Children and Youth Services Review 32 (2010) 840–846

Contents lists available at ScienceDirect

Children and Youth Services Review

j ourna l homepage: www.e lsev ie r.com/ locate /ch i ldyouth

Commitment to child welfare work: What predicts leaving and staying?

Kathleen Coulborn Faller ⁎, Marguerite Grabarek, Robert M. OrtegaUniversity of Michigan School of Social Work, 1080 S. University Ave., Ann Arbor, MI 48109-1106, United States

⁎ Corresponding author. Tel.: +1 734 998 9700; fax:E-mail addresses: kcfaller@umich.edu (K.C. Faller), g

(M. Grabarek), rmortega@umich.edu (R.M. Ortega).

0190-7409/$ – see front matter © 2010 Elsevier Ltd. Aldoi:10.1016/j.childyouth.2010.02.003

a b s t r a c t

a r t i c l e i n f o

Article history:Received 7 November 2009Received in revised form 3 February 2010Accepted 5 February 2010Available online 11 February 2010

Keywords:Child welfare workforceTurnover

This study describes results related to worker turnover from a longitudinal study of public and privateagency child welfare workers in one state. Findings from 460 new workers were examined for reasonsrespondents took their jobs and chose child welfare work, their commitment to their agencies and to childwelfare for two and five years, and the relationship of these variables and demographic variables to whetherthey were still in their positions at follow-up. Among the findings were that public agency workers endorsedsignificantly higher levels of commitment on three of the four commitment variables and were significantlymore likely to have taken their jobs because of good pay, benefits, and advancement opportunities thanprivate agency workers. In contrast, private agency workers endorsed taking the job because it was the onlyjob available and it was a good first job to take at significantly higher rates than public agency workers.Workers of color endorsed lower levels of commitment on three of the four of commitment variables,although race was not a significant predictor of actually having left the job. Variables that predicted stayingon the job were having viewed the state's Realistic Job Preview before taking the job, good supervision, andhigher job satisfaction.

+1 734 998 9710.rabarek@umich.edu

l rights reserved.

© 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Turnover in the child welfare workforce is a problem that hasplagued the child welfare field for at least four decades (e.g., Alwon &Reitz, 2001; Flower, McDonald, & Sumski, 2005; Strolin et al., 2006).Estimates are that approximately a fifth of frontline workers in publicchild welfare leave per year and in the voluntary sector about twicethat rate. In a national study conducted by Alliance for Children andFamilies, American Public Human Services Association, and ChildWelfare League of America of the child welfare workforce, on average20% of public child welfare workers and 8% of public child welfareagency supervisors turned over annually, and an astounding 40% ofworkers and 28% of supervisors in the private sector do so (Alliance forChildren and Families, American Public Human Services Association, &Child Welfare League of America, May, 2001). Similarly, Alwon andReitz (2001) report 15% of public agency staff and 27% in the voluntarysector leave every year. Finally, in 42 states participating in a recentstudy by the American Public Human Services Association (2005),22% of child protective services workers left per year, compared to 9%of state and local government workers. On average child welfareworkers in the United States last two years on the job (NASW, 2003;Strolin et al., 2006).

High turnover in staff has many detrimental consequences forthe quality of child welfare work. First, turnover results in frontline

workers with minimal experience making life and death decisionsabout child safety (Child Welfare League of America, 2002; Gonzalez,Faller, Ortega, & Tropman, 2009). Second, high turnover negativelyimpacts child well-being and permanency (Child Welfare Leagueof America, 2001, 2002, 2009; Flower et al., 2005; GAO, 2004).Third, high rates of turnovermean that there are uncovered caseloads,thereby increasing the burden on the remaining workers (Alwon &Reitz, 2001). Fourth, work overload results in lower quality of service(Child Welfare league of America;GAO, 2003). Finally, both the loss ofcolleagues and the increased workload burden have a negative impacton worker morale (Alwon & Reitz, 2001; Graef & Potter, 2002).

Childwelfareworker turnover is also costly. Training a newworkercosts on average one-third to one-half of a worker's annual salary(Ellett, Ellett, & Lerner, 2008; Graef & Hill, 2000; Graef & Potter, 2002;Landsman, 2008). Childwelfare agencies never have sufficient funds toprovide the needed services. Preventing unnecessary turnover couldhave positive impacts on child welfare budgets, as well as on childwelfare worker morale and quality of child welfare service delivery,thereby improving outcomes for children in the child welfare systemand their families (Flower, McDonald, & Sumski; GAO, 2003).

The U.S. General Accounting Office published two recent reports(GAO, 2003, 2006) highlighting the serious workforce problems inchild welfare. The GAO called upon the U.S. Department of Health andHuman Services (DHHS) to take a leadership role in addressing thechild welfare workforce crisis (GAO, 2003). One response by theChildren's Bureau within DHHS to this crisis was to make five yeargrants to eight schools of social work, who partnered with state childwelfare agencies to address recruitment and retention of child welfare

1 Only public child welfare applicants saw the Realistic Job Preview. Public childwelfareworkers hired after July, 2005, were sent a DVD of the Realistic Job Preview afterthey had made written application for a position. They were required to review andreturn the DVD, indicating their continuing desire for the position, before they receiveda behaviorally based interview. Private agency applicants did not review the RJP.

2 The response rate was 95.05%.3 We excluded workers who were “lateral transfers,” that is, workers who were

transferring between positions in their agency (e.g., from protective services to fostercare).

4 Although types of job (Protective Services or Foster Care) were entered into theregressions, it did not predict any differences on dependent variables.

841K.C. Faller et al. / Children and Youth Services Review 32 (2010) 840–846

workers (Child Welfare Staff Recruitment and Retention TrainingDiscretionary Grant Cluster, 2003–2008). Focused on training childwelfare supervisors in recruitment and retention strategies, most ofthese programs have also engaged in additional important endeavorsrelated to the child welfare workforce.

One such endeavor was developing Realistic Job Previews (RJPs)(Faller et al., 2009). Indeed, RJPs were one of the innovationsrecommended by the GAO (2003). Derived from the business field andthemilitary (Wanous, 1998, 1992), Realistic Job Previews aim to providethe job applicant with an accurate, but balanced picture of the positionhe/she is seeking, to improve prospective employee decision-makingabout whether or not to take a job, to reduce turnover that is caused bya failure to understand the job — especially the difficult aspects, andto improve job satisfaction because the job-seeker is not unpleasantlysurprisedbycomponents of the job (CPSHumanResources, Champnoise,& Masternak, 2004; Graef, 2005). Using Recruitment and RetentionGrant funding, four of the eight social work school grantees facilitatedthe development of RJPs, and one school gathered data to evaluate itsstate's already developed RJP (Faller et al., 2009).

Another key factor that affects child welfare turnover is commit-ment, both to the field of child welfare and to the organization. Theorganizational literature on commitment to work is relevant to currentefforts to understandworkforce turnover in childwelfare. A reduction incommitment to the organization is seen as resulting in loss of workerproductivity, especially if workers are driven by a perception of unfairpractices or problematic organizational behaviors, but lack of commit-ment also results in increased absenteeism and turnover (Knudson,Johnson, Roman & Martin 2003; Lambert, Pasukpuleti, Cluse-Tolar,Jennings, & Baker, 2006; Meyer & Allen, 1991; Akdogan & Cingoz 2009).

Commitment specifically in child welfare has also been examined(e.g., Landsman, 2001; Westbrook, Ellis, & Ellett, 2006). In a study ofpublic child welfare workers in Missouri, Landsman (2001) differen-tiated two types of commitment, commitment to the organization andto the occupation. Landsman'smodel demonstrates the relationship ofdifferent but overlapping factors to these two dependent variables.The distinction between organizational and occupational commit-ment is useful because there are a variety of child welfare jobs, withdiffering responsibilities and stresses (e.g., protective services, fostercare, and adoption).

This distinction is also useful, however, because there are agencyproblems, not directly related to child welfare work that can determinewhether workers are committed and intend to leave or stay. Agency-specific problems include poor supervision (e.g. GAO, 2003; Landsman,2007), high caseloads, highworkloads (ChildWelfare LeagueofAmerica,2002; Landsman, 2001), bureaucratic and punitive agency practices(Gonzalez et. al., 2009), and lack of resources to do the job. These workconditions, which may induce turnover, are different from thosewhich are endemic to child welfare work (e.g., Gonzalez et al., 2009).Child welfare-specific problems include constantly encountering childmaltreatment and occasionally encountering child death, experiencingpotential physical dangers from irate parents or the environmentsclients live in (ChildWelfare League of America, 2002;GAO, 2003, 2006),having to make potential life and death decisions about children, andbeing overwhelmed by the complexmix of family difficulties (e.g., GAO,2003, 2006; NASW, 2003).

In this article, we describe findings from a Longitudinal Study ofthe Child Welfare workforce in one state. We build upon the workof Landsman (2001) and Ellett and colleagues (e.g., Ellett, Ellett, &Rugutt, 2003) and explore new worker commitment to the profes-sional field of child welfare as well as commitment to a specificagency. We ask workers to indicate feasible timeframes for staying onthe job, rather than life commitment (e.g., Landsman, 2001, 2007,2008). In addition, unlike most studies of turnover, we examinewhether workers actually left, rather than intent to leave (e.g., Ellett,Ellett & Rugutt, 2003; Landsman, 2001, 2007, 2008; Strolin et al.,2006).We identify a range of factors that affect whether child welfare

workers leave or stay. These factors include reasons for taking the jobin the first place, demographic characteristics, exposure to a RealisticJob Preview before being interviewed for the job (CPS HumanResources, Champnoise, & Masternak, 2004; Graef, 2005; CPSHuman Resources & Champnoise, 2007; Faller et al., 2009),1 qualityof supervision (e.g., Collins-Camargo. & Kelly, 2007; Mor-Barak,Travis, Pyun, & Xie, 2009; Lietz, 2008; Landsman, 2007), and jobsatisfaction (e.g., Child Welfare League of America, 2002; Ellett et al.,2003, 2008; Landsman, 2001).

2. Study methods

The Longitudinal Study of ChildWelfare employees was conductedin one mid-western state. The study questionnaire was designedto gather data from respondents at four points in time — baseline,6-months, 12-months and 18-months. Specific topics covered in thequestionnaire variedwithdata collectionpoint. Respondents completedbaseline questionnaires at the end of their eight week new workertraining2. At the point of initial data collection (baseline), writtenconsent was obtained from study participants to contact them againin order to collect the follow-up data. The study was approved bythe relevant University of Michigan Institutional Review Board.

The questionnaire content was designed to explore in depth anarray of personal, interpersonal, and organizational factors whichhave been identified in the literature as being important predictors ofjob satisfaction and retention both in child welfare and the humanservices generally. We further refined the domains of inquiry usinginformation gathered from 13 focus groups conducted with childwelfare workers, supervisors, and administrators in the state. Most ofthe scales and items used in this study have been employed in priorresearch by numerous investigators.

2.1. Participants

Minimum requirements for childwelfare workers are that they holda bachelor's degree in a human service related field. There is a multi-stage, central hiring process for public child welfare workers. Privateagency workers are recruited by the individual agencies, often throughannouncements posted on a private agency federation website. Thescreening and selection process for private agencies varied by agency.All workers, both public and private, are required to attend newworkertraining sponsored by the public child welfare agency. BetweenNovember 2004 and April 2007, 651 child welfare workers in thestate completed baseline questionnaires. For this article, only workerswhowere new to their child welfare agencies (N=460)were included,327 public agencyworkers, and134private agencyworkers. Theprivateagencies were under contract to the public child welfare agency3.Demographic characteristics of the participants are found in Table 1.

As Table 1 indicates, like most child welfare workforce samples(CWLA, 2009), participants in this study are overwhelmingly female,and about 70% have only a bachelor's level education. However theproportion of participants who are workers of color is higher thantypically found in the child welfare workforce (35.4%) (NASW, 2003).Workers were about evenly split between child protection work andfoster care work4.

Table 1Demographic characteristics of new workers; public and private agency N=4601.

Characteristic Categories Number Percent

Gender Female 378 82.2Male 76 16.5

Race White 275 59.8Worker of color 163 35.4

Education Bachelors degree 320 70.5Master's degree 133 18.9

Job description Protective services 206 44.8Foster care 217 47.2

1Numbers do not sum to 460 because there are some missing data.

Table 2bReasons for working in child welfare.

1) Always wanted to help children2) Always wanted to help families3) My personal experiences4) I feel working with children is my calling5) Not satisfied with my initial career choice6) It was the only job available7) I was encouraged by others8) I wanted to improve the quality of child welfare services9) Child welfare work is consistent with my training10) Other ______________________________

842 K.C. Faller et al. / Children and Youth Services Review 32 (2010) 840–846

2.2. Procedure

Participants in the study were asked to indicate the top threereasons they chose to work in their current child welfare job and thetop three reasons they chose to work in child welfare. Table 2a and 2blists the options.

Lists of reasons for workers taking their current jobs and forworking in child welfare were derived from content analysis ofaudiotapes from the 13 focus groups conducted with child welfareworkers, supervisors, and administrators before the questionnairewas completed. In order to collapse the list of variables, factor analysiswas utilized to further categorize the responses. These categories arefound in Tables 4 and 5.

Second, the levels of commitment of these new child welfareworkers to both their current positions and to child welfare workwere ascertained. To determine the degree of commitment to theirrespective agencies and their commitment to work in child welfare atthe time they started work as protective services and foster careworkers, two questions were asked, prefaced by the followingstatement: On the scale below, check the box which best describesyour degree of commitment for remaining in ______ for at least 2-years;the second question, with same preface, asked about a commitmentfor 5-years. The questionnaires distributed to DHS workers had “DHS”in the blank space and those distributed to private agencyworkers had“your agency.” These questions were followed by two additionalquestions with the same preface, but referenced the degree ofcommitment to staying in childwelfare for 2- and5-years respectively.The scale for all four questions ranged from “Not committed at all”(1) to “Highly committed (7).

Lastly, until a cut-off date of January 2009, we tracked workers todetermine whether they were still in their jobs. Multiple means wereused to determine their employment status. In some cases, respondentsvoluntarily informed the study that they had left their agencies. Inaddition, DHS provided monthly updates on worker status for publicemployees. There is no centralized data source for private agencyworkers; therefore, it was not possible to determine the private

Table 2aReasons for taking the child welfare job.

1) Agency offers good job security2) Agency offers good pay3) Agency offers good benefits4) Always wanted to work for this agency5) It was the only job available6) My personal / family situation does not allow me to move7) I have heard good things about the work of this agency8) Gives me an opportunity to help children9) Gives me an opportunity to help families10) Agency offers good opportunities for advancement11) Agency offers job variety, e.g. protective services, foster home licensing, etc.12) Current job is a stepping stone to go work for DHS

workers' status in the same way as public agency workers. However,attempts were made to contact participants who failed to return followup surveys, using contact information provided at baseline. If thesecontact efforts were unsuccessful, the agency number was called.Overall, the exit status for public agencyworkerswasmore reliable thanfor private agency workers.

3. Results

As of January 2009, 78 of the 460 new workers (18.5%) (data weremissing for 38 participants in the study) had left their positions and344; (81.5%)were still on the job. Of thoseworkers who had departed,58 (17.7% of new public agency workers in the study) were publicagency employees, and 20 (21.1%of the private agency workers in thestudy) were private agency employees (N.S.).

Table 3 provides the bivariate findings on the relationship betweenwhether the worker was still at the agency and commitment to theagency for two and five years and commitment to child welfare fortwo and five years.

As Table 3 demonstrates, there was a statistically significantrelationship between leaving/staying by January, 2009, and levelof commitment on three of the four measures (commitment to theagency for two and five years and to child welfare for two years), andthe relationship approaches significance on the fourth (commitmentto child welfare for five years). In all four instances, higher levelsof commitment at baseline are found in those who were still on thejob.

Table 4 presents the bivariate (newDHS and private agency) tableson collapsed reasons for taking the job.

As Table 4 indicates, with the exception of the position providesan opportunity to help children and families, there were significantdifferences in the responses endorsed by public and private agencyworkers. Most importantly, private agency workers were significantlymore likely to have taken the job because it was the only job available,whereas public agency workers were significantly more likely to havetaken the job because of pay, benefits, job security, opportunities, andvariety. Significantly fewer public agency workers endorsed, as areason, that they heard good things about the agency, although thisreason was endorsed by a small minority of both types of workers.Finally, significantly higher proportions of private agency workersendorsed a good first job to take, although again a minority of bothgroups endorsed this reason.

Table 5 provides the results for reasons public and private agencyworkers chose child welfare as a type of work.

As Table 5 indicates, in terms of choosing to work in child welfare,more than 80% of both public and private agencyworkers selected as areason, to help children and families. A significantly higher percentageof private agency workers, but a minority, chose child welfare workbecause itwas the only position available; in contrast, a smallminority,but a significantly larger proportion of public than private agencyemployees selected child welfare because they were dissatisfiedwith their previous employment.

Table 4Reason for taking job (collapsed categories) by DHS versus private agency worker.

Worker type Total mentions

DHS(N=325)

Private(N=133)

N=458

Pay/benefits (sig.=.000) 73.5% 20.3% 266Help children/families (sig.=.119) 54.5% 62.4% 260Only job available/only family choice(sig.=.000)

8.3% 54.1% 99

Job security/opportunities/variety(sig.=.000)

61.5% 26.3% 235

Heard good things/always wanted towork @ agency (sig.=.008)

12.6% 22.6% 71

Good to take as 1st job (sig.=.001) 9.7% 22.6% 60

Table 3Commitment by exit status: new DHS and private agency workers.

Stayed on the job/left job Commitment to agencyfor at least 2 years(P=.000)

Commitment to agencyfor next 5 years(P=.000)

Commitment to child welfarework for at least 2 years(P=.027)

Commitment to child welfarework for next 5 years(P=.065)

Still at agency Mean 5.60 4.92 5.79 5.26N 321 308 325 311Std. dev. 1.614 2.039 1.399 1.799

Left agency Mean 4.71 3.92 5.38 4.81N 70 66 74 69Std. dev. 1.905 2.129 1.611 1.849

Total Mean 5.44 4.74 5.71 5.18N 391 374 399 380Std. dev. 1.701 2.086 1.447 1.814

Table 6Commitment to remaining at agency for next two years.

N=387 Unstandardizedcoefficients

Standardizedcoefficients

Adj. R2=.395 B Std. error Beta t Sig.

(Constant) 4.755 .479 9.918 .000Gender: male = 1 −.012 .195 −.002 −.060 .952Race: worker of color = 1 −.297 .152 −.079 −1.957 .051Public/private: public = 1 1.195 .200 .304 5.977 .000Education −.115 .163 −.029 −.709 .479Help children/families .249 .157 .069 1.579 .115Only job available/onlyfamily choice

−1.322 .238 −.295 −5.564 .000

Job security/opportunities/variety

.448 .172 .124 2.606 .010

Heard good things/alwayswanted to work @ agency

.342 .203 .072 1.687 .093

Good to take as 1st job −.714 .231 −.132 −3.097 .002

Table 7Commitment to remaining at agency for next five years.

N=373 Unstandardizedcoefficients

Standardizedcoefficients

Adj. R2=.432 B Std.error

Beta t Sig.

843K.C. Faller et al. / Children and Youth Services Review 32 (2010) 840–846

Multiple regressions were undertaken to examine variables thatpredicted commitment to the current job and to child welfare fortwo and five years. The models included new worker demographiccharacteristics and reasons for taking the job.

As Table 6 indicates, five variables were predictive of workers'commitment to remaining at the agency for two years. Higher levelsof commitment were positively associated with workers who took thejob because of job security and opportunities andwhowere employedin the public sector. Being a worker of color, taking the job because itwas the only one available, and choosing it because it was a good firstjob were predictive of lower levels of commitment. These variablesaccount for 40% of the variance in the model.

Table 7 is the regression for level of commitment to remain in thecurrent position for five years.

As Table 7 indicates, the same five variables were significantpredictors for commitment to the job for five years. Working for apublic child welfare agency and taking the job for security, opportu-nities, or varietywere predictive of higher levels of commitment. Beinga worker of color, taking the job because it was the only one available,and choosing it because it is a good first job to take were associated

Table 5Reason for choosing child welfare (collapsed categories) by DHS versus private agency.

Worker type Total mentions

DHS(N=325)

Private(N=133)

N=458

Help children/families/calling/improveCW services (sig.=.856)

81.2% 82.0% 373

Only job available (sig.=.004) 11.4% 21.8% 66Consistent w/ training/experience(sig.=.382)

44.3% 39.8% 197

Dissatisfied w/ previous job (sig.=.027) 7.7% 2.3% 28

with lower levels of commitment. These variables account for 43% ofthe variance.

Table 8 provides information about predictors of commitment tochild welfare for two years.

As Table 8 indicates, four variables were significant predictors ofcommitment to child welfare for at least two years. Race, which wasa significant predictor of commitment to the job/position for twoand five years, is also a significant predictor of commitment to childwelfare for two years; again being a worker of color was associated

1 (Constant) 4.431 .564 7.862 .000Gender: male = 1 −.020 .232 −.003 −.084 .933Race: worker ofcolor = 1

−.525 .180 −.117 −2.926 .004

Public/private:public = 1

1.746 .239 .369 7.306 .000

Education −.404 .188 −.086 −2.151 .032Help children/families/calling/ improve services

.167 .186 .038 .897 .370

Only job available/onlyfamily choice

−1.350 .278 −.257 −4.856 .000

Job security/opportunities/variety

.447 .204 .103 2.188 .029

Heard good things/wanted to work @agency

.355 .236 .062 1.506 .133

Good to take as 1st job −.987 .289 −.143 −3.413 .001

Table 8Commitment to child welfare for the next two years.

N=410 Unstandardizedcoefficients

Standardizedcoefficients

Adj. R2=.220 B Std. error Beta T Sig.

(Constant) 4.874 .412 11.822 .000Gender: male = 1 .015 .177 .004 .084 .933Race: worker of color = 1 −.350 .139 −.112 −2.523 .012Public/private: public = 1 1.031 .148 .311 6.954 .000Degrees collapsed −.176 .147 −.053 −1.198 .232Help children/families/calling/improve services

.785 .188 .197 4.174 .000

Only job available −.859 .207 −.197 −4.142 .000

Table 10Predictors of still being at the agency at follow-up in January 2009.

N=173 Unstandardizedcoefficients

Standardizedcoefficients

Adj. R2=.151 B Std.error

Beta t Sig.

(Constant) 2.106 .278 7.565 .000Gender: male = 1 .064 .068 .068 .943 .347Race: worker of color = 1 .088 .056 .113 1.584 .115Public/private: public = 1 .023 .125 .014 .182 .856Degrees collapsed: BA,BSW/MS, MSW

−.088 .061 −.106 −1.424 .156

Saw RJP video .164 .057 .215 2.851 .005Supervisor gives usefulinformation whenneeded: 1 = not at alltrue; 4 = very true

−.032 .041 −.084 −.782 .435

Supervisor helps with anew or unfamiliar task:1 = not at all true;4= very true

−.056 .039 −.148 −1.440 .152

Supervisor makes your lifedifficult 1 = not at alltrue; 4 = very true

−.094 .035 −.238 −2.644 .009

Job satisfaction @ 6 months:1 = low 4 = high

.091 .034 .220 2.722 .007

844 K.C. Faller et al. / Children and Youth Services Review 32 (2010) 840–846

with lower levels of commitment. Working for a public child welfareagency predicted a higher level of commitment. Reasons for choosingchild welfare were also significant. A chance to help children/families/improve services/the worker's calling, a combined variable fromfour reasons for taking the job, is a predictor of higher levels ofcommitment to child welfare for two years. Finally, that it was theonly job available continues to be a predictor of lower levels ofcommitment. Together, these four variables account for 22% if thevariance in level of commitment.

Table 9 provides the model for predicting commitment to childwelfare for five years.

As Table 9 indicates, being a public agency employee continues tobe significant in predicting higher levels of commitment to childwelfare for five years and only job available a significant predictorof lower levels of commitment to child welfare for five years. Raceis no longer a significant predictor, but a chance to help childrenand families continues to be a significant predictor of higher levels ofcommitment. These three variables account for 25% of the variance incommitment level.

Finally, Table 10 provides information on what variables predictedwhether workers were actually still at their jobs. In addition todemographic data, reasons for taking the job, and commitment,whether workers had viewed the RJP, rating of their supervision, andjob satisfaction were included in the model.

As Table 10 indicates, workers who saw the Realistic Job Previewvideo were significantly more likely to still be at their jobs, as werethose who felt greater job satisfaction. The RJP was designed torecruit and select public child welfare employees; therefore it wasnot viewed by private agency applicants. Workers who indicatedtheir supervisor made life difficult were significantly more likelyto have left the job. These three variables account for 15% of thevariance.

4. Discussion

The overall turnover rate of frontline workers hired over a 2.5 yearperiod (between November 2004 and April 2007) is relatively low,

Table 9Commitment to remaining in child welfare for next five years.

N=390 Unstandardizedcoefficients

Standardizedcoefficients

Adj. R2=.249 B Std. error Beta T Sig.

(Constant) 3.789 .501 7.559 .000Gender: male = 1 .002 .218 .000 .010 .992Race: worker of color = 1 −.174 .170 −.046 −1.023 .307Public/private: public = 1 1.040 .180 .260 5.782 .000Education −.148 .180 −.037 −.823 .411Help children/families/calling/improve services

1.354 .231 .279 5.873 .000

Only job available −1.191 .253 −.226 −4.713 .000

18.5%, 1 year 9 months to 4 years 2 months after baseline surveyswere completed. This low turnover and high retention rate isespecially positive when compared to annual national turnoverrates of about 20% in the public sector and 40% in the private sector(Alliance et al., 2001). The retention rate compares favorably to otherstates where turnover has been studied. For example, the turnoverrate for public agency workers in Georgia in 2004 was 66% and as highas 100% in some counties (Ellett, Ellett, & Lerner, 2008). Althoughthe state's private agency workers turned over at a higher rate thanpublic agency workers, the difference was not statistically significant.

The reasons for this relative job stability are not altogether clear.These findings preceded the current economic downturn; during thetimeframe of the study, the state unemployment rate varied between6.9 and 8%. Therefore, it is unlikely that the lack of available joboptions played a major role. During the time of the study, howeverthere was a class action lawsuit against the agency, which, in part,might explain retention. The settlement agreement from the lawsuitpromised sweeping changes in the work conditions for child welfarestaff, including dramatically reduced caseloads and support forcontinuing education (Dawyne B v Granholm, 2008).

The fairly high retention rate is consistentwith the fairly high level ofcommitment to both the job and to childwelfare.New frontlineworkersin our study were committed to their jobs (7 point scale: M=5.44 toagency for 2 years; M=4.47 for 5 years) and to child welfare work(M=5.71 for 2 years;M=5.18 for 5 years). Not surprisingly, thosewhohad left their positions by January 2009, had lower levels ofcommitment than those still in their positions (3 of 4 commitmentdifferences being statistically significant). Since commitment seems tobe an important factor in worker turnover, strategies for instillingcommitment (perhaps in worker training) and for recruiting workerswith high commitment deserve further study.

The findings related to reasons for working in a particular agencyand in child welfare are instructive. A promising finding is that morethan 80% of both public and private agency workers chose to work inchild welfare because they wanted a chance to help children andfamilies. Moreover, more than half chose the particular job in orderto help children and families. These findings suggest the state's childwelfare agencies are attracting individuals with the right career goals.There are interesting differences in why public and private agencyworkers took their positions. Public agency workers were motivated

845K.C. Faller et al. / Children and Youth Services Review 32 (2010) 840–846

by pay, benefits, job security, opportunities, and variety. In contrastprivate agency employees took the job because it was the only jobavailable, it was a good first job to take, and they had heard goodthings about the agency. That public agency workers would be morelikely to take jobs because of salaries and benefits is not surprisingbecause, in the state where this research occurred, salaries andbenefits for public child welfare workers are a good deal better than inthe private child welfare agencies (Recruitment and Retention ofChild Welfare Employees, 2008).

The only demographic variable that predicts commitment is race,withworkers of color having significantly lower levels of commitmenton 3 of 4 commitment questions. The state has made considerableefforts to recruit and hire workers of color, as reflected in the 35.4%workers of color statistic. The reasons for the lower commitment levelsfinding are not clear. One possible explanation for racial differences incommitment may be the absence of employment barriers in childwelfare, and the presence of barriers in other employment domains forprofessionals of color. Thus, workers of color may take child welfarepositions not so much out of commitment to the job and the field, butbecause they are discriminated against in other employment domains.An alternative explanation is that they are less committed to jobs intheir agencies and child welfare because the child welfare systemdifferentially substantiates reports ofmaltreatment involving childrenand families of color and disproportionately removes children of colorfrom their homes (e.g., Administration on Children and Families, 2009;Center for the Study of Social Policy, 2009). Thus, workers of colormayregard the child welfare system as discriminatory against children andfamilies of color and therefore be less strongly committed to thesystem. Moreover, these workers completed baseline questionnairesas they completed new worker training, which included federalstatutes such as the Multi-Ethnic Placement Act and the InterethnicAdoption Provisions (e.g., Heifetz Hollinger & the ABA Center onChildren and the Law, 1998) and derived state child welfare policies.Current federal child welfare policy prevents consideration of race inplacement decisions, except in narrowly defined circumstances.Arguably,workers of color, when compared towhiteworkers, responddifferently to a policy that is essentially race-blind. Nevertheless,when race is examined as a predictor of who had left the agency byfollow-up, race is not a significant predictor. Racial differences incommitment levels and possible explanations for them deserveadditional exploration.

On all four commitment variables, being a public agency employeewas predictive of a higher level of commitment. This finding resonateswith the lower turnover rates among public agency workers andhigher rates among private child welfare workers, noted in otherresearch (Alliance et al., 2001; Alwon & Reitz, 2001). However, whenthe variable, public/private agency personnel, was used as a predictorof whether the worker had remained in the job, the difference is notsignificant. But, having watch the RJP was related to still being on thejob, an opportunity that was only available to public child welfareemployees hired after July, 2005.

Although 12 states have developed Realistic Job Previews, only onestate has evaluated the impact of RJPs, the state in which this studywas conducted (Faller et al., 2009). The promising role of viewing RJPduring recruitment as a predictor of still being on the job needs toexamined in other states to deepen our understanding of RJPs. RJPsare home grown, state or region specific products. They vary in theircontent and media techniques. They also are based on the state childwelfare service delivery structure and reflect varying philosophiesabout child welfare work.

Two of the factors that predicted still being on the job at follow-upwere quite consistent with other research: job satisfaction (ChildWelfare League of America, 2002; Landsman, 2001) and goodsupervision (G.A.O., 2003; Lietz, 2008; Landsman, 2007; Mor-Barak,Travis, Pyun, & Xie, 2009; Potter & Brittian, 2009; Yankeelov, Barbee,Sullivan, Antle, 2009). Westbrook and colleagues (Westbrook, Ellis, &

Ellett, 2006) suggest a variety of personal, supervisory, and higheragency administrative factors lead to “committed survivors,” Theyundertook a qualitative study of “committed survivors,” who arepublic child welfare employees and stay in the Georgia public childwelfare system — one with an extraordinarily high turnover rates.

Thus, this study strengthens conclusions about factors alreadystudied in child welfare workforce turnover, the importance ofcommitment to the agency and to child welfare, good supervision,and job satisfaction in preventing turnover. The study also findspublic agencyworkers to bemore committed to their jobs and to childwelfare than private agency workers. It suggests that hiring staff whotake the job because it is the only job available and because it is a goodfirst job to take may not be an optimal strategy. It finds that being aworker of color is associated with lower levels of commitment to theagency and to child welfare; race of worker is a variable that deservesfurther study. Finally, the study finds having viewed the state'sRealistic Job Preview to be associated with still being on the job atfollow-up, a promising finding deserving further study.

5. Implications for policy and practice

1) Commitment to child welfare work leads to lower rates of workerturnover. Low rates of turnover usually are beneficial to both theagency and the clientele. High rates of turnover result in uncoveredcaseloads, increased workloads for remaining workers, low staffmorale, lack of timely case management and service delivery,longer time to permanency, and lack of safety for children.

2) Most child welfare workforce studies do not analyze data byworkerrace (e.g., CWLA, 2009; Mor-Barak et al., 2009; National Center onCrime & Delinquency, 2006). This study suggests the need forattention to hiring committed workers of color and retainingworkers of color. More research is needed, however, about thereasons workers of color are less committed before recruitment andretention strategies can be developed for professionals of color.

3) This study raises concerns about the motivation for taking childwelfare jobs by private agency staff and resonates with otherstudies documenting higher turnover rates in the private sector(Alliance for Children and Families, American Public HumanServices Association, & Child Welfare League of America, May,2001). This study's, findings along with those of other studies,raises a question about the advisability of contracting for childwelfare services with the private sector.

4) Findings also reinforce earlier research about the fundamentalimportance of good supervision for worker retention (e.g.,Landsman, 2007; Mor-Barak, et al., 2009). It is hoped that thenumerous efforts by the Children's Bureau to focus on supervisionin child welfare will positively impact on quality of supervision.

5) Finally, this study suggests the efficacy of a relatively lowinvestment intervention, the Realistic Job Preview (Faller et al.,2009; Graef, 2005) in recruitment, self-selection, retention, andworker morale.

Acknowledgement

The research for this study was funded by the Children's Bureau,U.S. Department of Human Services, Administration on Children,Youth, and Families, Discretionary Programs, Child Welfare TrainingGrant, Award # 90CT0115.

References

Administration on Children and Families. (2009, December). The AFCARS report RetrievedJanuary 24, 2010 from http://www.acf.hhs.gov/programs/cb/stats_research/afcars/tar/report16.htm

Akdogan, A., & Cingoz, A. (2009). The effects of organizational downsizing and layoffson organizational commitment: A field research. The Journal of American Academyof Business Cambridge, 14(2), 337−343.

846 K.C. Faller et al. / Children and Youth Services Review 32 (2010) 840–846

Alliance for Children and Families, American Public Human Services Association, &Child Welfare League of America (2001). The child welfare workforce challenge:Results from a preliminary study. Presented at Finding BetterWays 2001. Dallas, TX.

Alwon, F., & Reitz, A. (2001). Empty chairs: As a national workforce shortage strikeschild welfare: The Child Welfare League responds… http://www/cwla.org/programs/trieschman/emptychairs.htm. Accessed 6/14/2009.

Center for the Study of Social Policy. (2009). Race equity review: Findings from aqualitative analysis of racial disproportionality and disparity for African Americanchildren and families in Michigan's child welfare system Accessed Jan. 23, 2010from http://www.cssp.org/uploadFiles/michigan%20report%201%2014%2009%20FINAL.pdf

Child Welfare League of America. (2001). The child welfare workforce challenge: Resultsfrom a preliminary study. Washington, DC: Author.

Child Welfare League of America. (2002, Sept 1–8). Research round-up: The childwelfare workforce.

Child Welfare League of America. (2009). National data analysis system: Data reportGenerated June 2, 2009 from http://ndas.cwla.org/data_stats/access/predefined/Report.asp?PageMode=1& ReportID=243& GUID={EFFF7178-5CBF-4BD0-8B0D-E1F3772CBDEA}#Table.Children's Bureau (2003–2008).

Child Welfare Staff Recruitment and Retention Training Discretionary Grant Cluster.(2003–2008). Retrieved August 20, 2009 from http://www.childwelfare.gov/systemwide/funding/funding_sources/cluster.cfm

Collins-Camargo, C., & Kelly, M. (2007). Supervisor as informal mentor: Promotingprofessional development in public child welfare. The Clinical Supervisor, 25(1–2),127−146.

CPS Human ResourcesChampnoise, C. (2007). The RJP tool kit: A how-to guide fordeveloping a realistic job preview. Cornerstones 4 Kids & CPS Human ResourcesRetrieved on January 29, 2009, from http://portal.cornerstones4kids.org/stuff/contentmgr/files/b0dc9183b6b7bfe3baa9a4d25caef9c1/folder/rjp_toolkit_final.pdf

CPS Human Resources, Champnoise, C., & Masternak, M. (2004, June 17). Realistic jobpreview: A review of the literature and recommendations for Michigan familyindependence agency.: Anne E. Casey Foundation Retrieved from www.AECF.Org

Ellett, A. J., Ellett, C. D., & Rugutt, J. K. (2003,March).A study of personal and organizationalfactors contributing to employee retention and turnover in child welfare in Georgia.Athens, Georgia: School of Social Work, University of Georgia.

Ellett, A., Ellett, C., & Lerner, B. (2008). August 6). Paper given at Scaling the Summit,Denver, CO: Employee selection protocol to improve child welfare retention andpractice.

Faller, K. C., Masternak, M., Grinnell-Davis, C., Grabarek, M., Sieffert, J., & Bernatovicz, F.(2009). Realistic job previews in child welfare: State of innovation and practice.Child Welfare, 88(5), 23–48.

Flower, J., McDonald, J., & Sumski, M. (2005). Review of turnover in Milwaukee privateagency child welfare ongoing case management staff Retrieved June 26, 2009 fromhttp://dhfs.wi.gov/bmcw/progserv/Initiatives/turnoverstudy.pdf

General Accounting Office. (2003, March). Child welfare: HHS could play a greater rolein helping child welfare agencies recruit and retain staff. Washington, D.C: GeneralAccounting Office GAO 03-357.

General AccountingOffice. (2004, April). Child and family services reviews: Better use ofdata and improved guidance could enhance HHS's oversight of state performance.Washington, D.C: General Accounting Office GAO-04-333.

General Accounting Office. (2006, October). Child welfare: Improving Social serviceprogram, training, and technical assistance information would help address long-standing service-level and workforce challenges. Washington, D.C: GeneralAccounting Office GAO 07-75.

Gonzalez, R. P., Faller, K. C., Ortega, R. M., & Tropman, J. (2009). Exit interviews of childwelfare workers. Journal of Public Child Welfare, 3(1), 40−63.

Graef, M. (2005, January). “What am I getting into?” Using a realistic job preview toimprove retention. Invited presentation at the Western Regional Recruitment andRetention Project Institute, Denver, CO.

Graef, M., & Hill, E. (2000). Costing child protective services staff turnover. (Usingadministrative data in childwelfare) (statistical data included). ChildWelfare, 79(5),517−534.

Graef, M. I., & Potter, M. E. (2002). Alternative solutions to the child protective servicesstaffing crisis: Innovations from industrial/organizational psychology. http://www.americanhumane.org/Protecting Children, 17(3), 18−31 Retrieved February 20,2009 from http://ccfl.unl.edu/publications/cwtraining/Graef_Potter.pdf

Granholm, D. B. (2008, October). Settlement agreement Accessed May14, 2009 at http://www.childrensrights.org/reform-campaigns/legal-cases/michigan-dwayne-b-v-granholm/

Heifetz Hollinger, J.The ABA center on children and the law. (1998). A guide to theMultiethnic Placement Act of 1994 as amended by the Interethnic AdoptionProvisions of 1996. Washington, D.C: National Resource Center on Legal and CourtIssues.

Knudson, H. K., Johnson, J. A., Martin, K. J., & Roman, P. M. (2003). Downsizing survival:The experience of work and organizational commitment. Sociological Inquiry, 23(2),265−283.

Lambert, E. G., Pasukpuleti, S., Cluse-Tolar, T., Jennings, M., & Baker, D. (2006). Theimpact of work-family conflict on social work and human service worker jobsatisfaction and organizational commitment: An exploratory study. Administrationin Social Work, 30(3), 55−74.

Landsman, M. (2001, Sept). Commitment in public child welfare. Social Services Review(pp. 386−419).

Landsman, M. (2007). Supporting child welfare supervisors to improve retention. ChildWelfare, 86(2), 105−124.

Landsman, M. (2008, July 24). Strengthening the child welfare workforce for improvedchild and family outcomes. Presentation given at the North Dakota Children's JusticeSymposium, Fargo, ND.

Lietz, C. A. (2008). Implementation of group supervision in child welfare: Findings fromArizona's supervision circle project. Child Welfare, 87(6), 31−48.

Meyer, J. P., & Allen, N. J. (1991). A three-component conceptualization of organizationalcommitment. Human Resource Management Review, 1, 61−98.

Mor-Barak, M., Travis, D., Pyun, H., & Xie, B. (2009). The impact of supervision onworker outcomes: A meta-analysis. Social Service Review, 83(1), 3−32.

National Association of Social Workers (NASW). (2003). The child welfare workforceRetrieved July 8, 2009 from http://www.socialworkers.org/advocacy/updates/2003/082003_a.asp

National Center on Crime & Delinquency. (2006). Job turnover in child welfare andjuvenile justice: The voices of former frontline workers Retrieved October 15, 2009from http://www.cornerstones4kids.org/images/nccd_voices_306.pdf

Potter, C., & Brittian, C. (Eds.). (2009). Child welfare supervision: A practical guide forsupervisors, managers, and administrators. New York: Oxford University Press.

Recruitment and Retention of ChildWelfare Employees. (2008). Michigan child welfareworkforce study: A summary report of incoming workers Found on http://www.ssw.umich.edu/public/currentProjects/rrcwp/research.html

Strolin, J., McCarthy, M., Lawson, H., Smith, B., Caringi, J., & Bronstein, L. (2006). Should Istay or should I go? A comparison study of intention to leave among public childwelfare systems with high and low turnover rates Retrieved August 21, 2009 fromhttp://www.ocfs.state.ny.us/ohrd/swec/pubs/Turnover%20in%20Public%20Child%20Welfare.pdf

Wanous, J. P. (1989). Installing a realistic job preview: Ten tough choices. PersonnelPsychology, 42, 117−134.

Wanous, J. P. (1992). Organizational entry: Recruitment, selection and socialization ofnewcomers, (2nd ed.). Reading, MA: Addison Wesley Publishing Company.

Westbrook, T., Ellis, J., & Ellett, A. (2006). Improving retention among public childwelfare workers: What can we learn from insights and experiences of committedsurvivors? Administration in Social Work, 30(4), 37−62.

Yankeelov, P. A., Barbee, A. P., Sullivan, D., & Antle, B. F. (2009). Individual andorganizational factors in job retention in Kentucky's child welfare agency. Childrenand Youth Services Review, 31(5), 547−554.