Addressing Issues and Challenges in Counseling

blesd77

Reading4.pdf

Home >Psychology homework help >Addressing Issues and Challenges in Counseling

Promoting Adolescent Sense of Coherence: Testing Models of Risk, Protection, and Resiliency

WILLIAM PAUL EVANS*, SHAWN C. MARSH and DANIEL J. WEIGEL

University of Nevada, Reno, Mailstop No. 140, Reno, NV 89557, USA

ABSTRACT

Sense of coherence (SOC) is a dispositional trait that has been linked to well-being in a broad range of

populations and contexts. Little is known, however, about the factors associated with SOC

development and maintenance across the lifespan. Conceptualized as a condition of resiliency,

SOC in adolescents was explored via measures of risk and protection as reported on by 8th and 10th

grade students (N ¼ 1619). Employing cumulative indexes of protection and risk, analyses focused on testing four models of resiliency. Further, the relative and cumulative effects of protection and risk

were explored across ecological domains. Analyses revealed support for the compensatory model of

resiliency for males and females, while the challenge model also was supported for females. Analyses

also revealed that protection and risk influencing SOC emerge at multiple domains for males and

females. In addition, protective factors present at multiple domains were related to higher SOC while

risk factors present at multiple domains were related to lower SOC regardless of gender. Results

suggest a resiliency framework that considers multiple ecological domains is useful for under-

standing SOC in adolescents. Implications for additional research are presented. Copyright # 2008

John Wiley & Sons, Ltd.

Key words: adolescence; sense of coherence; resiliency

INTRODUCTION

As an alternative to the pathogenic model of health, the salutogenic model proposes

humans experience physical and mental well-being on a continuum, with illness at one end

and wellness at the other (Antonovsky, 1987). Sense of coherence (SOC) is a central

construct in the salutogenesis model (Antonovsky, 1987). Conceptualized as one’s

‘dynamic feeling of confidence that the world is comprehensible, manageable and

meaningful’ (McCubbin, 1998, p. xiii), SOC is believed to be a dispositional trait that

influences how well one handles unavoidable stress—which in turn determines how one

experiences well-being. The construct can be thought of as a broad measure of resiliency,

Journal of Community & Applied Social Psychology

J. Community Appl. Soc. Psychol., 20: 30–43 (2009)

Published online 19 December 2008 in Wiley InterScience

(www.interscience.wiley.com) DOI: 10.1002/casp.1002

* Correspondence to: William Paul Evans, Ph.D., Human Development and Family Studies, University of Nevada, Reno, Mailstop No. 140, Reno, NV 89557, USA. E-mail: [email protected]

with higher levels of SOC linked to positive health-related outcomes in various studies

among diverse samples (e.g. Fiorentino, 1986; Lutgendorf, Vitaliano, Tripp-Reimer,

Harvey, & Lubaroff, 1999).

Although some evidence suggests that SOC reaches a point of stability in adults around

the age of 30 (Kivamaki, Feldt, Vahtera, & Nurmi, 2000), little is known about the factors

associated with developing and maintaining high SOC across the lifespan. This has led

researchers to study youth populations to better understand the origins, context and

interrelationships of SOC (Margalit & Eysenck, 1990; Sagy & Antonovsky, 2000). To help

address this need for youth-focused research, the present study sought to expand our

understanding of factors associated with SOC in adolescent populations.

Given that the salutogenesis model is consistent with a resiliency perspective in that both

emphasize the ability of individuals to experience positive outcomes despite exposure to

risk; we adopted a resiliency framework to guide the study. Resiliency has been defined as

‘a pattern of positive adaptation in the context of past or present adversity (Wright &

Matsen, 2005, p. 18)’. Consensus on the conceptual definitions of risk and protective

factors, however, has only recently emerged in the literature. For the present study, we

agree with the definitions of Wright and Matsen (2005, p. 18) who define a risk factor as ‘a

measurable characteristic in a group of individuals or their situation that predicts negative

outcome on a specific outcome criteria’, for example, poverty, premature birth, or parental

divorce; and a protective factor as a ‘quality of a person or context or their interaction that

predicts better outcomes, particularly in situations of risk or adversity’, for example, family

cohesion or involvement in community activities. Thus, we were interested in selecting

independent risk and protective factors that had the above characteristics, not simply

factors that could be labelled risk or protection depending on where they fall on the same

continuum. This framework also capitalizes on the explanatory power that resiliency

modelling has had on coping and health-related outcomes. Adolescent health and

developmental outcomes have increasingly been studied from a resiliency perspective,

with investigators mapping the interactive relationships among risk and protective factors

that result in positive outcomes for youth (Eccles & Gootman, 2002; Lerner, 2004).

Researchers have proposed several models of resiliency including the compensatory

model, the challenge model, and two variations of the protective factor model (Garmezy,

Masten, & Tellegen, 1984; Masten, Garmezy, Tellegen, Pellegrini, Larkin, & Larsen, 1988;

Zimmerman & Arunkumar, 1994). These models each propose differing ways in which

risk and protection relate to each other to influence outcome variables (see Figure 1). In the

compensatory model, protective factors reduce negative outcomes, regardless of level of

risk exposure. This is an additive model in which risk and compensatory factors each have

an independent and direct effect without interaction. (Dubow & Tisak, 1989). The

challenge model describes a curvilinear relationship between risk and the outcome

variable, such that some risk exposure is actually more beneficial than no exposure to risk

in reducing the negative outcome (Zimmerman & Arunkumar, 1994). In the risk–protective

model, protective factors interact with risk factors in a buffering effect to reduce the

negative outcome (Dubow & Luster, 1990). Finally, the protective–protective model

describes a relationship in which the addition of each protective factor further reduces the

impact of risk on the negative outcome. (Hollister-Wagner, Foshee, & Jackson, 2001).

To our knowledge, only one recent study has been conducted to understand how risk and

protective factors work together to influence the development of SOC among adolescents

(Marsh, Clinkinbeard, Thomas, & Evans, 2007). In that study, the authors reported that

higher levels of social support and neighbourhood cohesion were significantly associated

DOI: 10.1002/casp

Sense of coherence and resiliency 31

with higher levels of SOC for males and females, while higher levels of anger expression

and family conflict were significantly associated with lower levels of SOC for males and

females. Further, positive community views regarding gang membership were predictors

of lower SOC in males only, while being older was associated with higher SOC in females

only. We are not aware, however, of any studies that have specifically tested models of

resiliency in relation to SOC or focused on how cumulative risk and protection factors from

differing ecological levels of adolescent life work in various manners to contribute to SOC.

To address this notable gap in SOC research, we developed three guiding research

objectives for the present study. The first research objective was to identify which of the

four resiliency models noted previously best explain SOC among adolescents using

comprehensive measures of risk and protection. The second research objective was to test

the relative influence of risk and protection on SOC by environmental domain. In other

words, we were interested in which ecological domains might be most strongly related to

SOC when examined by risk or protection. The third objective was to examine the

association between SOC and ecological domain through the lens of ‘comfort arenas’ (Call

& Mortimer, 2001). Conceptually, comfort arenas deals with the availability of sanctuaries

for youth to which they can withdraw to recuperate from developmental stressors. For

example, in the context of SOC, it would be expected that youth who experience

cumulative risk factors across more ecological levels will exhibit a lower SOC than youth

who experience cumulative risk in fewer ecological levels.

Pursuing these research objectives will broaden our understanding of factors associated

with SOC during an important developmental period. In doing so, we recognize individuals

are producers and products of multiple environments by including risk and protective

variables from the individual, peer, family, school and neighbourhood ecological levels.

Risk factors in this case may be threats or vulnerabilities, internal or external, which inhibit

Figure 1. Representative graphs of four models of resiliency in relation to SOC.

DOI: 10.1002/casp

32 W. P. Evans et al.

the development of a high SOC. Protective factors, on the other hand, would be those

resources which help to promote the development of a high SOC. Within this framework,

we explored the roles of risk and protective factors by sex given mixed findings about

gender and SOC (Antonovsky, 1998), and the reality that development often differs for

adolescent males and females (Hopkins, 1983).

METHOD

Procedure

Data for this study were collected via in-school surveys of 8th and 10th grade students

conducted between 1998 and 2001 in selected Arizona, California, Nevada and Wyoming

schools. These data are part of a longitudinal and multi-state project focused on resilience

to youth violence. Students with parental permission responded to a 142-item survey

designed to assess individual, family, school and community concerns and resources in

their everyday lives. More detailed explication of methods and procedures has been

previously published (Evans, Marte, Betts, & Silliman, 2001).

Participants

Data were collected from a sample of 1619 8th and 10th grade students attending schools in

the Western United States. Eighth grade students (n ¼ 1,147) were from rural and urban schools of Arizona (35.6%), Wyoming (26.8%), California (19.9%) and Nevada (17.8%).

Tenth grade students (n ¼ 472) were all from rural and urban schools in Nevada. Individual school district policies required surveys to be administered with passive consent in Arizona

and Wyoming and active consent in California and Nevada. As a result, over 90 per cent of

the students at each school site in Arizona and Wyoming completed the survey, compared

to a little over one-third of the students at each California and Nevada school. Analyses

were conducted to check for effects of type of consent procedure on key study variables

(e.g. SOC, socioeconomic status (SES) and ethnicity), but the differences were not

significant. A slight majority (52.4%) of the students resided in urban areas. The sample

consisted of 46.8% male and 53.2% female. More than half the sample (55.7%) was White,

followed by Hispanic (19.7%), Asian (7.6%), Multiethnic (7.0%), Native American (4.3%)

and Black (2.5%).

Measures

The measures for this study are derived primarily from prior research involving adolescent

resiliency that focuses on risk and protective factors at various levels of environment and

how they influence outcomes for youth (e.g. see Resnick et al., 1997; Richman & Fraser,

2001; Zimmerman & Arunkumar, 1994). Given the global orientation of SOC, our intent

was to select a mix of variables from the literature that reflect the universe of experiences

youth have in their day-to-day lives. Specifically, we selected variables that (a) appeared on

face or through prior research to be theoretically related to the SOC construct, (b)

represented both risk and protective factors in the lives of youth and (c) reflected

experiences or characteristics in five domains of environment (i.e. individual, home, peer,

DOI: 10.1002/casp

Sense of coherence and resiliency 33

school and community). Analyses were conducted separately by sex which was determined

from respondent’s endorsement of being either female or male. Variables included in the

present study are further detailed below.

Sense of coherence

SOC was the dependent variable of interest, and was measured via 13 items from the

Children Sense of Coherence Scale (Margalit & Efrati, 1996). The scale assesses to what

degree youth feel confidence in their world as expressed through their understanding of

their environment (e.g. ‘When someone gets mad at me, I understand why’); feelings of

control, and confidence that help will be available when needed (e.g. ‘When I need help,

there is always someone around to help me’); and motivation and interest in investing

personal efforts (e.g. ‘I care about what goes on around me’). Response options ranged

from 1 ¼ almost never to 4 ¼ almost always. After reverse coding several items, the item scores were averaged to yield the SOC score (a ¼ 0.71). Possible scores ranged from 1 to 4 with higher scores indicating stronger SOC.

Independent variables

Cumulative protective index. The cumulative protective index was created by

selecting measures of protective factors spanning the individual, home, peer, school and

community domains of adolescent environment (see Table 1). The indexing procedure

described here has been used by others in research similar to the present study (e.g. see

Table 1. Descriptors of protective indicators

Protective variable M males/ M females Domain Items Criterion

Protected males/Protected females (%)

Religion 3.23/3.07 Individual 1 �25th percentile 33.0/36.3 Emotional stability 1.94/2.23 Individual 7 �25th percentile 28.1/31.7 Parental monitoring 3.93/4.12 Home 3 �75th percentile 33.5/30.1 Place NA Home 1 Yes 71.7/73.5 Decisions NA Home 1 Yes 38.7/47.8 Parents care 4.68/4.57 Home 1 �75th percentile 81.0/77.1 Family understands 3.30/3.08 Home 1 �75th percentile 46.5/38.7 Want to run away 1.81/2.14 Home 1 �25th percentile 55.8/42.2 Family fun 3.49/3.32 Home 1 �75th percentile 53.6/48.0 Family pays attention 3.68/3.58 Home 1 �75th percentile 23.9/25.4 Friends care 3.76/4.18 Peer 1 �75th percentile 64.7/43.1 School activities 2.79/2.71 School 1 �75th percentile 34.4/28.7 Teachers care 3.12/3.23 School 1 �75th percentile 38.9/39.0 Like school 2.87/2.57 School 1 �25th percentile 74.8/55.3 Community activities 2.77/2.55 Community 1 �75th percentile 32.5/47.3 Neighbourhood cohesion 2.75/2.72 Community 7 �75th percentile 31.5/20.5

Note: Quartiles noted represent high positive/protective responses. Youth who met the protective criterion for any protective factor were coded 1 while all others were coded 0. The number of protective factors was summed for each youth to create a cumulative protective index score, which could range from 0 to 16. Male protective index: n ¼ 646, M ¼ 7.48, SE ¼ .125, SD ¼ 3.19. Female protective index: n ¼ 743, M ¼ 6.96, SE ¼ .125, SD ¼ 3.41.

DOI: 10.1002/casp

34 W. P. Evans et al.

Gerard & Buehler, 2004; Sameroff, Bartko, Baldwin, Baldwin, & Seifer, 1998). All items

included in the protective index were appropriately correlated with SOC in terms of

direction and strength (r for items ranged from 0.03 to 0.38 for males and 0.02 to 0.39 for

females). Individual domain protective factors were assessed by two measures, one

involving the influence of religion and one involving emotional stability. Home domain

protective factors were assessed by eight measures, including: parental monitoring; youth

feeling they have a place in the home that belongs to them; youth feeling they are involved

in decisions in the home; youth desire to run away from home; and the extent youth feel

parents care about them, their family understands them, their family has fun together, and

their family pays attention to them. A peer domain protective factor was assessed by one

item involving the extent youth feel friends care about them. School domain protective

factors included three items reflecting the extent of involvement in school activities, the

extent youth feel teachers care about them, and the degree to which they enjoy school.

Community level protective factors spanned two measures, one assessing the degree to

which youth are involved in community activities, and one assessing neighbourhood

cohesion. With the exception of emotional stability, parental monitoring, and

neighbourhood cohesion, all protective factor measures consisted of one item. Emotional

stability consisted of seven items measuring the degree to which various emotional

problems and complaints have bothered respondents during the past 30 days (e.g. feeling

blue or sad; Evans & Skager, 1992; a ¼ 0.80), parental monitoring of three items (i.e. parents know where youth is after school and at night, and youth discusses plans with

parents; a ¼ 0.75), and neighbourhood cohesion of seven items from Buckner’s (1988) Neighbourhood Cohesion Instrument (a ¼ 0.74). As seen in Table 1, youth who responded affirmatively to dichotomous items reflecting protection, or who scored in the extreme

positive (protective) quartile in their responses, were assigned a value of 1 meaning the

protective factor was present. All others were assigned a value of 0 meaning the protec-

tive factor was not present. Scores were summed to create the protective factor index, and

could range from 0 to 16, with 16 reflecting more protection (a ¼ 0.70 for males; a ¼ 0.74 for females).

Cumulative risk index. The same basic procedure was repeated to create the

cumulative risk index. Risk factors were selected at the individual, home, peer, school and

community domains (see Table 2). All items included in the risk index were appropriately

correlated with SOC in terms of direction and strength (r for items ranged from �0.06 to �0.37 for males and �0.02 to �0.44 for females). Individual risk factors were assessed through measures of involvement in fighting, having been threatened or attacked, using a

weapon to obtain something from someone, problem behaviours (i.e. delinquency), suicide

attempts, being a victim of crime, use of alcohol, use of marijuana and anger expression.

Home level risk factors included measures of living with one parent and family conflict.

Peer domain risk was assessed through four items, including negative peer influence, gang

involvement, having friends or family in a gang and peer attitudes toward drug use. School

domain risk was measured via one item involving the number of times the youth had

skipped school in the last month. Lastly, community domain risk factors included

community views regarding dropping out of school, community views regarding gang

membership and community views regarding violence as a way to solve problems. With the

exception of the fighting, problem behaviour, anger expression, family conflict and peer

influence risk factors, all risk measures consisted of one item. Involvement in fighting was

assessed via two items (i.e. been in a fight or started a fight in the last 6 months; r ¼ 0.57,

DOI: 10.1002/casp

Sense of coherence and resiliency 35

p � 0.01), problem behaviour by seven items (i.e. graffiti, vandalism, shoplifting, run away from home, burglary, threaten with a weapon and group fighting; a ¼ 0.80), anger expression by 12 items from Spielberger’s (1991) State-Trait Anger Expression Inventory

(a ¼ 0.84), family conflict by five items from a subscale of Bloom’s (1985) Family Functioning Scales (a ¼ 0.70), and peer influence by three items (i.e. fear of disapproval, doing things to be more popular and letting friends talk youth into doing things they do not

want to do; a ¼ 0.60). Youth who responded affirmatively to dichotomous items reflecting risk, or who scored in the extreme negative (risk) quartile in their responses, were assigned

a value of 1 meaning the risk factor was present. All others were assigned a value of 0

meaning the risk factor was not present. Scores were summed to create the risk factor index

and could range from 0 to 19, with 19 reflecting more risk (a ¼ 0.77 for males; a ¼ 0.78 for females).

Covariates. Age, SES and ethnicity were included as covariates to control for potential

developmental, resource and cultural effects on SOC levels. Age was measured in years

only and ranged from 12 to 16 years (M ¼ 14.29). Socioeconomic status was measured with a composite variable that included mother’s education, father’s education and perceived

family income. Education of mother/stepmother and father/stepfather was scored as

1 ¼ completed elementary, junior high, or high school; 2 ¼ some college or technical school or graduated from 2-year college or technical school; and, 3 ¼ graduated from a 4-year college, some school beyond a 4-year college degree, or professional or graduate

Table 2. Descriptors of risk indicators

Risk variable M males/ M females Domain Items Criterion

At-risk males/At-risk females (%)

Fighting 0.23/0.13 Individual 2 More than once 23.0/12.7 Threatened or attacked 0.71/0.55 Individual 1 �75th percentile 15.3/9.5 Used weapon 0.29/0.13 Individual 1 Yes 17.3/8.5 Problem behaviour 1.33/1.21 Individual 7 �75th percentile 20.7/19.6 Suicide attempt 1.15/1.31 Individual 1 �75th percentile 9.8/20.9 Victim of crime 0.77/0.48 Individual 1 Yes 45.9/32.0 Beer use 2.50/2.40 Individual 1 �75th percentile 37.5/36.1 Marijuana use 2.11/1.92 Individual 1 �75th percentile 23.0/21.2 Anger expression 2.16/2.09 Individual 12 �75th percentile 23.0/29.9 Living with one parent NA Home 1 Yes 22.6/25.2 Family conflict 2.27/2.36 Home 5 �75th percentile 33.8/29.1 Peer influence 2.01/1.85 Peer 3 �75th percentile 25.6/29.0 Gang member NA Peer 1 Yes 10.1/5.2 Friends/family in gang 0.50/0.56 Peer 1 Yes 31.8/34.9 Peer attitude re: drugs 1.91/1.96 Peer 1 �75th percentile 25.5/27.4 Times skipped school 1.55/1.52 School 1 �75th percentile 28.0/30.2 Views re: dropping out 0.41/0.44 Community 1 �75th percentile 35.8/40.0 Views re: gang membership 0.29/0.29 Community 1 �75th percentile 23.8/23.6 Views re: violence 0.42/0.39 Community 1 �75th percentile 35.1/32.7

Note: Quartiles noted represent high negative/risk responses. Youth who met the risk criterion for any risk factor were coded 1 while all others were coded 0. The number of risk factors was summed for each youth to create a cumulative risk index score, which could range from 0 to 19. Male risk index: n ¼ 592, M ¼ 4.72, SE ¼ .145, SD ¼ 3.53. Female risk index: n ¼ 695, M ¼ 4.56, SE ¼ .136, SD ¼ 3.58.

DOI: 10.1002/casp

36 W. P. Evans et al.

degree. To measure perceived family income, students were asked what their parents’

income was in comparison to most people. Responses were categorized as 1 ¼ lower than average or among the lowest, 2 ¼ about average and 3 ¼ higher than average or among the highest. The mean of the three variables comprised the SES variable. Since over half the

sample was White and some of the other categories were very small, ethnicity was

dichotomized as 0 ¼ White and 1 ¼ non-White.

RESULTS

Test of resiliency models

Hierarchical multiple regression procedures were used to test the four resiliency models

and followed analysis strategies described by Garmezy et al. (1984), Hollister-Wagner

et al. (2001), and Pollard, Hawkins, & Arthur (1999). Hierarchical multiple regression has

the advantage of partitioning the total variance in the dependent variable because it

identifies the unique contribution of each independent variable when included as a separate

block in the analysis. In the present analysis, the covariates, including ethnicity, SES, and

age were entered in block 1. Then the main effects for risk and protection were entered in

blocks 2 and 3, respectively. This step tested the strength of the compensatory model,

which holds that protective factors reduce negative outcomes, regardless of the level of risk

exposure (Zimmerman & Arunkumar, 1994). To test the risk–protective and protective–

protective models, the risk � protection interaction term was entered in block 4 of the hierarchical regression.]

1 The risk–protective model posits that protective factors interact

with risk factors in a buffering effect to reduce negative outcomes (Richman & Fraser,

2001; Zimmerman & Arunkumar, 1994). The protective–protective model is a variation of

the risk–protective model that suggests with the addition of each protective factor, the

relationship between the risk factors and outcome is further weakened (e.g. see Hollister-

Wagner et al., 2001). Finally, to test the challenge model the risk � risk quadratic term was entered in block 5. The challenge model predicts that the relationship between risk factors

and the outcome is curvilinear in that some exposure to risk factors increases the likelihood

of a negative outcome (Rutter, 1987). These regression analyses were conducted separately

for males and females (see Tables 3 and 4). Prior to analyses, the protective index was

weighted by a factor of 1.1875 (i.e. 19/16) given it consisted of 16 items versus 19 items in

the risk index.

Results of the hierarchical multiple regression analyses revealed that the compensatory

model was supported for both males and females, while the challenge model was supported

for females only. In regards to the challenge model, data plots revealed that SOC levels for

females in this sample increased with the presence of zero to approximately eight risk

factors, but beyond that point, SOC levels declined. The risk–protective and protective–

protective models were not supported for males or females. Age for females was the only

1 The protective-protective model is usually tested through a separate regression equation that replaces the risk � protection interaction term at block 4 with a risk � number of protective variables interaction term. In this case, however, the interaction term is the same due to how the indexes were constructed (i.e., ‘‘count’’ indexes), thus a separate equation is not necessary. Specifically, the risk-protective model [block 4] is assessed for significance immediately after that block is entered in the equation (significance at entry indicates support for the risk-protective model), while the protective-protective model [block 4] is assessed for significance after the final block [block 5] is entered in the equation (significance with all blocks entered indicates support for the protective- protective model).

DOI: 10.1002/casp

Sense of coherence and resiliency 37

significant covariate, with being older associated with higher levels of SOC. The adjusted

R 2 for the full model was 0.390 for males and 0.397 for females.

Ecological influences on SOC

Given evidence of significant protective and risk predictors across environmental domains,

we explored the nature of ecology in this context more directly by examining mean

differences in SOC scores among youth with zero, one, two, three, four and five affected

protective (or risk) domains. In other words, we sought to better understand the relationship

between SOC and comfort arenas of youth in this sample. Analyses again were conducted

separately by sex.

To explore the cumulative effect of protection across the five ecological domains, we

tallied the number of affected domains for each youth based on the presence of at least one

considered protective factor. This procedure was conducted separately by sex. This

resulted in five groups of males and six groups of females, with the number of domains

containing a protective factor ranging from one to five and zero to five, respectively. To

maintain consistency in controlling for ethnicity, SES, and age in all analyses, an analysis

of covariance (ANCOVA) was employed to explore for differences in SOC means across

domains. Overall models were significant for both males and females and significant

Table 3. Summary of regression analyses by sex testing relative influence of protective factors by domains predicting sense of coherence

Males (n ¼ 477) Females (n ¼ 551)

R 2

R 2 change R

2 R 2 change

Step 1: Covariates 0.011 0.011 0.052 0.052 ��

Step 2: Individual level 0.270 0.259 ��

0.350 0.298 ��

Step 3: Home level 0.414 0.144 ��

0.474 0.124 ��

Step 4: Peer level 0.436 0.022 ��

0.506 0.032 ��

Step 5: School level 0.442 0.006 0.511 0.006 Step 6: Community level 0.460 0.018

�� 0.518 0.006

�

Note: Each level contains all protective predictor variables in that domain. R 2 is presented for the model after each

block. � p � 0.05; ��p � 0.001.

Table 4. Summary of regression analyses by sex testing relative influence of risk factors by domains predicting sense of coherence

Males (n ¼ 436) Females (n ¼ 519)

R 2

R 2 change R

2 R 2 change

Step 1: Covariates 0.018 0.018 0.049 0.049 ��

Step 2: Individual level 0.333 0.315 ��

0.367 0.317 ��

Step 3: Home level 0.418 0.085 ��

0.424 0.057 ��

Step 4: Peer level 0.431 0.013 0.450 0.026 ��

Step 5: School level 0.431 0.000 0.451 0.001 Step 6: Community level 0.452 0.021

�� 0.455 0.004

Note: Each level contains all risk predictor variables in that domain. R 2 is presented for the model after each block.

�� p � 0.001.

DOI: 10.1002/casp

38 W. P. Evans et al.

contrasts were relatively stable across levels (see Table 5). Results indicate that more

domains with a protective factor present are related to higher SOC scores for both males

and females.

The same grouping and ANCOVA procedure was repeated with risk measures and

domains. Males and females were grouped based on how many of the five ecological

domains had at least one considered risk factor present. This resulted in six groups for both

males and females, with the number of domains with a risk factor present ranging from zero

to five. Overall models were significant for both males and females and significant contrasts

were relatively stable across levels (see Table 6). Results indicate that the more domains

with a risk factor present are related to lower SOC scores for both males and females.

Table 5. Summary of analysis of covariance by sex for sense of coherence across domains with protection

Group F-value Number of domains with protection M

Significant contrasts between domain levels

Males 31.50 ��

0 n/a 1 2.35 1 vs. 2, 3, 4, & 5 2 2.81 2 vs. 4 & 5 3 2.93 3 vs. 5 4 3.00 4 vs. 5 5 3.25

Females 20.72 ��

0 2.55 0 vs. 4 & 5 1 2.68 1 vs. 4 & 5 2 2.77 2 vs. 4 & 5 3 2.90 3 vs. 4 & 5 4 3.06 4 vs. 5 5 3.24

Note: Means reported are for SOC measure. Covariates include ethnicity (White vs. non-White), SES and age. All contrasts are significant at p � 0.05 with Bonferroni correction to address the possibility of Type I errors associated with the number of multiple comparisons conducted. ��

p < 0.001.

Table 6. Summary of analysis of covariance by sex for sense of coherence across domains with risk

Group F-value Number of

domains with risk M Significant contrasts

between domain levels

Males 23.56 ��

0 3.30 0 vs. 2, 3, 4, & 5 1 3.23 1 vs. 3, 4, & 5 2 3.08 2 vs. 4 & 5 3 2.97 3 vs. 4 & 5 4 2.81 5 2.64

Females 23.58 ��

0 3.29 0 vs. 2, 3, 4, & 5 1 3.16 1 vs. 3, 4, & 5 2 3.02 2 vs. 4, & 5 3 2.89 3 vs. 5 4 2.75 5 2.71

p < .001.

DOI: 10.1002/casp

Sense of coherence and resiliency 39

DISCUSSION

The present study was guided by three research questions involving adolescent SOC, risk,

protection and resiliency. The first question sought to identify which of four models of

resiliency best explain SOC among adolescents. Although SOC has been conceptualized as

a condition of resiliency, linked to well-being, and a product of developmental risk and

protective processes, no previous studies known to the authors have examined potentially

useful heuristic resiliency models for support. Analyses revealed that the compensatory

model was supported for both males and females, while the challenge model was supported

for females only. Support for the compensatory model indicates the independent and direct

effects of protective factors on SOC among adolescents, regardless of level of risk

exposure. This emphasizes the additive and central role protective factors play in the

development of SOC among both adolescent girls and boys, with an accumulation of

protective factors associated with linear increases in SOC. Interestingly, support for the

challenge model also was found for girls, revealing a curvilinear relationship where a small

exposure to risk was more conducive to SOC development than no exposure. Although

present findings underscore the general importance of protective factors to SOC, this

intriguing finding indicates, among girls, that a little experience with risk can help

positively shape SOC (at least until a ’tipping’ point is reached). This is consistent with

Antonovsky’s articulation of SOC development in terms of a person needing certain life

shaping experiences that help develop a personal balance for stress and self regulation

(Antonovsky, 1998). The support of this model only among girls, however, is difficult to

explain given our current results. This could be an artifact of the specific risk and protective

factors we selected, the variable indexing and clustering procedures we employed, or a

reflection of important differing SOC developmental processes among girls and boys. This

is a topic for future investigation.

The second research question dealt with the relative influence of protection and risk on

SOC across ecological domains. Findings suggest that protective and risk factors

significantly influencing SOC emerge from multiple domains in the environment for

both males and females. Given this, we then sought to explore the ecological context of

these factors for our last research question by illuminating the role of comfort arenas on

SOC for adolescents. Analyses revealed that the more ecological domains with protection,

the higher the SOC for both males and females; while the more ecological domains

with risk, the lower the SOC for both males and females. This last set of analyses

addressed the issue of how resources and stressors that occur simultaneously across

several life domains result in SOC. Our findings support the idea that comfort arenas also

play an important role in the expression of SOC. Thus, not only support for the

compensatory model was found (and in the case of the girls in our sample, the challenge

model) but also support for the contributions of risk and protection across ecological

domains, and further, whether these domains within an adolescent’s life are experienced as

a source of support or stress, also help determine the expression of SOC—with increased

SOC linked to more domains of support and lower SOC linked to more domains of stress.

Similar to the findings of Simmons, Burgeson, Carlton-Ford, & Blyth (1987) and Gerard

and Buehler (2004), this suggests that while protective factors appear critical in developing

high SOC, it also appears that having safe and nurturing social supports or perceived

ecological sanctuaries that can be a refuge from challenges and risk influences, aids in the

positive development of SOC (with the reverse also leading to lower levels of SOC). Thus

present findings confirm the global, ecological nature of this construct, with the selected

DOI: 10.1002/casp

40 W. P. Evans et al.

risk and protective factors, and ecological domains of this study each contributing to the

formation of SOC.

Cumulatively, these findings indicate that a resiliency approach appears to be useful in

exploring how SOC manifests among adolescents. They also provide intersecting evidence

that further bridges the gap between prevention science and youth development

frameworks. Current results reveal that risk and protection play major roles in SOC

expression and help provide support that prevention and youth development frameworks

are not fundamentally competitive, but rather can be mutually constructive in

understanding adolescent development and promoting positive outcomes. This endorses

the view of Catalano, Hawkins, Berglund, Pollard, & Arthur (2001), who conclude that

prevention science and youth development frameworks can be mutually beneficial. More

work will need to be done, however, to understand how present results correspond to

differing development periods and trajectories. For example, our data were age constricted

and thus we were not able to compare across developmental periods how risk, protection

and comfort arenas of adolescent life interact to affect SOC development or what factors

and life domains are most salient at differing developmental stages. Future studies need to

examine younger samples and longitudinally explore SOC development throughout

adolescence and into young adulthood. Although prior studies have found SOC to

stabilize in the late 20s (Kivamaki et al., 2000; Malmgren-Olsson & Bränholm, 2002),

little empirical data have been collected to confirm the developmental trajectory of

this construct, or how it relates to various physical and mental well-being outcomes over

time.

We developed models separately by sex because of developmental and gender

socialization process differences (Gilligan, 1982), as well as the lack of clarity regarding

sex and SOC (Antonovsky, 1998). Findings revealed similarities between females and

males on most of our research questions, with the exception of support for the challenge

model only among females. Longitudinal studies may elucidate these findings, and

determine if males, at different developmental periods, also reveal similar dynamics that

led to support of the challenge model among the females in our sample. Ethnicity also is an

important future SOC research topic; in our present analyses, we had to dichotomize

ethnicity due to sample characteristics, which likely masked important ethnic differences.

These endeavours will be critical, particularly since it has been found that sex,

developmental period and ethnicity each play significant roles in how adolescents cope

with social stressors (Gerard & Buehler, 2004).

Several limitations must be noted in interpreting results. First, the indexing of risk and

protective factors reduced variability and some of our capacity to recognize subtle

influences in the data, such as mediation between and relative weight of measures within

domains. Further, the nature of the available data resulted in variations in the number of

factors assessed in each domain, which could under- or over-emphasize the importance of

certain domains in relation to SOC or effect the likelihood youth experience risk or

protection in a given domain. Although measures of risk and protective factors were

carefully selected from the best available research, other tools and indicators may have

revealed different trends. In addition, the cross-sectional design and survey method limit

conclusions about causality and the processes by which resiliency occur. Future studies that

incorporate qualitative and longitudinal data may add depth to our understanding of SOC.

Third party assessment of SOC also could provide convergent validity to future studies, as

well as help validate the scales currently used for SOC research. Nonetheless, present

findings reveal strong support for the link between protective factors and SOC in this

DOI: 10.1002/casp

Sense of coherence and resiliency 41

sample, and suggest an asset-focused framework for those seeking to understand and

nurture this important health-related trait in adolescents as they transition to adulthood.

The future investigation of the etiology of SOC can only be justified by the continued

exploration of the correlates of SOC to important social- and health-related outcomes, and

how SOC might then be modified to produce positive social- and health-related trajectories

and interventions. Although a number of linkages have been found between SOC and

specific conditions (e.g. Amirkhan & Greaves, 2003), further investigation is needed to

identify how SOC is related to important health and psychosocial processes among adults,

as well as significant developmental outcomes among youth. Current findings reveal how

specific risk and protective factors interact across ecological domains to produce high

levels of SOC. Additional work needs to be done, however, in terms of what are the most

salient risk and protective factors in connection to SOC, and how they relate

developmentally into middle adulthood. Future studies that attempt to deconstruct such

a global construct as SOC so that these linkages can be highlighted and interventions

developed, will bring increased clarity to SOC and its relationship to well-being and

resiliency.

ACKNOWLEDGEMENTS

This study was supported in part by a grant from USDA-CSREES for Cooperative

Regional Research project W-193 (Resilience to Violence Among At-Risk Youth).

REFERENCES

Amirkhan, J. H., & Greaves, H. (2003). Sense of coherence and stress: The mechanics of a healthy disposition. Psychology and Health, 18, 31–62.

Antonovsky, A. (1987). Unraveling the mystery of health: How people manage stress and stay well. San Francisco: Jossey-Bass.

Antonovsky, A. (1998). The sense of coherence: An historical and future perspective. In H. I. McCubbin, E. A. Thompson, A. I. Thompson, & J. E. Fromer (Eds.), Stress, coping, and health in families: Sense of coherence and resiliency (pp. 3–21). Thousand Oaks, CA: Sage.

Bloom, B. (1985). A factor analysis of self-report measures of family functioning. Family Process, 24, 225–239.

Buckner, J. C. (1988). The development of an instrument to measure neighborhood cohesion. American Journal of Community Psychology, 16, 771–791.

Call, K. T., & Mortimer, J. T. (2001). Arenas of comfort in adolescence: A study of adjustment in context. Mahwah, NJ: Lawrence Erlbaum Associates.

Catalano, R. F., Hawkins, J. D., Berglund, M. L., Pollard, J. A., & Arthur, M. W. (2001). Prevention science and positive youth development: Competitive or cooperative frameworks? Journal of Adolescent Health, 31, 230–239.

Dubow, E. F., & Luster, T. (1990). Adjustment of children born to teenage mothers: The contribution of risk and protective factors. Journal of Marriage and the Family, 52, 393–404.

Dubow, E. F., & Tisak, J. (1989). The relation between stressful life events and adjustment in elementary school children: The role of social support and problem-solving skills. Child Devel- opment, 60, 1412–1423.

Eccles, J., & Gootman, J. A. (2002). Community programs to promote youth development. Washington, DC: National Academy Press.

Evans, W. P., Marte, R. M., Betts, S., & Silliman, B. (2001). Adolescent suicide risk and peer-related violent behaviors and victimization. Journal of Interpersonal Violence, 16, 1330–1348.

Evans, W. P., & Skager, R. (1992). Academically successful drug users: An oxymoron? Journal of Drug Education, 22, 353–365.

DOI: 10.1002/casp

42 W. P. Evans et al.

Fiorentino, L. M. (1986). Stress: The high cost to industry. American Association of Occupational Health Nursing, 34, 217–220.

Garmezy, N., Masten, A. S., & Tellegen, A. (1984). The study of stress and competence in children: A building block for developmental psychopathology. Child Development, 55, 97–111.

Gerard, J. M., & Buehler, C. (2004). Cumulative environmental risk and youth problem behavior. Journal of Marriage and Family, 66, 702–720.

Gilligan, C. (1982). In a different voice: Psychological theory and women’s development. Cambridge, MA: Harvard University Press.

Hollister-Wagner, G. H., Foshee, V. A., & Jackson, C. (2001). Adolescent aggression: Models of resiliency. Journal of Applied Social Psychology, 31, 445–466.

Hopkins, J. R. (1983). Adolescence: The transitional years. New York: Academic Press. Kivamaki, M., Feldt, T., Vahtera, J., & Nurmi, J. (2000). Sense of coherence and health: Evidence from two cross-lagged longitudinal samples. Social Science & Medicine, 50, 583–597.

Lerner, R. M. (2004). Liberty: Thriving and civic engagement among American youth. Thousand Oaks, CA: Sage.

Lutgendorf, S. K., Vitaliano, P. P., Tripp-Reimer, T., Harvey, J. H., & Lubaroff, D. M. (1999). Sense of coherence moderates the relationship between life stress and natural killer cell activity in healthy older adults. Psychology and Aging, 14, 552–563.

Malmgren-Olsson, E. B., & Bränholm, I. B. (2002). A comparison between three physiotherapy approaches with regard to health-related factors in patients with non-specific musculoskeletal disorders. Disability & Rehabilitation, 24, 308–317.

Margalit, M., & Efrati, M. (1996). Loneliness, coherence and companionship among children with learning disorders. Educational Psychology, 16, 69–79.

Margalit, M., & Eysenck, S. (1990). Prediction of coherence in adolescence: Gender differences in social skills, personality, and family climate. Journal of Research in Personality, 24, 510–521.

Marsh, S. C., Clinkinbeard, S. S., Thomas, R. M., & Evans, W. P. (2007). Risk and protective factors predictive of sense of coherence during adolescence. Journal of Health Psychology, 12, 281–284.

Masten, A. S., Garmezy, N., Tellegen, A., Pellegrini, D. S., Larkin, K., & Larsen, A. (1988). Competence and stress in school children: The moderating effects of individual and family qualities. Journal of Child Psychology and Psychiatry, 29, 745–764.

McCubbin, H. I. (1998). Series preface: Families at their best. In H. I. McCubbin, E. A. Thompson, A. I. Thompson, & J. E. Fromer (Eds.), Stress, coping, and health in families: Sense of coherence and resiliency (pp. xii–xv). Thousand Oaks, CA: Sage.

Pollard, J. A., Hawkins, J. D., & Arthur, M. W. (1999). Risk and protection: Are both necessary to understand diverse behavioral outcomes in adolescence? Social Work Research, 23, 145–158.

Resnick, M. D., Bearman, P. S., Blum, R. W., Bauman, K. E., Harris, K. M., & Jones, J. O., et al. (1997). Protecting adolescents from harm: Findings from the national longitudinal study on adolescent health. Journal of the American Medical Association, 278, 823–832.

Richman, J. M., & Fraser, M. W. (2001). Resilience in childhood: The role of risk and protection. In J. M. Richman, & M. W. Fraser (Eds.), The context of youth violence: Resilience, risk, and protection (pp. 1–12). Westport, CT: Praeger.

Rutter, M. (1987). Psychosocial resilience and protective mechanisms. American Journal of Orthopsychiatry, 57, 316–331.

Sagy, S., & Antonovsky, H. (2000). The development of the sense of coherence: A retrospective study of early life experiences in the family. Journal of Aging and Human Development, 51, 155–166.

Sameroff, A. J., Bartko, W. T., Baldwin, A., Baldwin, C., & Seifer, R. (1998). Family and social influences on the development of child competence. In M. Lewis, & C. Feiring (Eds.), Family, risk, and competence (pp. 161–185). Mahwah, NJ: Lawrence Erlbaum Associates.

Simmons, R. G., Burgeson, R., Carlton-Ford, S., & Blyth, D. (1987). The impact of cumulative change in early adolescence. Child Development, 58, 1220–1234.

Spielberger, D. (1991). State-trait anger expression inventory: Revised research edition professional manual. Odessa, FL: Psychological Assessment Resources.

Wright, M. O., & Masten, A. S. (2005). Resilience processes in development: Fostering positive adaptation in the context of adversity. In S. Goldstein, & R. B. Brooks (Eds.), Handbook of resilience in children (pp. 17–37). New York: Kluwer Academic/Plenum Publishers.

Zimmerman, M., & Arunkumar, R. (1994). Resiliency research: Implications for school and policy. Social Policy Report: Society for Research in Child Development, 8, 1–17.

DOI: 10.1002/casp

Sense of coherence and resiliency 43

1996 and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright

holder's express written permission. However, users may print, download, or email articles for individual use.