Sei sulla pagina 1di 10

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/258183834

The Caregiver Role Identity Scale A Validation Study

Article  in  Research on Social Work Practice · May 2005


DOI: 10.1177/1049731504272779

CITATIONS READS
24 672

2 authors:

Darcy Clay Siebert Carl Siebert


Rutgers, The State University of New Jersey Boise State University
39 PUBLICATIONS   601 CITATIONS    23 PUBLICATIONS   230 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Measuring Human Rights in Social Work View project

Primary Mathematics Assessment View project

All content following this page was uploaded by Carl Siebert on 10 June 2014.

The user has requested enhancement of the downloaded file.


RESEARCH ON SOCIAL WORK PRA
10.1177/1049731504272779 CTICE Siebert / IDENTITY SCALE VALIDATION

The Caregiver Role Identity Scale


A Validation Study

Darcy Clay Siebert


Carl F. Siebert
Florida State University
Objective: This article reports the validation of the Caregiver Role Identity Scale, designed to measure the promi-
nence of helping professionals’identity as personal and professional caregivers. The authors developed the measure
to test its application to burnout, depression, and professional impairment among social workers. Method: Data from
a probability sample of 751 practicing social workers were collected in an anonymous survey about social workers’
health and work issues. The authors split the sample to conduct exploratory and confirmatory factor analyses.
Results: The exploratory analysis demonstrated good internal consistency reliability, good content validity, and pre-
liminary discriminant validity. The confirmatory analysis demonstrated excellent factorial validity for a revised
scale, retaining reliability and demonstrating convergent validity. Caregiver role identity was related to burnout,
depression, professional impairment, and not seeking help for personal problems. Conclusions: This scale may be a
useful tool for early identification, prevention, and intervention strategies for impairment among social workers.

Keywords: social workers; role identity; scale development; professional impairment

Social services workers report experiencing the same are epidemiological, describing the prevalence of burnout
sorts of personal problems that face their clients (e.g., or alcohol and drug abuse among physicians or nurses,
depression, burnout, and high-risk alcohol and other drug along with a few demographic variables (Hughes et al.,
use; Siebert, 2003, 2004, in press-b). It is not surprising 1992; McAuliffe et al., 1991; Trinkoff, 1998; Trinkoff
that these social workers also report that these personal et al., 1999; Trinkoff et al., 2000). Few studies focus on
problems can lead to professional impairment (i.e., the help seeking among caregiving professionals, although a
quality of their work is diminished or impaired by their recent study of social workers suggests several barriers to
personal problems). Although social workers are trained help seeking, including denial (Siebert, 2003, in press-a).
to assist their clients who are experiencing personal prob- The conceptual literature from other disciplines supports
lems, social workers also report not seeking assistance for this notion, proposing that professional caregivers have
their own personal issues (Siebert, in press-a). This phe- difficulty recognizing and admitting that they have per-
nomenon is not limited to social workers; researchers sonal problems (Deutsch, 1985; Kottler & Hazler, 1996;
report that other caregiving professionals (e.g., nurses, Nace, 1995b; Thoreson, Nathan, Skorina, & Kilburg,
physicians, and psychologists) also experience these 1983).
issues (Centrella, 1994; Deutsch, 1984, 1985; Doyle, In the absence of empirical information, we turned to
1987; Duquette, Kerouac, Sandhu, & Beaudet, 1994; Fel- the theoretical literature to begin to explain this troubling
ton, 1998; Good, Thoreson, & Shaughnessy, 1995; Hood situation. Role identity theory (McCall & Simmons,
& Duphorne, 1995; Hughes et al., 1992; McAuliffe et al., 1978), a middle-range theory with roots in symbolic
1991; Pope, Tabachnick, & Keith-Spiegel, 1987; interactionism, provides an excellent framework for
Sherman & Thelen, 1998; Trinkoff, 1998; Trinkoff, understanding professional impairment and profession-
Eaton, & Anthony, 1994; Trinkoff, Storr, & Wall, 1999; als’ inability to seek treatment for themselves.
Trinkoff, Zhou, Storr, & Soeken, 2000; Vaillant, 1996;
Vaillant, 1972; Wood, Klein, Cross, Lammers, & Elliott, Role Identity Theory
1985)
The empirical literature about professional impair- Role identity theory posits that behavior is motivated
ment is thin. Studies of helping professionals typically by how one sees oneself in a variety of personal and pro-
fessional roles. These identities are hierarchically
Authors’ Note: This research was funded by the National Institute on Alcohol arranged in order of salience, the most salient having the
Abuse and Alcoholism.
most influence on behavior. Role identities are created
Research on Social Work Practice, Vol. 15 No. 3, May 2005 204-212 through interactions with others in the environment but
DOI: 10.1177/1049731504272779
© 2005 Sage Publications superimposed on personal characteristics. Social workers
204
Siebert / IDENTITY SCALE VALIDATION 205

have overlapping professional and personal identities as from professional impairment experts and survey
helpers of others, reinforced by the expectations of their methodologists, pretests, and pilot tests. The final survey
clients, friends, family, and their own expectations of instrument was composed of 84 items about personal
themselves. These multiple reinforcers of whom they problems (e.g., burnout and depression), professional
think they should be could make it difficult for them to ac- impairment (i.e., the professional consequences of per-
knowledge that they have personal problems similar to sonal problems), personal and occupational variables,
those of their clients—their idealized image of them- and help seeking. This analysis uses only the data that
selves would be at odds with the reality of their personal were relevant to role identity and impairment.
problems. This would leave them unable or unwilling to We mailed questionnaires to 1,000 practicing social
acknowledge and seek help for their personal problems workers who were members of the North Carolina Chap-
and thus, vulnerable to professional impairment. The ter of NASW. We created the sample by removing inac-
additional worries about potential practical consequences tive, retired, and student members from the most current
of professional impairment—for example, loss of status, membership list of 4,203 members. Using a systematic
loss of friends and clients, loss of the right to practice— probability sampling strategy, we chose every xth name,
can contribute to a kind of superdenial about their beginning with a randomly selected starting point.
personal distress. We followed survey research methods recommended
Thus, role identity theory well explains impairment in by Dillman (2000) for anonymous, cross-sectional sur-
social workers. It illustrates the issues that are idiosyn- veys, employing data collection strategies to improve the
cratic to the role of professional caregiver, yet it allows for likelihood of obtaining representative data. Potential
individual differences. Role identity theory underscores respondents were mailed a letter introducing the study,
the importance of self-expectations in concert with oth- and a questionnaire packet followed 3 days later. Included
ers’ professional and personal expectations. Most impor- in this packet was a postcard on which respondents could
tantly, it explains the apparent paradox of highly skilled check either that they had completed and returned the
and well-trained helping professionals being unable to questionnaire or that they were not interested in partici-
acknowledge and seek help for their own personal prob- pating. Because respondents returned the postcard sepa-
lems, which can ultimately lead to impaired performance. rately from the questionnaire, we could track their partici-
However, very few measures of role identity can be pation and retain their anonymity. Until potential
found in the literature, and no measures of caregiver role respondents returned the postcard, we mailed them
identity were available to test this theory’s utility in inves- reminder postcards and replacement questionnaire
tigating burnout, depression, professional impairment, packets every 2 weeks throughout the study.
and help-seeking among social workers. This article By the completion of the data collection, 822 (82%) of
reports the development and validation of the Caregiver the social workers had responded to the mailings. How-
Role Identity Scale, an instrument designed to measure ever, 75 of these responded by stating that they did not
the strength of helping professionals’identity as both per- wish to participate, 12 potential respondents were not
sonal and professional caregivers. locatable, and three questionnaires were determined
unusable. Thus, 751 of the 1,000 potential respondents
returned usable questionnaires, producing a survey com-
METHOD pletion rate of 75%.

The Caregiver Role Identity Scale was developed for


Work and Well Being: A Study of North Carolina Social MEASURES
Workers (Seibert, 2001). This anonymous survey gath-
ered information about occupational, personal, and In addition to the development of the Caregiver Role
health issues among practicing members of the North Identity Scale, we used a number of other measures to
Carolina Chapter of the National Association of Social explore the scale’s theoretical and empirical relationship
Workers (NASW). to distress and impairment among social workers.
The survey instrument development was guided by a
thorough review of the literature and focus groups of Role Identity
social workers and directors of professional assistance
programs in other disciplines. This was an iterative pro- We began the development of the measure by gather-
cess, including multiple revisions based on feedback ing data from interdisciplinary focus groups and an
206 RESEARCH ON SOCIAL WORK PRACTICE

exhaustive, systematic review of the literature (i.e., a Koeske, 1989, 1993). Previous structural equation mod-
dozen databases, including Medline, PsychInfo, Social eling analyses of the MBI have provided suggestions for
Services Abstracts, Sociological Abstracts, Health and improvement of the original measure, and several of these
Psychosocial Instruments, and others, were searched (Byrne, 1991; Yadama & Drake, 1995) have suggested
using multiple keywords, individually and in combina- the elimination of 2 items from the original 9-item scale.
tion). We developed a pool of items that operated the key We used this 7-item scale for our study, and it included
elements of role identity theory, including the profession- items such as, “I feel burned out from my work” and
als’ view of themselves as a helper or caregiver and their response options ranging from (0) never to (6) daily. The
perceptions of others’ view of them as a helper or care- score was the sum of the response options, and the mea-
giver. Potential items were collected and refined in the sure was reliable (a = .92).
iterative process previously described, including a pilot
test with a small convenience sample of social workers. Help Seeking
This process resulted in a scale containing 10 items
that we hypothesized would load onto the two conceptual We measured help seeking by asking respondents to
factors (self-expectations and the perception of others’ answer yes or no as to whether they had ever sought help
expectations for providing care or help) that make up role for personal issues other than addictive disorders.
identity. The scale included items such as the following:
“It is easier to care for others than to care for myself,” and Personal Characteristics
“It is my responsibility to be helpful to my family and
friends” and response options ranged from (1) strongly We hypothesized that personal characteristics men-
disagree to (5) strongly agree (see Table 1 for a complete tioned in theoretical and practice literature would be
list). Responses were summed to create a scale score. related to role identity and thus provide preliminary evi-
dence of convergent validity. These characteristics
Impairment included having a strong need for approval (Corey &
Corey, 1998), being a perfectionist (Kottler, 1993), feel-
We used a self-evaluative, general impairment ques- ing overly responsible for clients (Corey & Corey, 1998;
tion that has been used in other research (Pope et al., Kottler, 1993), and having difficulty asking for help. We
1987) for our measure of impairment. We asked respon- asked participants to describe themselves and respond
dents to respond on a scale of (1) strongly disagree to (5) from (1) strongly disagree to (5) to strongly agree.
strongly agree to the statement, “I have worked when I
was too distressed to be effective.” Demographics

Depressive Symptoms We also asked respondents to reply to standard demo-


graphic questions asking their ethnicity, birth year, mari-
Our measure of depressive symptoms, the Center for tal status, living situation, sexual identification, personal
Epidemiological Studies–Depression (CES-D) measure income, and whether they lived in rural areas, small
(Radloff, 1977), is a popular measure of depressive symp- towns, suburbs, or cities.
toms with well-researched properties that was used in a
similar study of nurses (Trinkoff, 1998). The 20-item
scale’s response options range from (0) rarely or never to DATA ANALYSIS STRATEGY
(4) most or all of the time and the scale score is the sum of
all responses. This measure was reliable with our sample We used Statistical Package for the Social Sciences
(a = .91). (SPSS) 10.0 for the exploratory analysis, and we allowed
missing values to remain missing because few such cases
Burnout existed (see Tables 1 through 4), and imputation did not
affect the results.
We used a revised version of the emotional exhaustion We compared the first 100 respondents with the last
subscale of the Maslach Burnout Inventory (MBI) as our 100 respondents on demographic characteristics and
burnout measure. Regarded as the heart of burnout, this variables of interest to explore response bias. This strat-
subscale is often used as the sole measure (Koeske & egy tested for effects of repeated mailings and was used to
Siebert / IDENTITY SCALE VALIDATION 207

TABLE 1: Descriptions of Items in Role Identity Scale (N = 376) FINDINGS


Variable M SD
Sample
I regularly help family members with 4.08 0.92
their problems and concerns The North Carolina Chapter of NASW sample was
Friends frequently turn to me when they 4.04 0.77
have problems or concerns
demographically representative of the national NASW
Work colleagues generally turn to me when 3.90 0.80 membership. It was 88% White, 8% African American,
they have problems or concerns and 3% other races and ethnicities. Respondents ranged
I have heard I am a natural helper or 4.03 0.87
caregiver
in age from 22 to 73 (M = 44 years) and were primarily
I began attending to others at an 3.82 1.05 female (84%) and heterosexual (94%). Most lived with a
early age spouse or partner (73%) and were married (66%). More
It is my responsibility to be helpful to 3.69 1.00 than half lived in suburbs or cities (58%), whereas the rest
my family and friends
It is easier to care for others than to 3.25 1.16 reported living in a small town or rural area (42%). Per-
care for myself sonal income averaged $40,000 per year. Although late
It is difficult to tell friends or family that 3.12 1.18 responders were slightly more likely to report working
I cannot help them with a problem
I would characterize myself as rescuer 2.56 1.14
when too distressed to be effective, early and late
Being a helper or caregiver is an 3.86 0.80 responders were demographically similar.
important part of who I am Descriptions were calculated for the role identity items
NOTE: Missing values range from 0 to 2. (see Table 1). Each item was significantly correlated at
Response options range from 1 (strongly disagree) to 5 (strongly agree).
greater than .35 with at least one other item.

Exploratory Analysis
estimate nonresponse bias (reported elsewhere) because
we were unable to gather data on nonresponders. The Initially, a principal components factor analysis was
strategy assumes that late responders are comparable to conducted. The Kaiser-Meyer-Olkin measure of sam-
nonresponders (Trinkoff & Storr, 1997) and that dispari- pling adequacy was excellent at .847. Bartlett’s Test of
ties between the early and late responders suggest poten- Sphericity, which tests whether the matrix is different
tial nonresponse bias. We used c2 tests for discrete cate- from the identity matrix, was 1,168 and significant at <
gorical variables (e.g., ethnicity, marital status, and .001 with 45 degrees of freedom. Two methods of evalua-
sexual identification) and ANOVAs for continuous vari- tion were used to estimate the number of potential fac-
ables (e.g., age, income, and years of social work experi- tors—checking the eigenvalue loadings > 1 and looking
ence). We also compared the sample demographic char- for breaks in the descending values, and checking the
acteristics with the national NASW membership to the Scree plots to identify the bend in the elbow. The inspec-
degree possible. tion of the loadings showed the largest values as 4.0 and
Next the sample was split, assigning cases to two 1.5, and a visual inspection of the Scree plot suggested
groups by randomly choosing a starting point in the sam- two factors as well. Additionally, when considering the
pling frame and then selecting every other case for mem- items conceptually, they were designed to fall into two
bership in Sample A, with the remaining cases assigned categories. Given these indications, principal axis factor
to Sample B. Sample A was used for an exploratory factor analyses next were conducted specifying two factors for
analysis and to test the scale’s psychometric properties. both orthogonal and oblique rotations.
Next, Sample B was used to conduct a confirmatory fac- Double loadings and small values were a problem with
tor analysis with structural equation modeling using the orthogonal rotation. The oblique rotation with two
Lisrel 8.54 to examine factorial validity. Finally, bivariate factors was clearly the best solution, and it makes concep-
associations among the revised role identity scale, other tual sense in that this rotation allows the factors to be cor-
personal characteristics, burnout, depression, impair- related. Conceptually, the first factor includes items that
ment, and help seeking were explored using all the cases describe the perception-of-self component of role iden-
to support the convergent and discriminant validity of the tity. The second factor includes the perception-of-how-
role identity measure. To adjust for multiple compari- others-see-me component of role identity. One item, “I
sons, a Bonferroni Holm correction was employed to began attending to the needs of others at an early age,”
adjust the level of significance used for each correlation had double loadings and conceptually was not a fit with
in the matrix. either of the two factors, and it was eliminated. The
208 RESEARCH ON SOCIAL WORK PRACTICE

TABLE 2: Results of a Two-Factor Oblique Factor Analysis of TABLE 3: Descriptions of Variables in the Exploratory
the Role Identity Scale Bivariate Analysis (N = 376)

Item Factor 1 Factor 2 n M SD

1. I characterize myself as a rescuer 0.780 Role identity scale 373 32.50 5.50
2. It is difficult to tell friend or family 0.777 Burnout 362 14.07 9.20
I cannot help Depression 366 28.57 7.93
3. I find it easier to care for others 0.691 Impairment 365 02.75 1.12
than myself Needs approval 370 03.75 0.85
4. It is my responsibility to be helpful to 0.464 Perfectionist 370 03.25 1.05
family and friends Difficulty asking for help 370 03.03 1.11
5. Being a helper or caregiver is who I am 0.457 Responsible for clients 370 03.58 0.93
6. Friends frequently turn to me when they 0.864
have problems Help seeking Frequency Percentage
7. Colleagues generally turn to me when 0.692
they have problems No 116 31.0
8. I have heard from others that I’m a 0.570 Yes 258 69.0
natural caregiver
NOTE: Missing values range from 0 to 14.
9. I regularly help family members 0.558
with problems
Variance explained (factors correlated
so no total variance) 2.65 2.53
possible violation of the assumptions of bivariate analy-
NOTE: Correlation = .47; Kaiser-Meyer-Olkin measure of sampling ac- sis. The statistics are presented in Table 3.
curacy = .830; Bartlett’s Test of Sphericity = 1,030, df =36, p < .001. Fac-
tor loadings < .25 are not reported. Content validity was established by a thorough review
of the scale items by experts in the substantive area and in
measurement, all of whom agreed that the items were
analysis was conducted again with this item removed, and grounded in the theoretical and conceptual literature.
the results are described in Table 2. Additionally, preliminary discriminant validity was
Next, an internal consistency reliability analysis was established, as the role identity scale score was not signif-
conducted on the role identity scale, and its alpha was icantly related to variables that conceptually should not
good at .81. The variable was normally distributed with a be related—for example, gender, sexual orientation, and
range of 15 to 45, a mean of 32.5, and a standard deviation place of residence.
of 5.5. Given these results, it was reasonable to include
this 9-item measure in the data analysis. A single scale Confirmatory Analysis
was retained for use in the data analyses because the the-
ory posits that both self-expectations and others’expecta- The predicted measurement model resulting from the
tions are integral to role identity, and the factor analysis exploratory factor analysis, with nine observed and two
supported the theoretical and conceptual coherence of the latent variables, was entered into LISREL 8.54. Using
measure. The two factors were correlated (r = .47, p < split Sample B, five cases were deleted using a case-wise
.00001). deletion method for missing data, adjusting the sample
Descriptive statistics were calculated for all the vari- size to 370. A case analysis of the data did not identify any
ables in the analysis, and the data were analyzed for problematic outliers or potential problems with serious

TABLE 4: Correlations Between Caregiver Role Identity Scale and Variables Hypothesized to be Related (N = 751)

1 2 3 4 5 6 7 8 9

1. Role identity
2. Burnout .18****
3. Depression .23**** .53****
4. Help Seeking¤ –.11* .06 .15***
5. Impairment .21**** .43**** .36**** .07
6. Needs approval .33**** .14** .22**** .13*** .13**
7. Perfectionist .24**** .19**** .16*** .03 .13** .31****
8. Difficulty asking for help .28**** .27**** .23**** .07 .22**** .23**** .33****
9. Responsible for clients .35**** .22**** .18**** –.04 .14** .34**** .26**** .20****
NOTE: Missing values = 6 to 34.
*p < .01. **p < .001. ***p < .0001. ****p <.00001.
Siebert / IDENTITY SCALE VALIDATION 209

Figure 1: Confirmatory Factor Analysis of the Caregiver Role Identity Scale (n = 370)

violations of assumptions of multivariate normality. The and a RMSEA value of .069. Additional fit index values
maximum likelihood method was used to estimate the included the Nonnormed Fit Index (NNFI = .95), the
level of data fit to the model. Consistent with the explor- Normed Fit Index (NFI = .95), the Comparative Fit Index
atory factor analysis, the latent constructs were allowed to (CFI = .97), the Goodness Of Fit Index (GFI = .97), and
correlate. The results included a c2 value of 90.61, df = 26 standardized root mean square residual (RMR = .058).
(p < .0001) that rejects the hypothesis that the model is These all indicated a good model fit that had potential for
correct, and a root mean square error of approximation additional improvement.
(RMSEA) of .082 that indicated a poor model fit to the The data suggested that correlating the error terms for
data. Item 2 (“Friends generally turn to me when they have
The data suggested a potential improvement in c2 if the problems”) and Item 3 (“Colleagues generally turn to me
error terms for three of the observed variables—Item 4 (“I when they have problems”) would improve the model fit.
have heard from others that I am a natural caregiver”), Similarly, the model fit would be improved by correlating
Item 6 (“It is my responsibility to be helpful to family and the error terms for Item 6 (“It is my responsibility to be
friends”), and Item 8 (“It is difficult to tell friends or fam- helpful to my family and friends”) and Item 1 (“I charac-
ily I cannot help”)—were correlated with the error term terize myself as a rescuer”). Given the potential overlap
for Item 10 (“Being a helper or caregiver is who I am”). between each of these sets of variables, it makes sound
This empirical evidence prompted a reconsideration of conceptual sense to correlate these pairs of error terms.
the conceptual and theoretical relationships among these Thus, the final model (see Figure 1) includes eight
variables. On reflection, Item 10 actually appeared to be a observed variables and two latent variables, correlating
proxy for the purpose of the scale—to determine level of the error terms of Item 2 to Item 3 and the error terms of
caregiver identity—rather than measuring an aspect of Item 6 to Item 1. This model resulted in a c2 value of
that identity, as is required for good scale items. As such, 25.40, df = 17 (p = .086), suggesting that the overall
it was inappropriate for inclusion with the other items, model is a good fit to the data. Additional fit indices all
and it made logical sense that the error term for Item 10 suggested a very good model fit: RMSEA = .037 and
was correlated with the error terms of the Items 4, 6, and RMR = .039 (below the .05 criteria); NFI = .98, NNFI =
8. The best decision in this circumstance was to remove .99, CFI = .99, and GFI = .98 (all above the .95 criteria);
Item 10 from the predicted measurement model, resulting and Expected Cross-Validation Index = .17 and Akaike
in an 8-item scale. Information Criterion = 63.40, both below their saturated
The maximum likelihood method was again used to values of .20 and 72.00, respectively. In addition, the larg-
estimate the level of data fit to the revised model. The est standardized covariance residual was improved from
results offered a c2 value of 52.44, df = 19 (p = .00006) 4.71 for the original model to 3.56 for the final model.
210 RESEARCH ON SOCIAL WORK PRACTICE

Thus, the adjustments for the original model resulted in prioritized by salience and the level of support received
an even better model fit. for each specific role. Social workers formulate identities
Finally, because no other measures of caregiver role of themselves as caregivers, often idealized, that develop
identity exist for comparison purposes, convergent valid- through an iterative process from their assessments of (a)
ity was examined by conducting a correlation analysis of their own performance as a personal and professional
all the cases that included the revised 8-item role identity caregiver and (b) others’ expectations of them in that role
scale and variables that were hypothesized to be related. (McCall & Simmons, 1978). The exploratory factor
The helper or caregiver role identity was associated with analysis of the measure clearly supports this notion
all potentially related variables using the Bonferroni because the factors fell cleanly along these two facets of
Holm correction for multiple comparisons (see Table 4). role identity—how the respondents saw themselves and
Another reliability analysis of the revised 8-item scale how the respondents perceived others’ perceptions of
was also conducted, and its reliability remained good, them. The confirmatory analysis empirically supported
although it dropped from .81 for the 9-item scale to .78 for the factor structure as well.
the 8-item scale. The scale exhibited additional sound psychometric
properties—reliability, face validity, and discriminant
validity. Of special interest are the findings of the conver-
DISCUSSION AND APPLICATION TO gent validity analysis because they support not only the
SOCIAL WORK RESEARCH AND PRACTICE measure but also its theoretical application in that respon-
dents scoring higher on the identity measure also reported
It is useful to consider certain conceptual and method- higher levels of depressive symptoms, burnout, and pro-
ological limitations of the study before discussing the fessional impairment. It makes intuitive sense that high
findings. Role identity theory is limited in that it does not scores on being perfectionists, needing approval, and
offer precise, easily measurable definitions of the con- feeling highly responsible for clients leave respondents
cepts and thus has not been frequently operationalized for vulnerable to burnout, depression, and impairment, but
research. The study achieved an impressive 75% usable these qualities were also related to higher scores on care-
response rate that resulted in a representative sample, but giver role identity. Of practical importance is that those
because the sample was limited to members of the North scoring higher on caregiver identity not only reported
Carolina Chapter of NASW, the findings cannot be gener- having greater difficulty asking for help, but they also
alized to all social workers. Furthermore, all the measures reported actually seeking help less frequently than those
were self-reported, and the measures used to examine with lower role identity scores.
convergent validity (i.e., depressive symptoms, burnout, Following the theory, it is likely that perfectionistic,
help seeking, impairment) asked for sensitive informa- responsible social workers have high expectations for
tion, so social desirability bias is possible, if not probable. their role performance, perceiving themselves as helpers,
Although the response rate was excellent and there was caregivers, and problem solvers, both personally and pro-
very little item nonresponse, it is possible that those who fessionally. Despite this idealized image (and, in part,
did not respond were somehow different from those who because of it), social workers are vulnerable to their own
did. distress, as evidenced by the findings of this and other
It also should be noted that although the Caregiver analyses (Siebert, 2003, 2004, in press-b). When this is
Role Identity Scale scores were normally distributed, the coupled with a reluctance to seek help (Siebert, in press-a),
disstribution had a slight negative skew that did not it is not surprising that social workers’ professional prac-
unduly influence the results. Furthermore, the correla- tice is negatively influenced, creating situations in which
tions with the scale score were generally moderate, as clients might receive less than effective services.
would be expected, as the variables in the analysis have a These findings have clear implications for the preven-
complex etiology (e.g., depression may have a large bio- tion of distress and impairment among social workers.
logical component, and much of burnout’s variance is The literature suggests that a large proportion of social
likely explained by working conditions) that would vary workers and other caregiving professionals are attracted
among respondents. to the profession because they occupied caretaking roles
Considering these potential limitations, the findings of in their families of origin (DiCaccavo, 2002; O’Conner,
the present study are quite promising and support role 2001; Vincent, 1996). These professionals and others
identity theory in a number of ways. Role identity theory with high scores on the caregiver role identity scale would
posits that individuals occupy an assortment of roles, benefit from early prevention intervention—ideally in
Siebert / IDENTITY SCALE VALIDATION 211

their early training in schools of social work—before the Deutsch, C. J. (1985). A survey of therapists’ personal problems and
rigors and stress of social work practice contribute to the treatment. Professional Psychology: Research and Practice, 16,
305-315.
depression or burnout that can result in impaired practice DiCaccavo, A. (2002). Investigating individuals’ motivations to
(Siebert, 2001, 2004, in press-b). Identifying social work become counseling psychologists: The influence of early
students with prominent caregiver role identities and pre- caretaking roles within the family. Psychology and Psychotherapy:
senting curriculum on self-awareness and self-care, with Theory, Research, and Practice, 75, 463-472.
a focus on personal distress and professional impairment, Dillman, D. A. (2000). Mail and internet surveys: The tailored design
method. New York: John Wiley.
might make a real difference to well-being of vulnerable
Doyle, B. B. (1987). The impaired psychiatrist. Psychiatric Annals,
professionals and their clients. Normalizing the experi- 17, 760-763.
ence of burnout or depression, for example, and the need Duquette, A., Kerouac, S., Sandhu, B. K., & Beaudet, L. (1994). Fac-
to seek help for oneself could be extraordinarily helpful tors related to nursing burnout: A review of empirical knowledge.
for professionals who might otherwise tend to deny that Issues in Mental Health Nursing, 15, 337-358.
they need assistance. Course content needs to be Felton, J. S. (1998). Burnout as a clinical entity—Its importance in
health care workers. Occupational Medicine, 48, 237-250.
developed and tested empirically with these issues in Good, G. E., Thoreson, R. W., & Shaughnessy, P. (1995). Substance
mind. use, confrontation of impaired colleagues, and psychological func-
An appreciation for the importance of caregiver role tioning among counseling psychologists: A national survey. The
identity could also assist treatment providers in the for- Counseling Psychologist, 23, 703-721.
mulation of assessments and interventions for distressed Hood, J. C., & Duphorne, P. L. (1995). To report or not to report:
Nurses’ attitudes toward reporting coworkers suspected of sub-
and impaired social workers once they have been identi-
stance abuse. Journal of Drug Issues, 25, 313-339.
fied. For professionals in need of treatment who score Hughes, P. H., Brandenberg, N., Baldwin, D. C., Storr, C. L., Williams,
high on caregiver role identity, behavior change could be K. M., Anthony, J. C., et al. (1992). Prevalence of substance abuse
targeted through the lens of idealized self-expectations among US physicians. Journal of American Medical Association,
and perceptions, improving the likelihood of effective 267, 2333-2340.
intervention with those professionals who are uncomfort- Koeske, G. F., & Koeske, R. D. (1989). Work load and burnout: Can
social support and perceived accomplishment help? Social Work,
able seeking help for themselves. 34, 243-248.
Although recent research has confirmed that profes- Koeske, G. F., & Koeske, R. D. (1993). A preliminary test of a
sional impairment is an issue for the social work disci- stress-strain-outcome model for reconceptualizing the burnout
pline, much more research is required to develop effective phenomenon. Journal of Social Services Research, 17, 107-135.
early identification, prevention, assessment, and inter- Kottler, J., & Hazler, R. J. (1996). Impaired counselors: The dark side
brought into light. Journal of Humanistic Education and Develop-
vention strategies that will target the idiosyncratic needs
ment, 34, 98-107.
of social work professionals. Conducting this research Kottler, J. A. (1993). On being a therapist. San Francisco:
and developing measures like the caregiver role identity Jossey-Bass.
scale will begin to address a concern that the social work McAuliffe, W. E., Rohman, M., Breer, P., Witshak, G., Santangelo, S.,
profession has long neglected—the health and well-being & Magnuson, E. (1991). Alcohol use and abuse in random samples
of our professionals and, by extension, the well-being of of physicians and medical students. American Journal of Public
Health, 81, 177-182.
our clients. McCall, G. J., & Simmons, J. L. (1978). The role-identity model. In
Identities and interactions: An examination of human associations
in everyday life (pp. 1-288). New York: Free Press.
REFERENCES Nace, E. P. (1995). The professional paradox. In E. P. Nace (Ed.),
Achievement and addiction: A guide to the treatment of profession-
als (pp. 1-7). New York: Brunner/Mazel.
Byrne, B. (1991). The Maslach Burnout Inventory: Validating facto-
National Association of Social Workers. (1996). Code of ethics. Wash-
rial structure and invariance across intermediate, secondary, and
ington, DC: Author.
university educators. Multivariate Behavioral Research, 26,
O’Connor, M. (2001). On the etiology and effective management of
583-605.
professional distress. Professional Psychology: Research and
Centrella, M. (1994). Physician addiction and impairment-current
Practice, 32, 345-350.
thinking: A review. Journal of Addictive Diseases, 13, 91-105.
Pope, K. S., Tabachnick, B. G., & Keith-Spiegel, P. (1987). Ethics of
Corey, M. S., & Corey, G. (1998). Becoming a helper. Pacific Grove,
practice: The beliefs and behaviors of psychologists as therapists.
CA: Brooks/Cole.
American Psychologist, 42, 993-1006.
Deutsch, C. J. (1984). Self-reported sources of stress among psycho-
Radloff, L. S. (1977). The CES-D scale: A self-report depression scale
therapists. Professional Psychology: Research and Practice, 15,
for research in the population. Applied Psychological Measure-
833-884.
ment, 1, 385-401.
212 RESEARCH ON SOCIAL WORK PRACTICE

Sherman, M. D., & Thelen, M. H. (1998). Distress and professional Trinkoff, A. M., & Storr, C. (1997). Collecting substance use data with
impairment among psychologists in professional practice. Profes- an anonymous mailed survey. Drug and Alcohol Dependence, 48,
sional Psychology: Research and Practice, 29, 79-85. 1-8.
Siebert, D. C. (2001). Work and well being: A survey of distress and Trinkoff, A. M., Storr, C. L., & Wall, M. P. (1999). Prescription-type
impairment among North Carolina social workers. Unpublished drug misuse and workplace access among nurses. Journal of
doctoral dissertation, University of North Carolina. Addictive Diseases, 18, 9-17.
Siebert, D. C. (2003). Denial of AOD use: An issue for social workers Trinkoff, A. M., Zhou, Q., Storr, C. L., & Soeken, K., L. (2000). Work-
and the profession. Health and Social Work, 28, 89-97. place access, negative proscriptions, job strain, and substance use
Siebert, D. C. (2004). Depression in North Carolina social workers: in registered nurses. Nursing Research, 49(2), 83-90.
Implications for practice and research. Social Work Research, 28, Vaillant, G. E. (1972). Some psychologic vulnerabilities of physi-
30-40. cians. New England Journal of Medicine, 28, 372-375.
Siebert, D. C. (in press-a). Help seeking and AOD use among social Vaillant, G. E. (1996). A long term follow-up of male alcohol abuse.
workers: Patterns, barriers, and implications. Social Work. Archives of General Psychiatry, 53, 243-249.
Siebert, D. C. (in press-b). Personal and occupational factors in burn- Vincent, J. (1996). Why ever do we do it? Unconscious motivation in
out among practicing social workers: Implications for researchers, choosing social work as a career. Journal of Social Work Practice,
practitioners, and managers. Journal of Social Services Research. 19, 63-69.
Thoreson, R. W., Nathan, P. E., Skorina, J. K., & Kilburg, R. R. (1983). Wood, B. J., Klein, S., Cross, H. J., Lammers, C. J., & Elliott, J. K.
The alcoholic psychologist: Issues, problems, and implications for (1985). Impaired practitioners: Psychologists’ opinions about
the profession. Professional Psychology: Research and Practice, prevalence, and proposals for intervention. Professional Psychol-
14, 670-684. ogy: Research and Practice, 16, 843-850.
Trinkoff, A. M. (1998). Substance use among nurses: Differences Yadama, G. N., & Drake, B. (1995). Confirmatory factor analysis of
between specialties. American Journal of Public Health, 88, the Maslach Burnout Inventory. Social Work Research, 19,
581-585. 184-192.
Trinkoff, A. F., Eaton, W. W., & Anthony, J. C. (1994). The prevalence
of substance abuse among registered nurses. Nursing Research, 40,
172-175.

View publication stats

Potrebbero piacerti anche