that have severely strained the

ability of the public and voluntary sectors to meet human needs; this shrinkage of funding

has probably also had a deleterious effect on the ability of anti-poverty organizations to

advocate. Given that the continued constriction of funds for services and concrete

194

assistance to those in need have been well documented in the last few decades (Gilbert &

Terrell, 2002), it is hard to imagine any lack of need for case advocacy among the poor

and the organizations that serve the poor. The Congressional Budget Office (2004) notes

that domestic discretionary spending (the part of the federal budget that funds the

majority of social services) decreased to only 3.4% of GDP in 2004, compared to 4.8% of

GDP in 1978. Devolution, privatization, tax cuts, and antagonism to welfare policies in

general have shifted the burden of meeting human needs onto states and local

communities. For all the above reasons, the nonprofit social service sector continues to

face the prospect of cutbacks, and perhaps for many, to the point of extinction (Steuerle &

Hodgkinson, 1999). As Covington notes, “under conditions of continued federal (and

state) neglect, there is a persistent tendency to ask the most marginalized communities to

solve problems not of their own making” (2001, p.51). In this situation, what are the

prospects for the poor to have a voice in the political process?

Using an empowerment approach would necessitate the inclusion of case

advocacy techniques and strategies to ameliorate this lack of benefits, both for short-term

efforts to aid clients, as well as for long term campaigns to convert such case advocacy

into cause advocacy. The data from this study suggest that strengthening case advocacy

efforts among organizations that serve the poor may be needed, since the anti-poverty

organizations were significantly less likely to conduct case advocacy. Since the

antipoverty sub-sample included organizations that by definition provided “hard”, or in-

kind benefits such as food and housing, this difference may in part reflect the nature of

195

the two types of nonprofit services that hold substantively different missions. Perhaps this

lack of case advocacy among anti-poverty organizations is a continued result of the

historic split between the delivery of social services and the provision of income

maintenance. Nonetheless, case advocacy that is connected to cause advocacy, such as

what empowerment-based advocacy ideally provides, is likely to be needed regardless of

the agency setting or type of service or benefit delivered.when negotiating to ensure that

the needs of the poor are met, especially given the ongoing struggle to put human needs

on the national agenda.

With the entitlement to welfare eliminated for poor women and children, the legal

underpinnings (based on the constitutional right to due process) that made case advocacy

so much more effective in the past are gone, making the need for policy change even

more essential. With an entitlement to basic welfare, poor parents could at least have due

process rights to force the state to provide for their basic needs if local social service

agencies denied them aid, which they often did. With that legal right gone, the number of

families who have fallen into deep poverty (under 50% of the federal poverty level) has

risen (Center for Budget and Policy Priorities, 2007). Prior to the implementation of the

TANF program, Legal Aid used to play a pivotal role in local communities in ensuring

that poor clients had access to entitlement based income support such as AFDC. In

addition, the federal policies that both cut the funds to Legal Aid nationally, as well as

forbid it to conduct any class action work or legislative issue advocacy, even with its own

funds, have been another hard blow to the social services system. Using the findings from

their large study on civic engagement, Verba, Scholzman and Brady (1995) conclude that

196

the inability of the poor to put their needs and basic human rights on the national agenda

and have them addressed creates a vicious cycle of cynicism and apathy among the poor

about the prospects for political change. If social service agencies are indeed so strapped,

as Abramovitz (2005) and other scholars have found, and they feel hampered or

intimidated by external or internal rules such that they cannot advocate for the needy or

educate and mobilize for policy change, then the only conclusion one can draw is that our

current social service system is woefully inadequate in terms of the ability to advocate for

poor clients. This dire situation clearly calls for policy reform.

Because government funding comes with a prohibition against using those

specific funds for lobbying, many scholars and practitioners argue that such restrictions

and reporting requirements have a deleterious effect on nonprofit advocacy, especially

given the confusing, overlapping morass of local, state and national regulation of

lobbying (Bass et. al., 2007; Berry, 2003, Paget, 1999). These prohibitions may be the

reason that the logistic regression results of this study found that high levels of

government funding suppressed nonprofit advocacy to some degree. That such a

dynamic appears to be revealed by this and several other recent studies is of great concern

to those who care about human needs that should be articulated and addressed by civil

society advocacy (Dahl, 1998; Berry, 2003; Child & Gronjberg, 2007).

In addition, campaign finance reform has utterly failed to have the intended

impact of mitigating the power of moneyed interests, given their vast sources of wealth

and access to policy makers that lends them so much power (Phillips, 2002; Krugman,

2008). It would not only simply be a shame if only the most well-heeled nonprofit

197

organizations holding the biggest megaphones could compete with the interests of the

wealthy to have their concerns heard by policy makers, but as Dahl (1998) notes in the

final paragraph of his seminal study on the democracies of the world, it could result in the

demise of U.S. democracy. Economic policies that favor the wealthy have long term

consequences for the health of our society, as economic devastation in inner cities and in

rural areas is fostering epidemic levels of substance abuse, child abuse and neglect, and

crime that eat away at the core of our society, and create generations of wounded people

unable to even subsist, let alone participate as citizens in the revitalization of their

communities. Yet the findings here imply that those nonprofits dependent solely or

mostly on government funds, such as those that represent the poor, or unpopular causes

such as the fight against sexual assault, may have no megaphone, or one with little

amplification. Verba, Scholzman and Brady (1995) conclude in their major studies of

civic engagement across race and class that the realization of a vibrant democracy that

includes, hears and is responsive to its poorest and most vulnerable members remains

elusive.

As Abramovitz & Morgen (2007) argue, as the cost of entitlement program swell

and crowd out discretionary spending, without stronger support in this country for

returning to a progressive taxation scale in order to create more funding for social

programs, many programs and agencies will evaporate in the next few decades. In the

U.S., we face either a dramatic shrinking of our social service system, or a massive

restructuring that shifts to a system that either houses social services within remaining or

new entitlement programs (perhaps as part of a more universal health care system), or

198

expands mutual support, cooperative and other forms of volunteer-based social services.

Are local human service providers a part of addressing this situation, or are they shirking

or unable to fulfill their purported roles as supporters and facilitators of democratic

participation, and of representing and mobilizing unpopular or unheard views? The data

in this study show that in this sample of local nonprofit human service agencies (a sample

that, as noted earlier, is biased in the direction of conducting advocacy due to their

connection to statewide advocacy organizations), only roughly half were

empowermentbased advocacy organizations. Since previous studies of nonprofits

advocacy have found much lower numbers of legislative issue advocacy, the higher rates

of advocacy found in this sample may be viewed as encouraging. Yet in light of the great

need for policy change, this author argues that much can and should be done to promote

more empowerment-based advocacy, particularly among those organizations serving the

poor.

That the data in this study indicates that nonprofits with a very high degree of

government funding may indeed be dissuaded from participating in the public

policy process provides support for Berry‟s (2003) bold and provocative proposal

that government limitations and restrictions against lobbying (but not

electioneering) should be completely removed for charitable nonprofits, because

their voice needs to be heard in public policy debates. Berry‟s (2003) overall

finding from his national representative study is that while many nonprofits do

participate in the policy process, they do so in small amounts, and often in limited

and tentative ways. His findings, as well as the findings in this study, add to the

199

cumulative scholarship in this area that finds that more often than not, all but the

largest and most sophisticated local nonprofits are only marginally, if at all,

involved in public policy processes, which obviously makes the political power of

the nonprofit human service sector much less than the power of moneyed interests

arrayed against them. Dr. Jeffrey Berry‟s recommendations carry much weight,

at least in academe, given that he is a leading national scholar in political science

whose life work for over 40 years had been to understand the power and influence

of interest groups in the U.S. political process. Again, the finding in this study

that showed that when government funding is at very high levels, nonprofits start

to participate less, provides some confirmation of Berry‟s proposal that policy

change is needed to address the low level of political engagement found in the

nonprofit sector.

In addition, many civil rights leaders have noted that the passage of time,

professionalization of the movement, and the abdication of public policy education efforts

via citizenship schools and other means has not been conductive to continued agitation

for policy change for the poor (Guinier, 1999). In a lecture in 1997 at the University at

Albany, soon after Dr. Mary Joe Bane resigned from second in command at the U.S.

Department of Health and Human Services in protest after President Clinton signed the

Personal Responsibility and Welfare Reform Act of 1996, Dr. Bane described how baffled

she and her colleagues were that any outcry against the proposal that an entitlement to

welfare be eliminated for the poor become law was so weak as to seem like just a whisper

in Washington, D.C. Indeed, the dearth of any significant advocacy activity against the

200

demise of this important piece of our national safety net was a marker of just how far the

issue of poverty has fallen off the radar screen in this country, as well as being an

indicator of how anti-poverty advocates are often as ignored, underresourced, and

disenfranchised as the poor (Bane, 1996; Gilbert & Terrell, 2002).

This inability to mobilize the poor, and the general population, against policies

that breed inequality and toward those that reduce it, are in part the result of our U.S.

culture that values individual rights over collective responsibility (Gilbert & Terrell,

2002). But, these results are also due to the superior advocacy capabilities of the very

well-financed, well-coordinated campaigns spawned by the conservative movement

(Covington, 2001; Paget, 1997; Bass et al, 2007). The Istook campaign to both defund

and muzzle the left through taking away the rights of nonprofit organizations that

received government funds to lobby, which came close to winning, is a clear example of

this movement (Bass et al., 2007). Against this political backdrop, Berry‟s (2004)

proposal that the legal limitations on charitable lobbying be eliminated clearly has great

appeal.

However, an unintended consequence of such a course of action might leave

nonprofit advocates that lobby only for increased funding, without educating the public

about community needs, more vulnerable to the charge by conservatives that their

lobbying campaigns are elite and only concerned with organizational longevity, rather

than with broader public policy concerns. This criticism would carry more resonance if

such lobbying campaigns reflected the hollow donut hole shown in these study results,

and resulted in an increase in nonprofit lobbying that was not accompanied by any public

201

policy education efforts. Currently, Berry (2003) argues that local nonprofits enjoy a

special status on the local level, in that they are generally regarded as community-based

agents that are most concerned with the general welfare of the local communities they

serve. This author believes that if nonprofit lobbying increased sharply as a result of such

easing of the lobbying restrictions, and was unaccompanied by local educational efforts

that provided solid empirical evidence for their claims, as well as a somewhat

representational process of deliberation, then the relatively honorable regard and

reputation most nonprofits enjoy could easily be sullied by the charge of elitist and

nonrepresentative lobbying. Instead, if lobbying by nonprofits is viewed by the general

public as a thoughtful, informed, judicious and caring response to community needs, then

the label of “special, pork-barrel interest” that will inevitably be thrown at them is less

likely to stick.

Therefore, instead of lifting the restrictions on lobbying for charitable nonprofits

that Berry (2003) recommends, this researcher proposes that the U.S. should create public

policies that strengthen and expand the political and social mechanisms for high quality

public policy education conducted by and within civil society. But how can public policy

education conducted by nonprofit human service providers be nurtured and expanded? By

both focusing on increasing the number of consumers in organizations, as well helping

nonprofits build the capacities and energy to engage in public policy education, the

number of empowerment based organizations that conduct a wider array of advocacy

could be increased. Linking such efforts to the rich sources of information government

collects could be added to strengthen such citizen policy education campaigns.

202

The national efforts to increase the number of nonprofits that lobby by educating

them about how to lobby have included education efforts that let nonprofits know that

such activities do not fall under the definition of lobbying, and that they are able to

conduct unlimited amounts of public policy education. Perhaps with this knowledge, as

well as funding to strengthen their public policy education capacities, local nonprofit

organizations will become more empowerment-based in their advocacy efforts. Funding

of national umbrella groups such as the National Association of Social Workers and other

national groups that represent human service providers and professionals to reach out to

their members and educate them about the importance of empowerment-based advocacy

practice could be one way to strengthen the sector‟s advocacy capacity.

It is beyond the scope of this report to present a detailed proposal here, but the

rough idea is that the federal government should make funds available that facilitate

public engagement of populations that are currently disenfranchised in the policy process.

Nonprofits or consortiums of nonprofits could apply for funding, as long as they agreed

to offer public policy education along with sane, constructive, and fair community spaces

for the discussion and deliberation of public policy issues. While the focus would be on

strengthening the capacity and support to local human service providers to engage in

public policy education, this program could encourage an empowerment-based advocacy

approach that also includes case and legislative issue advocacy. Foundations such as the

Pew Charitable Trust have developed effective models for the discussion and deliberation

of pressing policy issues among citizens, and each of these is characterized by a strong

emphasis on the inclusion of high quality information coupled with a constructive process

203

for dialog among a diverse and somewhat representative group of citizens over a

significant length of time (in other words, successful efforts do not consist of a series of

one-shot brief lectures). While certainly economic and environmental issues would be at

the top of such a list for a national program, the intertwined social problems of poverty

and violence should also be in the top ten. It is time to fund such civic educational and

engagement endeavors on a large, national scale that attend to some of the most pressing

national policy problems.

Human service organizations or local coalitions that represent them could play an

important role in such an initiative by doing more to raise awareness and educate the

public on the extent of human needs and social problems in their communities, especially

if such initiatives also supported the state and national umbrella advocacy groups that

have the capacity to conduct or compile research, and digest and prepare it for policy

makers, providers and the general public. An initiative like this could dovetail nicely

with the service learning and volunteer promotion programs currently under development

by the Obama administration, and would also offer an opportunity for social work

educators to play a role by offering curricula, training and higher education programs and

action-based research that provides the data and develops the skills needed by

professionals to implement these goals. Such university-community collaborations have

been successful in promoting public policy education, and the lessons learned from them

can inform the shape of such a policy (Kellogg Foundation, 2005).

To avoid charges of such efforts simply being propaganda campaigns, any public

programs or policies intended to strengthen the public policy function of the nonprofit

204

sector must use facts based on as solid evidence as possible. Mechanisms to provide such

sources of information, such as easy access to government documents, particularly those

produced by the independent oversight government agencies, should be provided in a

coordinated effort. Programs that incorporate mechanisms to involve consumers, selfhelp

and mutual benefit networks, the poor, and those affected by the public policies being

deliberated should be given priority for funding. It is also fortuitous that at this time,

efforts to advance evidence-based practice are well underway in many arenas of human

services and the social work profession, and the knowledge about promising practices in

advancing evidence-based practice gained from these efforts would ideally assist this

initiative.

Such a national initiative to expand public policy education and community

organizing on the grassroots level would not only increase civic engagement, but would

also likely result in an increase in the social capital and civic health of local communities.

In this era of shrinking resources, cooperatives & mutual benefit organizations are seen

by many as the wave of the future for all kinds of civic improvement endeavors. The role

of mutual aid in the web of social support is one that the social work profession has long

recognized, and this critical function is one that this proposed policy initiative should also

support. That self help efforts frequently lead to personal empowerment and greater skills

that translate to group work of all kinds bodes well for this policy trajectory.

The democratic skills involved in the self-governance of mutual aid efforts, and

consumer involvement with social service agencies both directly further civic

engagement, and continue the move currently underway in the social work profession that

205

rejects the passivity inherent in the traditional medical model (i.e. a model of service

delivery that views those it serves as patients rather than active aware consumers). Public

and private cooperation should be an essential part of any campaign to strengthen the

civic awareness and involvement of citizens and consumers. It may be that future

researchers will find, as some recent research has implied, that increased political

empowerment and engagement have marked and direct positive benefits on the health of

individuals that are more engaged and active (Putnam & DeWeever Jones, 2001; Klar &

Kasser, 2009).

Certainly, constructive progress toward the ideal political system articulated by

Dahl (1998), one that reflects adherence to the democratic institutions he describes as

being central to a vibrant and health society, is sorely needed. Effective and strong

advocacy and social change efforts by those people and organizations concerned with

social and economic justice is the only antidote to address the imbalance of power and

wealth that currently exist. In his exhaustive history of the rich, Kevin Phillips concludes:

As the twenty-first century gets underway, the imbalance of wealth and democracy in the

United States is unsustainable, at least by traditional

yardsticks. Market theology and unelected leadership have been displacing

politics and elections. Either democracy must be renewed, with politics

brought back to life, or wealth is likely to cement a new and less

democratic regime- plutocracy by some other name. Over the coming

decades, American exceptionalism may face its greatest test simply in

206

convincing the American people to continue to believe in its comfort and

reassurance. (2002, p. 422.)

Summary

Human service organizations are an important part of civil society, and as such

have a critical role to play in advancing civic engagement, especially for the poor and

oppressed. In addition, the impact of the conservative movement and the continued

shrinking of the US economy for the foreseeable future make the presence or absence of

political power a most critical issue for this sub-sector of civil society. One would think

that local human service nonprofits would engage in massive grassroots organizing to

counter the prevailing political winds that have produced so many cutbacks in the last

three decades, especially the cutbacks in services and resources to the poor. However, this

and other studies show that this may not be happening. For all of the above reasons, the

prevalence and vibrancy of civic engagement in the human services is essential, yet

achieving the strong political empowerment of both social welfare consumers and

professional social workers still seems to be far out on the horizon. Certainly the

conservative forces against social welfare are extremely well-funded and sophisticated,

and they benefit from having staked out a very simple position (less government, and of

course, less taxes), while progressive forces are under-funded, overburdened, and struggle

with having to prove both the need and the worth of these programs that address complex

human needs, which are often those of minority populations easily ignored by the

mainstream.

207

It seems that the story of advocacy in the human service nonprofit sector, while by

no means dismal, is still a story of more opportunities lost than gained. It is therefore

time to pay closer attention to the potential that local social service organizations have to

ensure they achieve their missions via involvement in the public policy process through

more deeply democratic means, and to do their part to increase the political power of

those they serve.

This chapter reviewed the findings of the study and the implications of these

findings in two sections. In the first section of this chapter, a discussion of the

implications for theory and research were discussed at the same time as the limitations for

the study were reviewed. The next section provided a discussion of the implications for

social work practice and policy. A brief summary of the main points in each section is

provided below, followed by a statement of the particular contributions this study offers

to the field of social welfare studies.

The importance of the role of mission as a variable in the constellation of factors

associated with being an empowerment-based advocacy organization was clearly

demonstrated by both multivariate models that found the mission variable to be a

significant predictor of being an empowerment-based advocacy organization. In this

sample of organizations, the fact that the anti-violence organizations were much more

likely to identify as being an advocacy organization than the anti-poverty organizations

indicated that the anti-violence programs appeared to have a more social change

orientation, perhaps because the crest of women‟s movement that spawned these

208

organizations was not as far in the past as the anti-poverty organizations. The greater

connection of the anti-violence organizations to formal statewide coalitions, as well as the

almost universal inclusion of case advocacy as a service, may have been other factors that

caused them to identify as advocates. Most importantly, more anti-violence organizations

were empowerment-based advocacy organizations. This may indicate that the anti-

violence movement as a whole has developed alternative service organizations that retain

a good deal of the movement vision and fervor that spawned them.

Of great concern is the finding that lower levels of empowerment-based advocacy

were found among the anti-poverty organizations in this sample. This finding can be seen

as being the result of several factors, including the continued political unpopularity of any

advocacy focused on redistributive policies in the U.S., the higher level of donations to

anti-poverty organizations that may promote an inclination to not disrupt the status quo

via advocacy, as well as the cutbacks in funding to this subsector of human service

nonprofits. That fewer anti-poverty nonprofits were empowerment-based appears to

indicate a need for increased support and encouragement of advocacy, particularly case

advocacy and public policy education. Again, anti-violence organizations were more

likely to receive a greater percentage of their funding from government. In sum,

contentious politics and lower capacities, such as lower education levels or other

resources of the board and staff, may explain why fewer anti-poverty organizations were

empowerment-based.

This study shows the utility of using a definition of empowerment-based advocacy

in research that incorporates multi-socio-political levels. This author asserts in this

209

chapter that case advocacy should be measured and seen as an essential part of the

political process within social welfare scholarship. Empowering case advocacy helps

people meet basic needs as it simultaneously helps them see the big picture of how social

problems are often either caused or exacerbated by social and political structures, and

crucially, how to step into the big picture and change it. An empowerment approach to

service delivery ideally not only ensures that case advocacy is made available to

consumers of human services, but would also challenge the power relations inherent in a

professional-client relationship by including consumers as citizens in the organization‟s

higher level advocacy processes. Such a practice approach seeks to dissolve what is seen

as a false dichotomy between individual and systems change (Sachs & Newdom, 1999).

When connected to public policy education and legislative issue advocacy efforts, case

advocacy is often the first steps toward individual political empowerment. This was the

first study this researcher could locate that measures case advocacy alongside other forms

of advocacy conducted by nonprofits. Most importantly, the data in this study show a

significant number of nonprofits in this sample engaged in empowerment-based

advocacy.

However, some important limitations to the definition of empowerment-based

advocacy used in this study were identified. First, the function of community organizing

itself was not included in the definition of advocacy activities. Secondly, to more

completely understand what is happening on the local level, advocacy needs to be studied

in an inter-organizational context. Nonetheless, the manner in which empowermentbased

advocacy was defined in this study did capture and describe advocacy conducted by the

210

nonprofits in this sample, and shows some promise as a definition to be used in future

research.

There were other lessons learned from this study that scholars should consider in

formulating future research in this area. To facilitate ease of answering and thereby

hopefully increase response rates, the variable income from government could be split

into five categories, using quartiles and adding the variable, no income from government,

without losing the variable effect it has on advocacy. Using the identity variables in

future research on nonprofit advocacy may shed some light on the degree of orientation

or connection an organization may have to a social movement that fuels social and

political change efforts. Other methods of measuring the construct of diversity may reveal

that the racial and ethnic make up of an organization may indeed be a marker of its

orientation to advocacy activities. Perhaps assessing and considering the role of how

diverse the surrounding environment may be would be a related factor to consider.

The variable organizations with consumers was found to play a key role in the

constellation of factors that predicted empowerment-based advocacy in the multivariate

models. Making this variable a dichotomous one also facilitates the ease of answering for

a nonprofit organization. However, other more qualitative and sensitive measures must

be used to assess the degree to which client perspectives are incorporated in the direction

and substance of an organization‟s advocacy goals.

This study offers perhaps the first quantitative analyses showing that in addition

to very low levels of government funding negatively affecting a nonprofit‟s participation

211

in the policy process, very high levels may have a deleterious effect as well. This finding

demonstrates the need for continued education about the rules of nonprofit participation

in the political process, but also indicates a need for policy reform. This author argues

that government funding directed toward human service providers as well as other

nonprofits for high quality public policy education would help involve both nonprofits

and the populations they serve, and would facilitate the raising of voices that are not

currently heard in the policy process.

Overall, the importance of the findings generated by this study shows that

empowerment-based advocacy practice is indeed evident in two major subsectors of the

social welfare sector of nonprofit organizations. Thus, the empowerment theory used in

this study shows potential for further use in research on advocacy. In particular, the

proposition that there exists a unique role for consumers in organizations, and that these

consumers may play a role in influencing whether or not a nonprofit service provider

conducts certain patterns of advocacy, was reflected in the finding that organizations with

consumers were significantly more likely to be empowerment-based. At this point, the

data from this study do not clearly indicate whether government funding is the primary

force that drives the likelihood of an organization being empowerment based, or whether

the addition of consumer involvement, via government funding or otherwise, is the

driving force. Indeed, these phenomena may be intertwined.

While the design of this sample and study does not allow any causal inferences to

be made, the important role of the three key variables used in the multivariate model, an

organization‟s mission, the degree of funding from government, and whether or not

212

consumers are involved, certainly contribute to the growing literature in the field of social

welfare advocacy. Lastly, the finding that less public policy education is conducted

among the politically active nonprofits in this sample is cause for concern, and perhaps

may warrant some policy change to address this hole in the continuum of advocacy

activities.

213

Appendix A. Variables Table

Full variable name

(as referred to in text)

Type of

Vars

Values& labels

Description

survey number

nominal

tax status

nominal

type of organization

nominal

mission

nominal

0=Anti-poverty

1=Anti-violence

Recoded from old values

(1=AV; 2=AP)

issues of interest

nominal

1=yes

2=no

iv35a- general; iv35a2-

poverty/ econ devel; iv35a3-

violence against women;

iv35a4- other

Organizational

Identity

nminal

0=No; 1=Yes

Transformed original,was

1=yes, 2=no

“ “

nominal

0=No; 1=Yes

Transformed original from

1=yes, 2=no

“ “

nominal

0=No; 1=Yes

Transformed original from

1=yes, 2=no

Information Source

statewide advocacy

connection

nominal

0=No; 1=Yes

Transformed original from

1=yes, 2=no

coalition connection

nminal

0=No; 1=Yes

ConnCoal (ii.22e ) original

values:1=yes, 2=no

Knowledge Advocacy

Rules

knowledge of IRS rules

nminal

(dichot)

0=Not

knowledgeable; 1=

Very/somewhat

know.

takes „h‟ election

nominal

(dichot)

0=No, or not sure;

1=Yes

Source of Income (%)

% income from

government

Cont.

0-100%

% income from

donations

Cont.

0-100%

214

% income from

foundations

Cont.

0-100%

% income from fees

Cont.

0-100%

% income from United

Way

Cont.

0-100%

% income from other

Cont.

0-100%

Capacity Vars

Age (as of 1999)

Cont.

Total Staff

Cont.

Total Expenditures

Cont.

12.

Size

Ordinal

Small= under

$499,999

Medium=

$500,000999,999.

Large=$1,000,000+

Diversity vars

% Board ethnicity

% Staff ethnicity

% Organizational

ethnicity (only

organizational ethnicity

used for multivariate)

Cont.

0-100%

Represents the value of the

percent of diverse staff (& if

the staff value is missing- it

took the value of the percent

of diverse board members)

Categories of

organizational ethnicity

(by %)

0=0%; 1=1-20%;

2=21-50%;

3=51-99%; 4=100%

15a

Connection to all ethnic

communities

Ordinal

(for MV,

Cont.)

Values: 2-16; Includes

connection to: AA

orgs & communities;

Hispanic “ “, NA “

“, Asian “

Values for how well an org is

connected to a community

were assigned as follows: Not

at all=1; Somewhat=2

Moderately=3; Very=4

15b

Connection to

AfricanAmerican &

Hispanic

Comms

Ordinal

(for MV,

Cont.)

Values: 2-8; Includes

connection to: AA

orgs & communities;

Hispanic “ “ “

Same as above

Organizations w/social

workers as advocates

Nominal

(dichot)

0=No; 1=Yes

17.

Organizations with

consumers

Nominal

(dichot)

0=No; 1=Yes

215

Dependent Variables

Case advocacy

Nominal

(dichot)

0=No; 1=Yes

Public Policy education

Nominal

(dichot)

0=No; 1=Yes

Legislative issue

advocacy

Nominal

(dichot)

0=No; 1=Yes

Empowerment-based

advocacy organization

Nominal

(dichot)

0=No; 1=Yes

Students also viewed