Dynamics of a virtual team

Rochester Institute of Technology RIT Scholar Works

Articles

2010

Team virtues and performance: An Examination of transparency, behavioral integrity, and trust Michael Palanski

Kahai Surinder

Francis Yammarino

Follow this and additional works at: http://scholarworks.rit.edu/article

This Article is brought to you for free and open access by RIT Scholar Works. It has been accepted for inclusion in Articles by an authorized administrator of RIT Scholar Works. For more information, please contact [email protected].

Recommended Citation Journal of Business Ethics, 2010

Team Virtues and Performance:

An Examination of Transparency,

Behavioral Integrity, and Trust

Michael E. Palanski Surinder S. Kahai

Francis J. Yammarino

ABSTRACT. Virtue-based research in business ethics

has increased over the last two decades, but most of the

research has focused on the actions of an individual per-

son. In this article, we examine the associations among

team-level virtues using data from two studies. Specifically,

we investigate whether transparency (usually thought to

be an organizational- or collective-level construct),

behavioral integrity (usually thought to be an individual-

level construct), and trust (usually thought to be an

individual-level construct) can be conceptualized and

operate at the team level of analysis and, if so, what their

relationships are to team performance. Using Partial Least

Squares (PLS) analysis, we found in both studies that team

transparency was positively related to team behavioral

integrity, which in turn was positively related to team

trust. We also found evidence of a positive relationship

between team trust and team performance. Implications

of these findings for future teams and ethics research are

discussed.

KEY WORDS: integrity, performance, teams, transpar-

ency, trust, virtues

The recent spate of organizational and individual

ethical lapses in business and government has been

met with an ever-growing chorus of calls for

increased ethical awareness and action. As a result of

such calls, organizational researchers have begun to

consider different ethical frameworks for, as Chun

(2005) points out, most research in applied business

ethics has focused on Kantian (rule-based) or Utili-

tarian (cost/benefit) theoretical approaches. Over

the past decade, an increasing amount of research has

considered a more virtue-based approach (Solomon,

2003). An approach based on virtues, which are

simply dispositions that constitute good character

(MacIntyre, 1984), focuses on positive patterns of

behavior across time and situations.

Despite this burgeoning interest in virtues, most

research has tended to focus on the virtues (or lack

thereof) of individual persons, or on the virtues

(or lack thereof) of organizations, or on the relation-

ship between the virtues of individuals and organi-

zations (e.g., Audi and Murphy, 2006; Cameron et al.,

2004; Caza et al., 2004; Chun, 2005). Although ethics

research at the individual and organizational levels of

analysis is undoubtedly important and potentially

fruitful, we believe that that limiting research to these

two levels omits another potentially important level of

analysis: the team, defined as two or more interde-

pendent individuals who work jointly to accomplish

one or more tasks. Given the ever-increasing

importance of teams in the knowledge-based econ-

omy (Baker et al., 2006), a great amount of research

has focused upon the performance and effectiveness of

teams (Guzzo and Dickson, 1996). Much of this

research has focused on team-level performance

antecedents such as team efficacy and potency (Gully

et al., 2002; Howell and Shea, 2006), and newer

research has focused on ethics within teams

(Schminke et al., 2002), but, to our knowledge, no

research has yet examined the effects of team-level

ethical constructs on team performance. In response

to this lack of exploration, we propose that consid-

eration of team-level ethics is important, and that a

virtue-based approach is a viable option for doing so.

Given that virtues typically have been considered

in frameworks conceptualized at either the individ-

ual or organizational levels, an examination of team

virtues should begin with a usable conceptualization

at that level of analysis. Thus, we draw on research

about multiple levels of analysis (cf. Dansereau et al.,

Journal of Business Ethics � Springer 2010 DOI 10.1007/s10551-010-0650-7

1984; Klein et al., 1994) to develop a theoretical

framework in which to explore team virtues. We

start with the assumption that virtues are funda-

mentally isomorphic – that is, they have the same

basic structure and function across levels of analysis

(Kozlowski and Klein, 2000). Specifically, we

develop a model which considers key virtues at the

team level of analysis: transparency (usually consid-

ered as an organizational- or collective-level virtue),

behavioral integrity (usually conceptualized as an

individual-level virtue), and trust (usually considered

as an individual-level virtue).

We begin by reviewing the literature about levels

of analysis and each respective virtue (transparency,

behavioral integrity, and trust). We next develop an

initial theoretical model which considers the rela-

tionship among these variables; specifically, we

propose that team transparency leads to team

behavioral integrity, which in turn leads to team

trust and finally to team performance. We then test

this model with data from two studies. Study 1 is a

longitudinal study of temporary work teams, while

Study 2 is a field study of nurses in a healthcare

network. We conclude with a discussion of the

results and future directions for research.

Literature review and theoretical

development

Levels of analysis

Yammarino et al. (2005, p. 882) define levels of

analysis as ‘‘the entities or objects of study.’’ As Klein

et al. (1994) point out, organizations inherently

include multiple levels of analysis, as individuals may

work alone, in dyads, in teams, and/or in organi-

zations, all of which may in turn interact with other

individuals, dyads, teams, and organizations. Many

variables of interest may be considered at different

levels of analysis. For example, performance is often

considered at individual, dyad, team, and organiza-

tional levels. In contrast, other variables of interest

are exclusive at a particular level of analysis. Group

cohesion, for example, by definition is conceptual-

ized exclusively to the group level of analysis.

Over the last two decades, there have been

increasing calls for more explicit consideration of

levels of analysis. Explicit consideration of levels of

analysis is a critical task for two primary reasons

(Yammarino et al., 2005). First, failure to consider

levels of analysis leaves theory building and theory

testing incomplete, and may lead to incorrect con-

clusions. For example, consider a company whose

sales are lagging. Senior management believes that its

sales force is unmotivated and decides to send all

salespersons for individual motivational training (a

response aimed at the individual level of analysis).

After much time and effort, senior management is

dismayed to discover that sales have not improved.

Unknown to senior management, the problem was

not the motivation of individual salespersons, but

rather the existence of red tape and needless rules that

inhibited the entire sales team (a problem which

occurred at the team or organizational level of anal-

ysis). Second, consideration of different levels of

analysis may provide important insight. For example,

Yammarino et al. (2005, p. 881) point out that a

revolution occurred in physics when physicists pro-

posed, and subsequently demonstrated, that quantum

mechanics operate at a level lower than the atomic

level.

Klein et al. (1994) show that it is necessary to

consider three elements when conducting multi-

level research: the level of theory, the level of

measurement, and the level of statistical analysis.

First, the level of theory describes the target which a

researcher wishes to describe or investigate. For

example, our model is focused on teams as the target

level of analysis. More specifically, we investigate the

four constructs of transparency, behavioral integrity,

trust, and performance at the team level. Thus, each

of these four constructs is a property of the teams

themselves, not the individuals that compromise the

team, nor the organizations within which the teams

are embedded. In other words, to say that Team A

has high behavioral integrity is to say that Team A as

a whole unit in and of itself has high behavioral

integrity. Certainly, it seems plausible that Team A

might have high behavioral integrity because it is

composed of high behavioral integrity individuals or

because it is embedded in an organization whose

culture reflects high behavioral integrity, but these

types of issues which involve multiple levels of

analysis are not considered in the current model. In

addition to consideration of teams as whole units,

our model also includes an assumption of isomor-

phism, meaning that each construct has the same

Michael E. Palanski et al.

structure and function (i.e., relationship to other

constructs) at different levels of analysis. For instance,

empirical evidence shows that behavioral integrity is

related to trust at the individual level (Simons et al.,

2007); the assumption of isomorphism leads to the

assertion (which we test) that behavioral integrity is

related to trust at the team level as well.

Second, the level of measurement describes the

actual source of the data which are to be analyzed.

For example, in both studies presented here, data

were collected from individual team members (with

the exception of team performance). Following the

method used by Hofmann and Jones (2005) for

adapting the individual-level Big Five personality

types to the team level, we used what Chan (1998)

has called a referent-shift compositional model. In

other words, we adapted existing individual-level

measures to allow individuals to rate the team as a

whole unit. In contrast, team performance data were

collected from third-party sources which provided a

single, team-level rating of performance.

Finally, the level of statistical analysis describes the

treatment of the data in statistical analysis. In both

studies presented here, individual-level data were

aggregated to the team level (again, with the

exception of team performance, which was collected

and analyzed at the team level). Mean individual

scores were used for aggregation after justifying

within-group agreement via within-group (rwg) and

intraclass correlation coefficient (ICC(1))statistics.

Transparency

Transparency has been discussed under different

labels in several areas of the literature. While the

core idea in all these conceptualizations is the same

[i.e., that transparency is a virtue (Murphy et al.,

2007) which involves openness, availability, or dis-

closure of information], they differ in terms of what

is being disclosed and the level of analysis at which

they are defined. For instance, in the accounting

area, the notion of accounting transparency is con-

cerned with public dissemination of information

about business transactions and exists at the national

level (Bushman and Smith, 2003). Political trans-

parency is concerned with public disclosure of

information the government possesses or collects as

well as public disclosure of information about gov-

ernment actors, the decisions they make, and the

reasons for those decisions (Balkin, 1999). It can exist

at the national, state, and local government levels

(Balkin, 1999). In the computer-supported coopera-

tive work literature, a subarea within information

systems, researchers focus on contextual awareness that

results at the team level when team members disclose

information about their current work context (e.g.,

how busy they are, and what they are currently doing)

so that others in the team can decide whether or not to

disturb them (Dabbish and Kraut, 2008). Within the

organizational behavior area, transparency may be

conceptualized at the organizational level as informa-

tional justice, which entails providing explanations

about organizational procedures and being thorough,

candid, timely, and considerate toward others’ specific

needs in communications about those procedures

(Colquitt, 2001).

We are concerned with transparency at the team

level, and our focus is on the sharing of enough

information and explanations that enable group/

team members to carry out their components of the

group’s/team’s task. Accordingly, we define team

transparency as the sharing of relevant information

and explanations within a team to enable its mem-

bers to carry out their responsibilities within the

team. Similar to Eggert and Helm’s (2003) descrip-

tion of relationship transparency, we focus on the

degree to which there is an atmosphere in which

team members inform one another with information

and explanations about decisions which are made.

Behavioral integrity

Palanski and Yammarino (2007) have shown that

there is a great deal of misunderstanding and differ-

ence surrounding the meaning of word integrity.

They showed that integrity has been used in man-

agement, applied psychology, and business ethics

literature to mean many things, including wholeness,

authenticity, consistency in adversity, consistency

between words and actions, and moral/ethical

behavior. They proposed that integrity should be

considered as a virtue within the framework of

moral philosophy as a way to resolve this misun-

derstanding and difference of opinion. Based on this

framework, they suggested that integrity should be

defined as ‘‘the consistency of an acting entity’s

Team Virtues

words and actions’’ (Palanski and Yammarino, 2007,

p. 17). In addition, Palanski and Yammarino (2009)

have developed a multi-level theory of integrity

which, among other things, explicitly considers the

integrity of teams. Palanski and Yammarino’s (2007,

2009) definition of integrity is very similar to Simons

(2002, p. 19) definition of behavioral integrity, the

‘‘perceived pattern of alignment between an actor’s

words and deeds.’’ Because behavioral integrity is

more precise terminology and has a nascent stream of

research at the individual level (e.g., Dineen et al.,

2006; Simons et al., 2007), we retain the behavioral

integrity terminology, but draw on Palanski and

Yammarino’s (2009) model as the basis for our the-

oretical discussion and hypotheses.

Behavioral integrity at the individual level has

been linked theoretically to organizational citizenship

behaviors and willingness to accept change (Simons

et al., 2007) and has been shown both to directly affect

and to moderate the effect of supervisory guidance on

employee conduct (Dineen et al., 2006). Behavioral

integrity of individual leaders has also been linked to

increased follower trust, both theoretically (Simons,

2002) and empirically (Simons et al., 2007).

Because most theory about behavioral integrity

focuses on the individual level of analysis, Palanski

and Yammarino (2009) distinguished between the

integrity of individual team members and the

integrity of the overall group (team). They described

the integrity of the group as the integrity of an acting

entity; in other words, although group-level integ-

rity may emerge from the individual integrity of the

team members, group-level integrity refers to the

integrity of the team as a separate, autonomous

entity which is irreducible to the individual level

of analysis. Simons’ (2002) concept of behavioral

integrity includes two theoretical sub-dimensions:

consistency between espoused and enacted values

and consistency between promises made and

promises kept. Palanski and Yammarino (2009) re-

tained these two sub-dimensions, noting that team

values may be espoused in several forms, such as

through team mission statements or mottos, in team

lingo, or in team goals. Likewise, a team as a whole

may also make and keep promises (e.g., the infor-

mation technology team at a company promises to

resolve issues within a specified period of time), thus

demonstrating that team-level behavioral integrity is

plausible.

We surmise that both aspects of transparency,

amount of information shared and amount of expla-

nation for decisions made, will enhance behavioral

integrity. Simons (2002) describes behavioral integ-

rity as a perceived pattern of alignment between words

and deeds. In order for individuals to formulate per-

ceptions, the actual words and deeds of others must be

made salient. Similarly, the words and deeds of a team

must be made salient for ascriptions of team behav-

ioral integrity. We believe that higher levels of

information sharing will lead to more salient aware-

ness of behavioral integrity. While increased salience

may strengthen the relationship between transpar-

ency and behavioral integrity, the direction of this

relationship (positive or negative) is still open to de-

bate. For instance, a team might be high in transpar-

ency and yet have very low behavioral integrity. In

such a case, the team’s high transparency would likely

lead to an even lower assessment of its behavioral

integrity as the team’s word/deed misalignment

comes to light. However, based on evidence that

virtues tend to enjoy a positive relationship with one

another (Palanski and Yammarino, 2007), we suspect

that such instances will be rare and that the transpar-

ency-behavioral integrity relationship will usually be

positive. Moreover, the ‘‘explanation of decisions

made’’ aspect of transparency may also support a

positive relationship, as others may overlook a pos-

sible word/deed misalignment provided there is an

adequate explanation. For example, a team may

espouse the value of ‘‘never be late for a meeting with

another team,’’ only to show up late to a meeting.

Technically, this action would indicate a misalign-

ment between word and action; however, if the

offending team offers a reasonable explanation (e.g.,

they stopped to help at the scene of a traffic accident),

the perception of behavioral integrity may not be

affected. Thus, we propose

Hypothesis 1: Team transparency will be positively

related to team behavioral integrity.

Trust

Research on trust, particularly follower trust in a

leader, has expanded greatly over the past 15 years.

Research has shown that trust is an important

antecedent for a number of key outcomes, including

Michael E. Palanski et al.

job performance, organizational citizenship behav-

iors, organizational commitment, job satisfaction,

and turnover intentions (Dirks and Ferrin, 2002).

Different researchers have proposed different con-

ceptualizations of trust, which have resulted in

confusion and disagreement about the meaning and

impact of trust. Following Dirks and Ferrin (2002),

we define trust as ‘‘a psychological state comprising

the intention to accept vulnerability based upon

positive expectations of the intentions or behavior of

another’’ (p. 395).

Palanski and Yammarino (2009) point out that a

number of studies have examined team-level trust

(e.g., Peters and Karren, 2009), but most of these

studies do not consider trust in the team as a whole

(Dirks and Ferrin, 2002). However, Currall and

Inkpen (2002) examine the issues surrounding trust

at various levels of analysis, proposing that an indi-

vidual, a team, or a firm (organization) may be either

a trustor or a trustee. Thus, trust may exist, for in-

stance, between two different firms, between an

individual (e.g., employee) and a firm, or between

two teams within an organization (Palanski and

Yammarino, 2009).

According to Palanski and Yammarino’s (2009)

model, the notion that team behavioral integrity will

lead to trust in the team is based on an assumption of

isomorphism (Kozlowski and Klein, 2000) in which

the higher-level constructs (i.e., team behavioral

integrity and team trust) have the same structures

and functions as the lower-level constructs (i.e.,

individual behavioral integrity and individual trust).

Given that team behavioral integrity has the same

structure (consistency of words and actions) as

behavioral integrity at the individual level, and that

team trust has the same structure (willingness to be

vulnerable) as trust at the individual level, we

propose that the function of the two constructs is

the same at both the individual and team levels.

Specifically, we surmise

Hypothesis 2: Team behavioral integrity will be

positively related to team trust.

Given the Aristotelian tradition upon which vir-

tue ethics is based, one might legitimately ask the

question, ‘‘What is/are the purpose(s) of teams?’’ Just

as the virtues of a knife (e.g., sharpness and strength)

enable it to fulfill its purpose (i.e., to cut), so too

should team-level virtues (e.g., transparency,

behavioral integrity, and trust) enable a team to fulfill

its purpose(s). While teams may exist to fulfill mul-

tiple purposes, at the very least in a business context

a team exists to accomplish specific tasks – i.e., to

perform. Thus, team performance is a logical out-

come to consider in conjunction with team virtues.

De Jong and Elfring (2010) demonstrated that

intrateam trust (perceptions of trust that a team

member has in his or her fellow team members) is

positively related to team performance. Similarly,

Dirks and Ferrin (2002), describing follower trust in

leaders, articulated two mechanisms by which trust

may positively impact performance. First, as trust

develops, people will spend less time ‘‘covering their

backs’’ and more time focusing on their jobs. Sec-

ond, drawing upon social exchange theory, they

suggested that individuals who trust one another will

develop higher quality social relationships. In these

relationships, individuals will help one another

and go above and beyond the call of duty, actions

which should lead to higher levels of performance.

Although Dirks and Ferrin (2002) were describing

leader–follower relationships, we believe that the

same reasoning applies to teams. Members who

belong to a team in which there is a high level of

trust (in other words, where the team itself is the

trustee) can devote more time to doing their jobs

and less time to attempting to manipulate team

dynamics to protect themselves. Similarly, teams

which have high levels of trust are also likely to have

high-quality social relationships in which team

members help one another to fulfill job require-

ments. Thus, given the individual-level evidence

that trust is related to performance and the plausible

assumption that performance is isomorphic across

levels of analysis, we propose

Hypothesis 3: Team trust will be positively related to

team performance.

Method and results

We tested the hypotheses using data from two

related empirical studies: a lab study with temporary

work teams (Study 1) and a field study with ongoing

work teams (Study 2). We chose a two-study

Team Virtues

approach in an attempt to balance the relative

strengths and weaknesses of each study design.

Specifically, the lab design of Study 1 was designed

to test the model with a sample of temporary, ad hoc

teams, to utilize a somewhat larger sample size

(n = 35 teams vs. n = 16 teams, respectively), and to

allow for more rigorous testing by including data

collected at different times. The field setting of Study

2 was designed to support inferences about gener-

alizability of the results by examining permanent,

ongoing teams who engaged in a variety of tasks and

to replicate the construct, internal, and external

validities from Study 1. We present the methods and

results for each study independently, followed by a

discussion which integrates both studies.

Study 1: laboratory study

Participants and procedures

Participants were 149 students (mostly juniors) in an

introductory organizational behavior class at a

medium-size public university in the Northeast

United States who were randomly assigned to teams

of 3–5 members each. Thirty-six students were as-

signed as team leaders by the researchers based on

the results of previous personality test results. Spe-

cifically, students who scored highest on the team in

previously administered extraversion and conscien-

tiousness testing were assigned to be team leaders.

The remaining students (N = 113) were assigned as

team members. Missing data reduced useable num-

ber of participants to 148 participants (35 leaders and

113 followers) working in 35 teams of 3–5 members

each. Participants had a mean age of 21 years, were

split evenly between men and women, and had a

mean of 4.8 years of at least part-time work

experience.

In this study, participants were placed into teams

with the purpose of working on one major class

project. In an effort to simulate ‘‘real-world’’ expe-

riences and enhance generalizability of the results, the

team leader was responsible for scheduling and con-

ducting team meetings and communicating with the

class instructor and teaching assistants. Team leaders

were given limited reward and coercive power (e.g.,

the option to reallocate extra credit based on contri-

bution and performance). In addition, team members

were given individual responsibilities within the team

(e.g., communicating with professors and teaching

assistants). Data in the form of surveys were collected

online through survey software.

Measures

Transparency. We used a five-item scale adapted from

Kernan and Hanges (2002) designed to measure the

amount and thoroughness of information and

explanations that are shared within a team. All five

items were measured on a five-point scale ranging

from 0 (‘‘strongly disagree’’) to 4 (‘‘strongly agree’’).

An example item was, ‘‘People on this team provide

each other with enough details for us to do our

jobs.’’ Data used in theoretical testing were self-

reports of team members collected at Time 1 (about

halfway through the 8-week project).

Behavioral integrity. We developed a two-item scale

based on Simons’ Behavioral Integrity (BI) scale

(Simons et al., 2007). Simons’ BI scale is focused on

the BI of individual leaders and is designed to

measure both promise-keeping (e.g., ‘‘When this

person promises something, I can be certain that it

will happen’’) and consistency between espoused

and actual values (e.g., ‘‘This person conducts

himself/herself by the same values that he/she talks

about’’). Our scale retains these two theoretical sub-

dimensions with two questions (i.e.., ‘‘How often

does this team keep promises?’’ and ‘‘How often

does this team act in a way that shows that these

values are actually important?’’, respectively) with

behaviorally anchored responses ranging from

0 = ‘‘not at all’’ to 4 = ‘‘frequently, if not always.’’

Data used in theoretical testing were self-reports of

team members collected at Time 1.

Trust. Mayer et al. (1995) have noted that various

definitions of trust tend to confuse elements of trust

and elements of its antecedents and consequences,

and suggest using a narrower conceptualization of

trust based on the willingness to be vulnerable to

another. Mayer and Gavin (2005) have developed a

trust scale that was modified for use in this study. All

three items are measured on a five-point scale

ranging from 0 (‘‘strongly disagree’’) to 4 (‘‘strongly

agree’’). Trust items were modified slightly to reflect

a shift from manager to the team. For example, ‘‘If

someone questioned my motives, my manager

would give me the benefit of the doubt.’’ was

Michael E. Palanski et al.

changed to ‘‘If someone questioned my team’s

motives, I would give my team the benefit of the

doubt.’’ Data used in theoretical testing were self-

reports of team members collected at Time 2 (at the

conclusion of the 8-week project).

Team performance. Independent third-party (i.e.,

teaching assistant) reports of team performance were

obtained using Mott’s (1972) scale, which is

designed to measure the quantity, quality, and effi-

ciency of job performance. Answers were assessed on

a scale ranging from 0 (‘‘their quality is poor’’) to 4

(‘‘their quality is excellent’’). To minimize response

error, each team’s performance was rated by two

teaching assistants who were asked to reach a con-

sensus and provide a single rating for each team at

Time 2.

Results

Table I contains means, standard deviations, corre-

lations, and square roots of average variance ex-

tracted (AVE) (see PLS explanation below) for all the

variables. Table II contains factor loadings and cross-

loadings for all the variables.

To test the assertion that BI, trust, and transpar-

ency are all team-level constructs, we computed the

rwg within-group agreement statistic and intraclass

correlation coefficient, or ICC(1), for team behav-

ioral integrity, team trust, and team transparency.

The rwg statistic (James et al., 1993) is calculated by

comparing the observed group variance to an ex-

pected random variance and is useful for analyzing

the variance within a single group. Typically, the

expected random variance is assumed to be uniform

which, as Bliese (2000) points out, is not without

theoretical problems. Still, the comparison to a

uniform variance has become somewhat of a stan-

dard and is associated with a cutoff value of 0.70. All

the variables in the study had an average rwg of 0.84

across teams. The ICC(1) is the proportion of total

variance that can be explained by team membership.

TABLE I

Means, standard deviations, reliabilities, and correlations for Study 1

Variables Mean SD 1 2 3 4

1. Team transparency (Time 1) 3.19 0.34 0.83

2. Team integrity (Time 1) 3.32 0.40 0.78** 0.88

3. Team trust (Time 2) 2.83 0.40 0.35 0.43** 0.82

4. Team performance 3.35 0.76 0.20 0.40** 0.24* 0.84

Note: The boldfaced values on the diagonal represent the square root of the AVE.

n = 35, *p > 0.01, **p > 0.01.

TABLE II

Factor and cross-factor loadings, AVE, and internal con-

sistency reliability (ICR) of items in Study 1

Item Factor

1 2 3 4

Team behavioral integrity (AVE = 0.78; ICR = 0.88)

1 0.92 0.47 0.71 0.34

2 0.84 0.34 0.66 0.04

Team trust (AVE = 0.67; ICR = 0.86)

1 0.39 0.76 0.32 0.22

2 0.32 0.82 0.19 0.24

3 0.37 0.87 0.33 0.46

Team transparency (AVE = 0.69; ICR = 0.91)

1 0.70 0.25 0.89 0.10

2 0.73 0.38 0.82 0.08

3 0.64 0.32 0.85 0.27

4 0.68 0.28 0.89 0.20

5 0.43 0.22 0.70 0.25

Team performance (AVE = 0.70; ICR = 0.88)

1 0.00 0.18 0.08 0.76

2 0.33 0.46 0.22 0.93

3 0.13 0.23 0.15 0.82

Note: Factor loadings are indicated in boldface.

AVE = average variance extracted.

ICR = internal consistency reliability.

Team Virtues

Although no clear standard cutoff for the ICC(1)

exists, Bliese (2000) reports that most field studies

result in an ICC(1) ranging from 0.05 to 0.30. The

ICC(1)’s for the variables in this study were in this

range, as team transparency was 0.28, team behav-

ioral integrity was 0.23, and team trust was 0.19.

For testing of both the measurement and the the-

oretical models, Partial Least Squares (PLS) using

SmartPLS (Ringle et al., 2005) was employed as the

primary data analysis technique. PLS is widely used

for exploratory data testing and has several advantages

over other techniques (Chin and Newsted, 1999).

PLS does not require multivariate normal distribution

and is especially suitable for the analysis of small

samples. Moreover, PLS can help reduce measure-

ment error by weighing the individual indicators of a

multi-indicator variable (Sosik et al., 2009). Other

forms of path modeling, such as covariance-based

structural equation modeling, are generally used in

confirmatory model testing and may be susceptible to

error in situations where there is a low construct-to-

sample size ratio, as was generally the case in this

study. PLS also has the ability to test both the mea-

surement model and theoretical model simulta-

neously. This ability makes PLS preferable to multiple

regression analysis in which the measurement model

and theoretical models must be tested independently.

The test of the measurement model includes three

primary parts: (1) individual item reliability, (2)

internal consistency, and (3) discriminant validity.

Tables I and II include results for all the three parts.

Individual item reliability was assessed by examining

the factor loadings of each measure on its corre-

sponding construct. Fornell and Larcker (1981)

suggest accepting items which have more explana-

tory power than error variance. In practice, the

generally accepted cutoff is 0.70 or greater. All factor

loadings in Study 1 were equal to or greater than

0.70; thus, individual item reliability was generally

quite robust for the constructs in these studies.

Construct internal consistency may be assessed by

composite internal scale reliability, which is similar

to Cronbach’s a. Fornell and Larcker (1981) suggest a cutoff of 0.70 for internal consistency. A second

way to measure internal consistency is with AVE,

which is a measure of variance accounted for by the

underlying construct. Fornell and Larcker (1981)

suggest a cutoff of 0.50 for AVE. All constructs met

both criteria for internal consistency.

Discriminant validity in PLS is assessed in two

ways. First, each item should load higher on the

construct that it is supposed to measure than on any

other construct (Carmines and Zeller, 1979). All

items in the study met this criterion. Second, each

construct should share more variance with its items

than with any other construct in the model (Barclay

et al., 1995). This criterion is usually assessed similar

to a multi-trait/multi-method approach. Specifi-

cally, the square root of the AVE of a construct

should be greater than the construct’s correlation

with any other construct in the model. For Study 1,

an examination of Table I (in which the square root

of the AVE is located on the diagonal) demonstrates

that this criterion was also met. Based on both of

these criteria, the variables in this study showed

strong discriminant validity.

Results of the test of the theoretical model are

shown in Figures 1 and 2. The standardized beta

coefficient for each path in the model was obtained

from the PLS algorithm in SmartPLS. Statistical

significance of each path in the theoretical model

was determined by the t-value for a given bivariate

relationship based on a bootstrapping technique with

500 iterations. Results showed that team transpar-

ency was positively related to team behavioral

integrity (b = 0.78, p < 0.01); thus, Hypothesis 1

was supported. The relationship between team

behavioral integrity and team trust was also positive

(b = 0.43, p < 0.05); thus, Hypothesis 2 was

supported. Similarly, in support of Hypothesis 3,

there was a significant positive relationship between

team trust and team performance (b = 0.40,

p < 0.05).

Study 2: field study

Participants and procedures

Data were collected from clinical nurses and their

managers in a regional healthcare organization.

Eighty-three nurses from 18 offices (which were

directed by a total of five managers, each of whom

was responsible for 1–5 offices and approximately 20

nurses) participated in the study with a response rate

of 80%. Because of low response rates in two offices,

the number of usable teams (defined as the staff in a

given office) was 16. Participants were 95.3%

women with a mean age of 45.4 years. Average

Michael E. Palanski et al.

organizational tenure was 6.59 years, average job

tenure was 4.04 years, and average tenure with

manager was 2.02 years. All the participants had at

least three months’ tenure with the team. Potential

non-response bias was assessed by comparing demo-

graphics of usable responses with demographics

of non-usable (anonymous) responses and non-

participants. Results of a MANOVA for age and

gender revealed no significant differences on these

variables. Data were collected on site in paper-and-

pencil format. Participants who were not able to

complete the surveys on-site were provided with

postage-paid envelopes which were addressed di-

rectly to the principal investigator.

Team Behavioral Integrity

S1 r2 = .61

S2 r2 = .70

Team Transparency

Team Trust

S1 r2 = .19

S2 r2 = .45

Team Performance

S1 r2 = .16

S2 r2 = .73

S1 = .78**

S2 = .84**

S1 = .43*

S2 = .67**

S1 = .40*

S2 = .86**

Figure 1. Results from Study 1 and Study 2. Results on path lines are standardized b weights. S1 = Study 1;

S2 = Study 2, *p < 0.05; **p < 0.01.

Team Transparency

Team Behavioral Integrity

Team Trust

Team Performance

x1 x2 x3 x4 x5

x1 x2 x3 x4 x5

Method Factor (Social desirability)

x14

x6 x7

x6 x7

x8 x10

x8 x10 x11 x12 x13

x9

x9

Figure 2. PLS model for assessing common-method bias. Note: Team transparency, team behavioral integrity, team

trust, and team performance are substantive constructs. The method factor is added to the theoretical model to assess

common-method bias. x1–x14 are indicators of constructs. Indicators x1–x10 were obtained using a survey. x14 is a

scale based on items obtained from the same survey that was used to obtain x1–x10. The small circles below substan-

tive constructs are single-indicator constructs that were added to represent survey-based indicators.

Team Virtues

Measures

For consistency across studies, we used the same

scales from Study 1 for team transparency, team

behavioral integrity, team trust, and team perfor-

mance. Data for transparency, BI, and trust were

cross sectional and obtained from team members,

but, in an effort to reduce common-source bias,

team performance was assessed by a director who

was not a part of the teams and responsible for

overseeing all the 18 offices.

Results

Table III contains means, standard deviations, cor-

relations, and square roots of AVE (see PLS expla-

nation below) for all the variables. Table IV contains

factor loadings and cross-loadings for all the vari-

ables.

To test the assertion that transparency, BI, and trust

are all team-level constructs in Study 2, we again

computed the rwg within-group agreement statistic

and intraclass correlation coefficient, or ICC(1), for

team transparency, team behavioral integrity, and

team trust. All the variables in the study had an

acceptable rwg average across teams (team behavioral

integrity = 0.75, team trust = 0.77, and team trans-

parency = 0.75, respectively). The ICC(1)’s for the

variables in this study were also in the range reported

by Bliese (2000), as team behavioral integrity was

0.17, team trust was 0.08, and team transparency was

0.10.

For testing of both the measurement and the

theoretical models, we once again used PLS with

SmartPLS (Ringle et al., 2005). We followed the

same methods for assessing the measurement and the

theoretical models as for Study 1. Tables III and IV

include results for all the three parts. Individual item

reliability exceeded the 0.70 threshold suggested by

Fornell and Larcker (1981). All the constructs also

met both criteria (0.70 for internal scale reliability

and 0.50 for AVE) for internal consistency. Finally,

discriminant validity was demonstrated as each item

TABLE III

Means, standard deviations, reliabilities, and correlations for Study 2

Variables Mean SD 1 2 3 4

1. Team transparency 3.03 0.61 0.96

2. Team integrity 3.32 0.56 0.67* 0.98

3. Team trust 3.10 0.40 0.83* 0.67* 0.87

4. Team performance 3.34 0.65 0.64* 0.49* 0.84* 0.85

Note: The boldfaced values on the diagonal represent the square root of the AVE.

n = 16, *p > 0.01.

TABLE IV

Factor and cross-factor loadings, AVE, and ICR of

items in Study 2

Item Factor

1 2 3 4

Team behavioral integrity (AVE = 0.96; ICR = 0.98)

1 0.98 0.69 0.81 0.54

2 0.98 0.61 0.84 0.42

Team trust (AVE = 0.75; ICR = 0.90)

1 0.51 0.84 0.64 0.80

2 0.76 0.94 0.79 0.71

3 0.43 0.81 0.46 0.61

Team transparency (AVE = 0.93; ICR = 0.99)

1 0.79 0.67 0.94 0.51

2 0.84 0.75 0.97 0.51

3 0.82 0.66 0.96 0.68

4 0.75 0.78 0.97 0.71

5 0.85 0.81 0.97 0.63

Team performance (AVE = 0.70; ICR = 0.88)

1 0.63 0.68 0.64 0.73

2 0.31 0.73 0.45 0.93

3 0.31 0.73 0.45 0.93

Note: Factor loadings are indicated in boldface.

AVE = average variance extracted.

ICR = internal consistency reliability.

Michael E. Palanski et al.

loaded the highest on its designated construct and all

square roots of each construct’s AVE were greater

than each respective construct’s correlation with

other constructs.

Results of the test of the theoretical model are

shown in Figure 2. The standardized beta coefficient

for each path in the model was obtained from the

PLS algorithm in SmartPLS. Statistical significance of

each path in the theoretical model was determined

by the t-value for a given bivariate relationship based

on a bootstrapping technique with 500 iterations.

Results showed that team transparency was posi-

tively related to team behavioral integrity (b = 0.84,

p < 0.01); thus, Hypothesis 1 was supported. The

relationship between team behavioral integrity and

team trust was also positive (b = 0.67, p < 0.01);

thus, Hypothesis 2 was supported. Similarly, in

support of Hypothesis 3, there was a significant

positive relationship between team trust and team

performance (b = 0.86, p < 0.01).

Discussion

Before discussing the results from hypothesis testing,

it is important to note some of the construct validation

results from PLS testing in both studies. First, variables

which were measured at the individual level and

aggregated to the team level (i.e., transparency, BI,

and trust) all had very robust support for aggregation

both from rwg and ICC(1) statistics. Further, these

variables also displayed robust internal consistency

and reliability, as well as acceptable discriminant

validity. We highlight these results because, to our

knowledge, these are the first studies which contain

team-level transparency and BI as focal constructs,

and the data here offer initial empirical support for

their operation at the team level of analysis.

Hypothesis 1 suggested a positive relationship

between team transparency and team behavioral

integrity, as the free exchange of information may be

necessary for the recognition of BI. PLS results

showed strong support for the paths in both studies

(b = 0.78 and b = 0.84, respectively), as well as the

practical significance of transparency as an anteced-

ent to BI, with a high amount of variance explained

in both studies (r 2

= 0.61 in Study 1 and r 2

= 0.70

in Study 2). Although the PLS results showed

support for the discriminant validity of the two

variables, we were concerned that the magnitude of

their relationship might be exaggerated by common-

method bias, as both variables were collected at the

same time in the same way from the same raters in

both studies. As such, we performed post hoc anal-

yses to test for the possible effects of common-

method bias as follows.

Except for team performance, all other criterion

and predictor variables were measured using group

members’ responses to a survey, resulting in a concern

about common-method bias. In a structural equation

model, the presence of common-method bias can be

assessed by including a common-method factor in the

model being tested and linking this factor to indicators

measured by this common method. Williams et al.

(2003) suggest that the common-method factor may

be modeled using a scale for social desirability, which

can be a potential reason for the common-method

bias. The surveys administered in both the laboratory

and field studies included 10 true–false questions that

measured social desirability (Strahan and Carrese

Gerbasi, 1972). Items measuring this variable in-

cluded ‘‘I like to gossip at times.’’ and ‘‘I never resent

being asked to return a favor.’’ A scale for social

desirability was constructed from these items and was

used to indicate the common-method factor that we

included in our model to assess the level of common-

method bias.

Unlike covariance-based structural equation

modeling (which is implemented using packages

such as LISREL and AMOS), PLS does not allow

one to directly link more than one construct with an

indicator. Therefore, to model the effects of a sub-

stantive construct and the common-method factor

on any survey-based indicator, one has to finesse the

structural equation model using a method suggested

by Liang et al. (2007). This method involves the

modeling of survey-based indicators that need to be

tested for common-method bias as new constructs.

Essentially, one adds new constructs to the model:

one for each survey-based indicator. Each of these

new constructs is modeled with a single but different

survey-based indicator and is represented as being

dependent on the substantive construct (that the

indicator is representing) and the common-method

factor. See Figure 2 for a depiction of the resulting

model. Liang et al. (2007) argue that this finessing is

valid because the variance that is shared between a

survey-based indicator and its substantive construct

Team Virtues

remains unaltered and is now represented as

the variance shared between two constructs, i.e., the

new single-indicator construct representing the

indicator and the substantive construct represented

by the indicator. Similarly, the variance shared be-

tween a survey-based indicator and the common-

method factor is unaltered and is now represented as

the variance shared between the new single-indica-

tor construct representing the indicator and the

common-method factor.

According to Williams et al. (2003), common-

method bias may be suggested when (a) the com-

mon-method factor has a significant effect on the

new constructs representing survey-based indicators,

and (b) the new constructs share more variance with

the common-method factor than with their respec-

tive substantive constructs. Results of testing the

new model suggest that common-method bias is

probably not a serious concern for this study. Spe-

cifically, in the laboratory study, none of the 10

paths from the common-method factor to the new

constructs representing survey-based indicators were

significant (the magnitudes of path coefficients ran-

ged from 0.036 to 0.129); in the field study, only

one path out of ten (the path from the common-

method factor to a construct representing an

indicator of transparency) was significant (the mag-

nitudes of path coefficients ranged from 0.017 to

0.213). In addition, in the laboratory study as well as

the field study, none of the new constructs shared

more variance with the common-method factor

than it did with their respective substantive con-

structs. In the laboratory study, the average sub-

stantively explained variance was 0.70, while the

average method-based variance was 0.02. Thus, the

ratio of substantively explained variance to method-

based variance was 35:1. In the field study, the

corresponding variances were 0.08 and 88, and the

ratio was 11:1.

The second path in the model (Hypothesis 2)

suggested a relationship between team behavioral

integrity and team trust. This relationship was

predicated on an assumption of isomorphism based

on prior BI research at the individual level of analysis

(Simons, 2002; Simons et al., 2007). Results from

both studies showed support for both the paths

(b = 0.43 and b = 0.67) and variance explained

(r 2

= 0.19 and r 2

= 0.45). Although the post hoc test

described above did not reveal common-method

bias between BI and trust, the relatively lower path

weights and variance explained in Study 1 may be in

part attributed the fact that these variables were

collected at different times, whereas in Study 2 they

were collected simultaneously. Some of the differ-

ences are also likely attributable to the differences

between the samples. Sample 1 consisted of tem-

porary, ad hoc teams in which trust may not be fully

developed and may not play a major role for the

team. In contrast, Study 2 consisted of more per-

manent and established teams in which consistent

behavior, as demonstrated by BI, is more important

for developing trust in the team over time and across

different situations.

Relative differences in the results for Hypothesis 3

may also be attributed in part to differences between

the two samples. The results in Study 1 showed a

solid path weight (b = 0.40) and variance explained

(r 2

= 0.19), indicating that trust plays an important

role in determining team performance. In contrast,

results from Study 2 indicate a very strong rela-

tionship between team trust and team performance,

both in terms of path weight (b = 0.86) and variance

explained (r 2

= 0.73). In retrospect, this makes a

good deal of sense, for the teams in Study 2 are

providing healthcare services which intrinsically

require a great deal of trust in the team (especially on

the part of patients).

Despite the differences in magnitude of the results

for Hypothesis 3 between the samples, the fact that

the paths in both samples were significant is some-

what striking, given that team performance was rated

by independent, third-party raters in both samples.

Trust in the individual has tended to play a modest

role with respect to individual performance. For

example, in their meta-analysis, Dirks and Ferrin

(2002) found a correlation of 0.16 between trust in

leader and job performance at the individual level.

Based on correlations (r = 0.24 in Study 1 and

r = 0.84 in Study 2) in our two studies, there may be

a more significant association between trust and

performance at the team level.

Limitations and future directions

Although we tried to balance strengths and weak-

nesses through a two-study design, several limita-

tions remain. First, although Study 1 employed data

Michael E. Palanski et al.

collection at a different time for trust, data from both

studies were mostly cross sectional. Even third-party

ratings of team performance were collected at the

same time as other variables. Thus, inferences about

how and when constructs develop and influence one

another with respect to teams should be interpreted

with caution.

Second, although the level of theory and the level

of data analysis are at the team level, data for trans-

parency, BI, and trust were collected at the indi-

vidual level and aggregated. Although aggregation to

the team level was supported, perceptions of team

transparency, BI, and trust were not totally

homogenous; thus, there may be individual differ-

ences in perception which are important. Moreover,

data about transparency, BI, and trust were collected

from team members themselves. One could make an

argument, however, that perceptions of these con-

structs, especially BI and trust, should be measured

by parties outside the group. For example, in Study

2, it would be interesting to obtain patient percep-

tions of BI and trust for the teams of nurses. After all,

in that type of situation, patient trust might be the

most important outcome.

Finally, the model presented here is not the only

conceivable model, especially with respect to BI and

transparency. Research at the individual level of

analysis has posited and subsequently demonstrated

that BI’s impact on performance is mediated by trust

(e.g., Simons, 2002). We have made assumptions of

isomorphism such that team-level constructs in this

model will function in the same manner as con-

structs at the individual level of analysis; however,

other alternatives may be possible. As such, we also

tested an alternative model in which team-level BI

has a direct association with team performance. The

results are described in Figure 3 (Alternative 1).

Likewise, while we have posited that transparency

is necessary to demonstrate BI, transparency may also

be viewed as a risk of sorts. For example, when a

person discloses information about himself/herself,

that information may be potentially used against the

person in a harmful way. If the same person is faced

with the choice of whether to disclose information

on behalf of the team, then he/she may be more

likely to disclose the information whether he/she

trusts the team. As a result, we may be able to make a

case that transparency (when viewed as a risk) should

be an outcome resulting from high trust, rather than

an antecedent to BI. Thus, we also tested an alter-

native model in which team behavioral integrity is

associated with team trust, which in turn is associated

with team transparency and subsequently team per-

formance. The results from this alternative model are

described in Figure 3 (Alternative 2).

In Study 1, the results indicate that the proposed

alternative models’ paths were not significant, thus

lending support to the fact that the hypothesized

model is a better fit to the data. In Study 2, the

proposed alternative models’ paths were significant,

indicating that both of these models may provide

alternative explanations. Although PLS does not

produce model-fit matrices (as might be found in

covariance-based SEM), the variance explained may

Alternative 1

Alternative 2

Team Transparency

S1 r2 = .13

S2 r2 = .60

Team Behavioral Integrity

Team Trust

S1 r2 = .20

S2 r2 = .47

Team Performance

S1 r2 = .05

S2 r2 = .51

S1 = .45*

S2 = .68**

S1 = .36

S2 = .78**

S1 = .21

S2 = .71**

Team Behavioral Integrity

Team Performance

S1 r2 = .11

S2 r2 = .34

S1 = .34

S2 = .58*

Figure 3. Alternative model results from Study 1 and Study 2. Results on path lines are standardized b weights.

S1 = Study 1; S2 = Study 2. *p < 0.05; **p < 0.01.

Team Virtues

be used to compare models in terms of explanatory

power (Chin, 1998). In the hypothesized model, the

team performance r 2

is 0.73. In Alternative 1, the

team performance r 2

is 0.34, while in Alternative 2 it

is 0.51. Thus, while both alternatives may provide

alternative explanations, neither provides as much

explanatory power as the hypothesized model. The

result from Study 2, when coupled with the null

alternative results in Study 1, lend further support to

the hypothesized model as the best explanation.

Overall, the model and results presented here

suggest that the analysis of virtues at the team level of

analysis has the potential for contributing to our

knowledge on team performance and team dynam-

ics. The validity of the model would be strengthened

by considering other perceptions (e.g., customer) of

a team’s transparency, BI, and trust in future studies.

Other research should also consider possible cross-

level effects on team performance and other variables

in the model. For example, in consideration of ‘‘top-

down’’ effects, what effect might transparency at a

higher level (e.g., organization) have on team

functioning? With respect to integrity, Palanski and

Yammarino (2009) discussed the differences be-

tween a compositional entity (the compositional

‘‘makeup’’ of the team based on the level of integrity

of each of its members) and an acting entity (the

integrity of the team in and of itself). It is likely that

there are similar ‘‘bottom-up’’ cross-level aspects to

transparency and trust, and future research should

consider these aspects. Also, just as a team leader’s

integrity likely affects the team’s integrity (Palanski

and Yammarino, 2009), research is needed about

how a team leader’s transparency and trustworthiness

impact the same elements at the team level.

Finally, the model and methods presented here

may serve as a guide to future research which ex-

pands virtues from the individual level to the team

level. For example, at the individual level, trans-

parency is closely related to informational justice.

Thus, our approach may be helpful for expanding

other forms of justice (e.g., interpersonal) to higher

levels of analysis. Similarly, it would be interesting to

explore the role that other virtues (e.g., honesty,

courage, and compassion) play in driving team

outcomes. Similarly, future research should examine

outcomes other than simply performance; for

example, overall team ethical conduct would be

worthy of future consideration. For now, though,

the initial support for team transparency, BI, and

trust presented here is a promising first step.

References

Audi, R. and P. E. Murphy: 2006, ‘The Many Faces of

Integrity’, Business Ethics Quarterly 16, 3–21.

Baker, D. P., R. Day and E. Salas: 2006, ‘Teamwork as

an Essential Component of High-Reliability Organi-

zation’, Health Services Research 41, 1576–1598.

Balkin, J. M.: 1999, ‘How Mass Media Simulate Political

Transparency’, Cultural Values 3, 393–413.

Barclay, D., C. Higgins and R. Thompson: 1995, ‘The

Partial Least Squares (PLS) Approach to Causal Mod-

eling, Personal Computer Adoption and Use as an

Illustration’, Technology Studies 2, 285–309.

Bliese, P. D.: 2000, ‘Within-Group Agreement, Non-

Independence, and Reliability: Implications for

Data Aggregation and Analysis’, in K. J. Klein.

S. W. J. Kozlowski (eds.), Multilevel Theory, Research

and Methods in Organizations: Foundations, Extensions,

and New Directions (Jossey-Bass, San Francisco, CA),

pp. 349–381.

Bushman, R. M. and A. J. Smith: 2003, Transparency,

Financial Accounting Information, and Corporate

Governance. SSRN eLibrary. Retrieved from http://

papers.ssrn.com/sol3/papers.cfm?abstract_id=795547.

Cameron, K., D. Bright and A. Caza: 2004, ‘Exploring

the Relationships Between Organizational Virtuous-

ness and Performance’, American Behavioral Scientist 47,

766–790.

Carmines, E. G. and R. A. Zeller: 1979, Reliability and

Validity Assessment (Sage, Beverly Hills, CA).

Caza, A., B. A. Barker and K. Cameron: 2004, ‘Ethics

and Ethos: The Buffering and Amplifying Effects of

Ethical Behavior and Virtuousness’, Journal of Business

Ethics 52, 169–178.

Chan, D.: 1998, ‘Functional Relations Among Con-

structs in the Same Content Domain at Different

Levels of Analysis: A Typology of Composition

Models’, Journal of Applied Psychology 83, 234–246.

Chin, W. W.: 1998, ‘Issues and Opinion on Structural

Equation Modeling’, MIS Quarterly 22, vii–xvi.

Chin, W. W. and P. R. Newsted: 1999, ‘Structural

Equation Modeling Analysis with Small Samples Using

Partial Least Squares’, in R. Hoyle (ed.), Statistical

Strategies for Small Sample Research (Sage, Thousand

Oaks, CA), pp. 307–341.

Chun, R.: 2005, ‘Ethical Character and Virtue of

Organizations: An Empirical Assessment and Strategic

Implications’, Journal of Business Ethics 57, 269–284.

Michael E. Palanski et al.

Colquitt, J. A.: 2001, ‘On the Dimensionality of

Organizational Justice: A Construct Validation of a

Measure’, Journal of Applied Psychology 86, 386–400.

Currall, S. C. and A. C. Inkpen: 2002, ‘A Multilevel

Approach to Trust in Joint Ventures’, Journal of Inter-

national Business Studies 33, 479–495.

Dabbish, L. and R. Kraut: 2008, ‘Awareness Displays and

Social Motivation for Coordinating Communication’,

Information Systems Research 19, 221–238.

Dansereau, F., J. A. Alutto and F. J. Yammarino: 1984,

Theory Testing in Organizational Behavior: The Varient

Approach (Prentice-Hall, Englewood Cliffs, NJ).

De Jong, B. A. and T. Elfring: 2010, ‘How Does Trust

Affect the Performance of Ongoing Work Teams?

The Mediating Role of Reflexivity, Monitoring and

Effort’, Academy of Management Journal 53(3), 535–549.

Dineen, B. R., R. J. Lewicki and E. C. Tomlinson: 2006,

‘Supervisory guidance and behavioral integrity: Rela-

tionships with employee citizenships and deviant

behavior’, Journal of Applied Psychology 91, 622–635.

Dirks, K. T. and D. L. Ferrin: 2002, ‘Trust in Leadership:

Meta-Analytic Findings and Implications for Research

and Practice’, Journal of Applied Psychology 87, 611–628.

Eggert, A. and S. Helm: 2003, ‘Exploring the Impact of

Relationship Transparency on Business Relationships: A

Cross-Sectional Study Among Purchasing Managers in

Germany’, Industrial Marketing Management 32, 101–108.

Fornell, C. and D. F. Larcker: 1981, ‘Evaluating Struc-

tural Equation Models with Unobserved Variables and

Measurement Error’, Journal of Marketing Research 18,

39–50.

Gully, S. M., K. A. Incalcaterra, A. Joshi and J. M.

Beaubien: 2002, ‘A Meta-Analysis of Team-Efficacy,

Potency, and Performance: Interdependence and Level

of Analysis as Moderators of Observed Relationships’,

Journal of Applied Psychology 87, 819–832.

Guzzo, R. A. and M. W. Dickson: 1996, ‘Teams in

Organizations: Recent Research on Performance and

Effectiveness’, Annual Review of Psychology 47, 307–338.

Hofmann, D. A. and L. M. Jones: 2005, ‘Leadership,

Collective Personality, and Performance’, Journal of

Applied Psychology 90, 509–522.

Howell, J. M. and C. M. Shea: 2006, ‘Effects of Champion

Behavior, Team Potency, and External Communica-

tion Activities on Predicting Team Performance’,

Group & Organization Management 31, 180–211.

James, L. R., R. J. Demaree and G. Wolf: 1993, ‘rwg: An

Assessment of Within-Group Interrater Agreement’,

Journal of Applied Psychology 78, 306–309.

Kernan, M. C. and P. J. Hanges: 2002, ‘Survivor Reactions

to Reorganization: Antecedents and Consequences of

Procedural, Interpersonal, and Informational Justice’,

Journal of Applied Psychology 87, 916–928.

Klein, K. J., F. Dansereau and R. J. Hall: 1994, ‘Levels

Issues in Theory Development, Data Collection, and

Analysis’, Academy of Management Journal 19, 195–229.

Kozlowski, S. W. J. and K. J. Klein: 2000, ‘A Multilevel

Approach to Theory and Research in Organizations:

Contextual, Temporal, and Emergent Processes’, in

K. J. Klein and S. W. J. Kozlowski (eds.), Multilevel

Theory, Research, and Methods in Organizations: Foun-

dations, Extensions, and New Directions (Jossey-Bass, San

Francisco), pp. 3–90.

Liang, H., N. Saraf, Q. Hu and Y. Xue: 2007, ‘Assimi-

lation of Enterprise Systems: The Effect of Institutional

Pressures and the Mediating Role of Top Manage-

ment’, MIS Quarterly 31, 59–87.

MacIntyre, A.: 1984, After Virtue, 2nd Edition (University

of Notre Dame Press, Notre Dame, IN).

Mayer, R. C., J. H. Davis and F. D. Schoorman: 1995,

‘An Integrative Model of Organizational Trust’,

Academy of Management Review 20, 709–734.

Mayer, R. C. and M. B. Gavin: 2005, ‘Trust in Man-

agement and Performance: Who Minds the Shop

While Employees Watch the Boss?’, Academy of Man-

agement Journal 48, 874–888.

Mott, P. E.: 1972, The Characteristics of Effective Organi-

zations (Harper & Row, New York).

Murphy, P. E., G. R. Laczniak and G. Wood: 2007, ‘An

Ethical Basis for Relationship Marketing: A Virtue

Ethics Perspective’, European Journal of Marketing 41,

37–57.

Palanski, M. E. and F. J. Yammarino: 2007, ‘Integrity and

Leadership: Clearing the Conceptual Confusion’,

European Management Journal 25, 171–184.

Palanski, M. E. and F. J. Yammarino: 2009, ‘Integrity and

Leadership: A Multi-Level Conceptual Framework’,

Leadership Quarterly 20, 405–420.

Peters, L. and R. J. Karren: 2009, ‘An Examination of the

Roles of Trust and Functional Diversity on Virtual

Team Performance Ratings’, Group & Organization

Management 34, 479–504.

Ringle, C. M., S. Wende and A. Will: 2005, Smart PLS

2.0. University of Hamburg, Hamburg, Germany.

http://www.smartpls.de.

Schminke, M., D. Wells, J. Peyrefitte and T. C. Sebora:

2002, ‘Leadership and Ethics in Work Groups: A

Longitudinal Assessment’, Group & Organization Man-

agement 27, 272–293.

Simons, T. L.: 2002, ‘Behavioral Integrity: The Perceived

Alignment Between Managers’ Words and Deeds as a

Research Focus’, Organization Science 13, 18–35.

Simons, T. L., R. Friedman, L. A. Liu and J. McLean-

Parks: 2007, ‘Racial Differences in Sensitivity to

Behavioral Integrity: Attitudinal Consequences,

In-Group Effects, and ‘‘Trickle Down’’ Among Black

Team Virtues

and Non-Black Employees’, Journal of Applied Psy-

chology 92, 650–665.

Solomon, R. C.: 2003, ‘Victim of Circumstances? A

Defense of Virtue Ethics in Business’, Business Ethics

Quarterly 13, 43–62.

Sosik, J., S. Kahai and M. Piovoso: 2009, ‘Silver Bullet or

Voodoo Statistics? A Primer for Using the Partial Least

Squares Data Analytic Technique in Group and

Organization Research’, Group & Organization Man-

agement 34, 5–36.

Strahan, R. and K. Carrese Gerbasi: 1972, ‘Short,

Homogeneous Versions of the Marlowe–Crowne

Social Desirability Scale’, Journal of Clinical Psychology

28, 191–193.

Williams, L. J., J. R. Edwards and R. J. Vandenberg:

2003, ‘Recent Advances in Causal Modeling Methods

for Organizational and Management Research’, Journal

of Management 29, 903–926.

Yammarino, F. J., S. D. Dionne, J. U. Chun and

F. Dansereau: 2005, ‘Leadership and Levels of Analysis:

A State-of-the-Science Review’, Leadership Quarterly

16, 879–919.

Michael E. Palanski

Rochester Institute of Technology,

Rochester, U.S.A.

E-mail: [email protected]

Surinder S. Kahai and Francis J. Yammarino

State University of New York at Binghamton,

Binghamton, U.S.A.

Michael E. Palanski et al.