8
Individual and organizational factors in job retention in Kentucky's child welfare agency Pamela A. Yankeelov , Anita P. Barbee, Dana Sullivan, Becky F. Antle Kent School of Social Work, University of Louisville, Louisville, KY, 40292, United States abstract article info Article history: Received 6 August 2008 Received in revised form 17 October 2008 Accepted 21 October 2008 Available online 31 October 2008 Keywords: Retention Child welfare Supervision Predictors A major concern in the eld of child welfare continues to be the high rate of employee turnover. The purpose of this research was to examine a variety of individual and organizational factors that differentiate leavers from the stayers in child welfare using a prospective design and data from the Kentucky Cabinet for Health and Family Services (CHFS). The results of this study show that the stayers (N = 448) and leavers (N = 275) in this child welfare agency did not differ in their gender, race, or cumulative G.P.A. MSWs were more likely to leave the agency and employees with majors other than psychology or social work were more likely to stay. Stayers reported being more attached to their supervisors and receiving more guidance. Rural workers also perceived greater support from their supervisors than did urban workers. Implications for practice are discussed. Published by Elsevier Ltd. 1. Introduction A major concern in the eld of child welfare continues to be the high rate of employee turnover (Drake & Yadama, 1996; US GAO, 2003; Rycraft, 1994). The strength of the child welfare workforce is of great national concern and has an impact on services provided to abused and neglected children and their families (CWLA, 2002; Pew Commission, 2004). Other professions also experience burnout and turnover, but it is particularly problematic for a eld like child welfare, which concentrates its productive capacity in human capital, that is, the work is concentrated on the knowledge, skills, and ability of people, working with other people such as namely, children and families (Balfour & Neff, 1993). High worker turnover has been found to have great impact on the staff and functioning of the child welfare agency (Graef & Hill, 2000). Vacancies are created and it can be difcult to nd trained replacement workers who can immediately take over a caseload, leading to additional workload burden on existing staff and lack of continuity of work with families. Agencies also have to bear the nancial cost that results from high turnover, related to processing employees out of the system upon resignation, the administrative costs to ll a vacant position, and the cost of training a new worker. It has been estimated that the average cost per vacancy is around $10,000 (which was based on 1995 gures) (Graef & Hill, 2000). Zlotnik, DePanlis, Daining and Lane (2005) published a systematic review on the factors that inuence retention of child welfare staff. The review included six qualitative studies, twelve quantitative studies and six mixed methodology studies of which 52% were unpublished manuscripts. Zlotnik et al. (2005) concluded that retention was impacted by personal and organizational factors, as well as, Title IV-E strategies. At that time, Zlotnik et al. (2005) identied approximately seven quantitative studies which noted the impact of both personal and organizational variables on retention (i.e., Dickinson & Perry, 2002; Ellett, 2000; Ellett, Ellett, & Rugutt, 2003; Harris et al., 2000; Nissly, Mor Barak, & Levin, 2005; UALR SSW, 2002a, b) and only one prospective study which explored the effects of the dimensions of burnout, a personal factor, on retention (i.e. Drake & Yadama, 1996). Since that time, three prospective, published studies could be located that included a wide variety of both personal and organizational factors (e.g. Curry, McCarragher, & Dellman-Jenkins, 2005; Smith, 2005; Weaver, Chang, Clark, & Rhee, 2007). The purpose of this research was to add to this small body of literature by examining a variety of individual and organizational factors that differentiate the leavers from the stayers in child welfare using a prospective design and data from the Kentucky Cabinet for Health and Family Services (CHFS). Of the quantitative studies, regardless of their design, the personal factors that have consistently been studied include education (e.g. Curry et al., 2005; Nissly et al., 2005; Weaver et al., 2007), job satisfaction (e.g. Cahalane & Sites, 2004; Dickinson & Perry, 2002; Ellett et al., 2003; Weaver et al., 2007), burnout (e.g. Cahalane & Sites, 2004; Dickinson & Perry, 2002; Smith, 2005), personal and professional commitment (e.g. Ellett et al., 2003; UALR SSW, 2002a,b; Weaver et al., 2007) and role conict (e.g. Cahalane & Sites, 2004; Weaver et al., 2007). The most consistently studied organizational factors included co-worker support, supervisor support and quality of supervision (e.g. Curry et al., 2005; Dickinson & Perry, 2002; Ellett, 2000; Ellett et al., 2003; Nissly et al., 2005; Smith, 2005; UALR SSW, 2002a,b; Weaver et al., 2007), workload (e.g. Ellett, 2000; Smith, 2005), salary (e.g. Ellett et al., 2003, Weaver et al., Children and Youth Services Review 31 (2009) 547554 Corresponding author. Tel.: +1 502 852 0426; fax: +1 502 852 0422. E-mail address: [email protected] (P.A. Yankeelov). 0190-7409/$ see front matter. Published by Elsevier Ltd. doi:10.1016/j.childyouth.2008.10.014 Contents lists available at ScienceDirect Children and Youth Services Review journal homepage: www.elsevier.com/locate/childyouth

Individual and organizational factors in job retention in Kentucky's child welfare agency

Embed Size (px)

Citation preview

Page 1: Individual and organizational factors in job retention in Kentucky's child welfare agency

Children and Youth Services Review 31 (2009) 547–554

Contents lists available at ScienceDirect

Children and Youth Services Review

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

Individual and organizational factors in job retention in Kentucky's childwelfare agency

Pamela A. Yankeelov ⁎, Anita P. Barbee, Dana Sullivan, Becky F. AntleKent School of Social Work, University of Louisville, Louisville, KY, 40292, United States

⁎ Corresponding author. Tel.: +1 502 852 0426; fax:E-mail address: [email protected] (P.A.

0190-7409/$ – see front matter. Published by Elsevier Ldoi:10.1016/j.childyouth.2008.10.014

a b s t r a c t

a r t i c l e i n f o

Article history:

A major concern in the field Received 6 August 2008Received in revised form 17 October 2008Accepted 21 October 2008Available online 31 October 2008

Keywords:RetentionChild welfareSupervisionPredictors

of child welfare continues to be the high rate of employee turnover. The purposeof this research was to examine a variety of individual and organizational factors that differentiate leaversfrom the stayers in child welfare using a prospective design and data from the Kentucky Cabinet for Healthand Family Services (CHFS). The results of this study show that the stayers (N=448) and leavers (N=275)in this child welfare agency did not differ in their gender, race, or cumulative G.P.A. MSWs were more likelyto leave the agency and employees with majors other than psychology or social work were more likely tostay. Stayers reported being more attached to their supervisors and receiving more guidance. Rural workersalso perceived greater support from their supervisors than did urban workers. Implications for practice arediscussed.

Published by Elsevier Ltd.

1. Introduction

Amajor concern in the field of child welfare continues to be the highrate of employee turnover (Drake & Yadama, 1996; US GAO, 2003;Rycraft, 1994). The strength of the child welfare workforce is of greatnational concern and has an impact on services provided to abused andneglected children and their families (CWLA, 2002; Pew Commission,2004). Other professions also experience burnout and turnover, but it isparticularly problematic for a field like childwelfare, which concentratesits productive capacity in human capital, that is, thework is concentratedon the knowledge, skills, andabilityof people,workingwithotherpeoplesuch as namely, children and families (Balfour &Neff,1993). Highworkerturnoverhasbeen found tohavegreat impacton the staff and functioningof the child welfare agency (Graef & Hill, 2000). Vacancies are createdand it can be difficult to find trained replacement workers who canimmediately take over a caseload, leading to additionalworkload burdenon existing staff and lack of continuity of work with families. Agenciesalsohave tobear thefinancial cost that results fromhigh turnover, relatedto processing employees out of the system upon resignation, theadministrative costs to fill a vacant position, and the cost of training anew worker. It has been estimated that the average cost per vacancy isaround $10,000 (which was based on 1995 figures) (Graef & Hill, 2000).

Zlotnik, DePanfilis, Daining and Lane (2005) published a systematicreview on the factors that influence retention of child welfare staff.The review included six qualitative studies, twelve quantitativestudies and six mixed methodology studies of which 52% wereunpublished manuscripts. Zlotnik et al. (2005) concluded that

+1 502 852 0422.Yankeelov).

td.

retention was impacted by personal and organizational factors, aswell as, Title IV-E strategies. At that time, Zlotnik et al. (2005)identified approximately seven quantitative studies which noted theimpact of both personal and organizational variables on retention (i.e.,Dickinson & Perry, 2002; Ellett, 2000; Ellett, Ellett, & Rugutt, 2003;Harris et al., 2000; Nissly, Mor Barak, & Levin, 2005; UALR SSW, 2002a,b) and only one prospective study which explored the effects of thedimensions of burnout, a personal factor, on retention (i.e. Drake &Yadama, 1996). Since that time, three prospective, published studiescould be located that included a wide variety of both personal andorganizational factors (e.g. Curry, McCarragher, & Dellman-Jenkins,2005; Smith, 2005; Weaver, Chang, Clark, & Rhee, 2007). The purposeof this research was to add to this small body of literature byexamining a variety of individual and organizational factors thatdifferentiate the leavers from the stayers in child welfare using aprospective design and data from the Kentucky Cabinet for Health andFamily Services (CHFS).

Of the quantitative studies, regardless of their design, the personalfactors that have consistently been studied include education (e.g. Curryet al., 2005; Nissly et al., 2005;Weaver et al., 2007), job satisfaction (e.g.Cahalane & Sites, 2004; Dickinson & Perry, 2002; Ellett et al., 2003;Weaver et al., 2007), burnout (e.g. Cahalane & Sites, 2004; Dickinson &Perry, 2002; Smith, 2005), personal and professional commitment (e.g.Ellett et al., 2003; UALR SSW, 2002a,b; Weaver et al., 2007) and roleconflict (e.g. Cahalane & Sites, 2004; Weaver et al., 2007). The mostconsistently studied organizational factors included co-worker support,supervisor support and quality of supervision (e.g. Curry et al., 2005;Dickinson & Perry, 2002; Ellett, 2000; Ellett et al., 2003; Nissly et al.,2005; Smith, 2005; UALR SSW, 2002a,b; Weaver et al., 2007), workload(e.g. Ellett, 2000; Smith, 2005), salary (e.g. Ellett et al., 2003,Weaver et al.,

Page 2: Individual and organizational factors in job retention in Kentucky's child welfare agency

548 P.A. Yankeelov et al. / Children and Youth Services Review 31 (2009) 547–554

2007) and organizational commitment (e.g. Cahalane & Sites, 2004;Ellett, 2000; Smith, 2005).

The Kent School of Social Work, University of Louisville has had acontract with the CHFS to evaluate child welfare training for the last16 years. One ten-year study of the Kentucky PCWCP program (Barbeeet al., in press) included the collection of individual and organizationalfactors that could impact PCWCP graduates. The individual measuresincluded variables explored in other retention studies such as gender,and education, but also included variables rarely or less often exploredincluding worker's region, specifically urban and rural, and astandardized measure of personality. The organizational measuresincluded a multi-dimensional measure of perceived workplaceenvironment, supervisor and co-worker technical and social support,as well as, the impact of the educational program. It was found in thatstudy that co-worker and supervisory support affect intent to stay andactual retention, especially in urban areas. For this current study,similar measures were administered during new employee trainingand personnel records at CHFS were later reviewed for these traineesto identify employee retention and movement through theorganization.

With rates of employee turnover in public child welfare as high as40% (US GAO, 2003) and the funding opportunities afforded by theChildren's Bureau to support supervisory interventions, it is importantto continue to explore more deeply the individual and organizationalfactors that differentiate child welfare workers who stay versus thosewho choose to leave with the use of longitudinal designs andmultidimensional, standardized scales, so that the most appropriateand comprehensive intervention strategies can be implemented toimprove retention rates. This study aims to further the understandingof employee turnover because of its unique combination of variables.The following section will discuss more specifically the impact ofvarious individual and organizational factors relevant to this currentstudy on employee turnover in child welfare.

1.1. Education and training

The Pew Commission (2004), a bipartisan group charged withdeveloping recommendations to improve outcomes for children infoster care, recognized the importance of the strength of the childwelfare workforce in meeting this goal and recommended givingfinancial incentives for states that acted to improve the quality of theworkforce by improving the competence level (including increasingthe number of workers with BSW and MSW degrees). A few studiescomparing the education and experiences of child welfare workersfound that workers with BSWs and MSWs rated themselves as betterprepared for thework in child welfare than did workers with differenteducational backgrounds (Dhooper, Royse, & Wolfe, 1990; Lieberman,Hornby, & Russell, 1988). Dhooper (1990) also found that child welfareworkers who had an educational background in social work werebetter prepared than workers who had other degrees. Unfortunately,greater intent to leave child welfare or to actually leave child welfarehas been found among workers with graduate degrees in comparisonto bachelor degrees (e.g. Balfour & Neff, 1993; Curry et al., 2005; Nisslyet al., 2005). Furthermore, Rosenthal et al. (1998) found greaterturnover among child welfare workers with a Master in Social Work(MSW)when compared to workers with aMasters in a human servicefield. By contrast, Weaver et al. (2007) found that although MSWswere more likely to report a greater intention to leave; they werenot more likely to do so. Likewise, Landsman (2001) and Ellett et al.(2003) found support that child welfare workers with MSWs tendedto remain on the job. Collectively, these studies present inconsistentpatterns of the impact of graduate education on retention. Conversely,Title IV-E programs have tried to combat retention issues by providingspecialized training in child welfare and have noted significantlybetter retention rates even after pay back periods have ended(Rosenthal & Waters, 2006, Barbee et al., in press).

1.2. Personality and values

In this current study, a standardized measure of personality wasincluded to aid in the understanding of what type of employees werebest suited for sustained child welfare employment. Only one studycould be found that explored the relationship between personalityand turnover among social workers. Koeske and Kirk (1995) foundthat only psychological well being was related to the retention ofsocial workers. In a meta-analysis on the effects of personality oncounterproductive behaviors for a diverse group of workers notspecific to the child welfare profession, Salgado (2002) found that allfive factors of the Big Five Personality Inventory were valid predictorsof a lack of turnover with emotional stability, conscientiousness, andagreeableness as the most predictive of the lack of turnover.

Other studies have explored various traits that predict turnoveramong child welfare workers. Landsman (2001) found that beingcommitted to the field of social work was the strongest contributor tobeing satisfied with the job and the second strongest contributor tocommitment to the occupation. Landsman (2001), Samantrai (1992),and Rycraft (1994) suggest that having an “orientation to service” isthe most important factor in being committed to child welfare, and atrait that should be looked for when recruiting new workers who arelikely to remain on the job, and be committed to the agency and field.

1.3. Job satisfaction and salary

Job satisfaction has been measured using a myriad of uni- andmulti-dimensional scales in retention studies. In general, job satisfac-tion has been found to be positively correlated with intention toremain employed among child welfare workers (Ellett et al., 2003;Landsman, 2001; Mor Barak et al., 2001; Weaver et al., 2007). Incontrast, higher salaries and satisfactionwith salary has been found attimes to be related to intention to remain employed or actual turnoverin child welfare (e.g. Dickinson & Perry, 2002; Jayaratne & Chess,1984;Koeske & Kirk, 1995), while other studies have failed to show arelationship between salary and turnover (Mor Barak et al., 2001;Smith, 2005).

1.4. Regional differences

Sundet and Cowger (1990) found that rural, in comparison tourban, child welfare workers thought that their setting was onlymoderately problematic or stressful, however they did indicate feelingprofessionally isolated and that there were few opportunities foradvancement in the profession. In a series of focus groups, Westbrook,Ellis & Ellet (2006) explored the experiences of urban, suburban, andrural child welfare “committed survivors”. Rural and suburban“committed survivors” emphasized the importance of leadershipmore than the urban “committed survivors”. The rural workers, incomparison to urban workers, also reported greater faith in theirsupervisors, emphasized the importance of teamwork, and discussedless frequently the importance of the ability to move through theagency as valuable survival influences.

1.5. Organizational factors

Previous research in this area had consistently identified super-visor support and co-worker support as critical to preventingemployee burnout and intention to leave or actual turnover (Curryet al., 2005; Davis-Sacks, Jayaratne, & Chess, 1985; Dickinson & Perry,2002; Mor Barak et al., 2001; Rycraft, 1994; Samantrai, 1992; Smith,2005;Weaver et al., 2007). There has been discussion as to the reasonsfor the existence of this relationship in child welfare. Ellett, Collins-Carmargo, and Ellett (2006) delineated the critical components ofquality or competent supervision in public child welfare whichincludes not only an introduction to the culture of the organization

Page 3: Individual and organizational factors in job retention in Kentucky's child welfare agency

549P.A. Yankeelov et al. / Children and Youth Services Review 31 (2009) 547–554

but also instruction on how to strive for the most effective practice inthe face of differing work activities, large volume of work, stress of thedaily work and challenges faced in child welfare work. In support ofthis conceptualization, Reagh (1994) also found that the balance ofworking with people and completing all necessary paperwork wasdifficult, and that a better balance may contribute to higher retentionrates. Drake and Yadama (1996) also found that emotional exhaustionhad a direct effect on exiting from the job, not just intention toturnover, and that depersonalization was not related to retention inchild welfare. This means that although some workers do havefeelings of depersonalization, they tend to remain on the job, whichcould be detrimental to working with clients and result in sub parcasework. More recently, Conrad and Kellar-Guenther (2006) exploredthe relationship between compassion fatigue (distress experienced bythe helper), burnout, and compassion satisfaction (fulfillment fromhelping others and positive relationship with colleagues) for childwelfare workers. They found that child welfare workers with highcompassion satisfaction had lower levels of compassion fatigue andlower levels of burnout. They suggest that social support may becrucial in buffering workers from their emotional draining daily work.Additionally, DePanfilis (2006) concluded in her review of theliterature on compassion fatigue that child welfare managers shouldbe cognizant of their important role in helping workers effectivelycoping with secondary trauma. In relation to training support,Wehrmann, Shin, and Poertner (2002) found that with greateropportunity to use the new skills learned at training on the job andgreater availability of peer support given to workers post-training, thebetter the transfer of training knowledge. Supervisors can definitelycreate a structure to allow practice of new skills and can encourage anenvironment of open discussion and support. Collectively, thesestudies suggest that child welfare supervisors can aid their workers inbalancing the paperwork and interpersonal work with clients, providesocial support including opportunities to vent the stress of the work,refocus the purpose of the work and lastly, and support the skills andknowledge gained at training.

Ellett et al. (2006) found that quality supervision should increasethe self-efficacy of the worker. Drawing on Bandura's principles of selfmastery, they identified ways to accomplish this instruction by 1)supervisors offering work tasks to staff that meet their skilldevelopment, 2) supervisors modeling best practices, and 3) super-visors using verbal persuasion that is motivating and demonstratesaffective support of their workers as a person and a professional. In aneffort to explore these varying dimensions of supervision and themost important aspects of supervision in child welfare on retentionand turnover, supervisor support has been measured in a variety ofways across studies including competency (e.g. Ellett et al., 2003;Smith, 2005; Weaver et al., 2007), social support (e.g. Dickinson &Perry, 2002; Ellett, 2000; Nissly et al., 2005; Smith, 2005), professionalsharing (e.g. Ellett, 2000) and support for attending training and useof the knowledge and skills gained at training (e.g. Curry et al., 2005).The use of various measurement tools lends credence to theawareness that supervision is a multi-dimensional construct.Although peer and supervisor support have been measured usingmultidimensional scales, there has a tendency to collapse theemotional and technical dimensions in favor of a combined measureor only to measure one dimension over the full range of options.Through the use of a multidimensional measure of peer and super-visor support, this study hopes to explore more fully the effects ofsupervisory support on retention among child welfare workers.

2. Methodology

2.1. Research question

This study focused specifically on retention among a sample of childprotection workers in the state of Kentucky. The primary research

question was the following: what are the individual and organizationalfactors that predict job retention in child welfare workers in Kentucky?The first hypothesis was that BSWs would stay longer than those withother undergraduate degrees. We were uncertain as to the impact ofMSWs on retention since those findings were mixed in the literature.The second hypothesis was that the personality traits of conscientious-ness, emotional stability and agreeableness would be positivelycorrelated with retention. The third hypothesis was that supervisorand co-worker support would be positively correlated with retention.The fourth hypothesiswas that a positive viewof thework environmentwould be positively correlated with retention. The fifth hypothesis wasthat supervisory interest in trainingwould be positively correlated withretention.

2.2. Design

This research utilized a non-experimental, prospective researchdesign. In Phase One of this study, trainees were requested tocomplete a pre-training questionnaire prior to the start of the coretraining and a post-training questionnaire immediately following thelast unit of this core training. On both occasions, the trainees wereinformed that the purpose of the questionnaire was to explore factorsthat could affect job outcomes. They were informed that theirindividual responses to all of the questions would be held in thestrictest confidence. They were also told that their individualresponses would not be shown to their supervisors or to membersof the Department's Central Administration. Lastly, they wereinformed that any data presented to Central Administration wouldbe presented in aggregate form across all training participantsstatewide. In Phase Two of this study, employee retention recordswere obtained and matched to each individual's Phase one data.Groups were identified on the basis of worker retention—those whoremained with the organization and those who were no longer withthe organization. Group differences on demographics, personality,social support from co-workers and supervisors, and trainingeffectiveness measures were examined. The study was conducted incompliance with our institution's Internal Review Board (IRB).

2.3. Sampling

The sample was drawn from the population of all new employeeshired by the agency between the years of 1996 and 1999. Individual,training and work environment measures were taken prior to andimmediately after the completion of an initial training for newemployees and then their personnel records were examined in the Fallof 2001. Seven hundred twenty-three employees participated in thetraining during this time period, accepted the invitation to participatein the study, and provided their social security number on both the preand post test measures. The participants were classified into twodistinct categories, specifically stayers (N=448) and leavers (N=275).These terms were applied to child welfare personnel retentionpatterns by Rycraft (1994). Each category had approximately fivesub categories, including those who remained in their same position,those who were promoted both within their unit and outside of theirunit, moved from region to region within Kentucky, and those whomoved across Cabinets (Health versus Family Services versus JuvenileJustice). Approximately 62% of the sample was categorized as stayers,while 38% of the sample was categorized as leavers. Table 1 providesthe retention rates for each cohort by year. The overall retention ratefor the sample during the first year of employment was 86%. Theretention rate dropped to 72% during the second year of employment,and down to 64% during the third year of employment. Approximately81% of the entire sample was female, and 85.2% of the sample wasCaucasian, while 14.8% were people of color. The average age of thesample was 30.01 with a standard deviation of 8.19. The sample had amean undergraduate G.P.A. of 3.16 (SD=.43). Approximately 41.3% had

Page 4: Individual and organizational factors in job retention in Kentucky's child welfare agency

Table 1Percentage of employees retained and counts for each cohort by year.

Hire year Total # b1 year 12–24months

25–36months

37– 48months

49–60months

1996 150 91% (137) 83% (124) 75% (112) 67% (101) 49%(73)1997 127 86% (109) 70% (89) 62% (79) 48% (61) –

1998 157 80% (126) 69% (108) 51% (80) – –

1999 287 88% (253) 66% (189) – – –

550 P.A. Yankeelov et al. / Children and Youth Services Review 31 (2009) 547–554

a Bachelor of Social Work, 25.1% had a Bachelor of Arts in Psychology,28.6% had a Bachelor of Arts in variety of other majors and 5% had aMaster of Social Work degree.

2.4. Measures

As previously noted, participants completed a pre and post trainingquestionnaire in Phase One. In the pre-training questionnaire, traineeswere asked to complete the Big Five Personality Inventory (Goldberg,1992), the Social Provision Scale (Cutrona & Russell, 1987), a multi-dimension workplace environment inventory (Herzberg, 1966), anditems assessing the quality of work assistance provided by theirsupervisor in four areas. In the post-training questionnaire, traineeswere asked to rate the trainer's and training qualities, and to rate theirsupervisor's attitude toward training on author generated indices. InPhase Two, the employee identification number, which was docu-mented on Phase One's measures was used to match employmentdata from personnel records. The following provides a description ofeach of the Phase One measures.

2.5. The Big Five

The Big Five Personality Inventory (Goldberg, 1992) is a standar-dized 100 item scale measuring 5 personality attributes, specifically,extroversion, emotional stability, openness to experience, agreeable-ness and conscientious. Each of the five attributes is assessed by 20adjectives, 10 that describe the presence and 10 that describe theabsence of the attribute. Trainees were asked to indicate the degree towhich each adjective represents their general approach on a scalefrom 1 (strongly disagree) to 5 (strongly agree). The Cronbach alphasfor the 5 subscales are as follows: .50 (extraversion), .81 (emotionalstability), .89 (conscientiousness) .89 (openness to experience), and.90 (agreeableness). The reliability and validity of this scale wereestablished through large-scale studies comparing Italians andAmericans. The average weighted mean coefficient for the five sub-scales is 0.75 (Viswesvaran & Oanes, 2000). Construct validity of thisscale was supported through high correlations with the similar NEO-PI scale (Barbaranelli, Caprara, & Maslach, 1997).

2.6. Social Provisions Scale

The Social Provisions Scale (Cutrona & Russell, 1987) measures theperceived social functions or “provisions” received from relationshipswith others. Trainees were asked to rate the degree to which they felteach of 18 statements reflected the type of relationship they have withtheir co-workers and with their supervisor on a scale from 1 (stronglydisagree) to 4 (strongly agree). The 18 statements are grouped into 5areas, approximately 3 or 4 items per area, and address the degree towhich their relationship supplies guidance (advice or information),reliable alliance (assurance that the other can be relied upon for tangiblesupport), reassurance ofworth (recognition of one's expertise and skill),social integration (a sense of belonging to a group that shares similarinterests, concerns and leisure activities), and attachment (emotionalcloseness and sense of security). Past researchhas provided evidence forthe validity of this scale in predicting coping with stressful experiencesfor the elderly, postpartumwomen, public school teachers and hospital

nurses (for a review, see Cutrona & Russell, 1987). The internalconsistency of the scale for co-worker relationships was .93 withreliabilities of the co-worker subscales ranging from .70 (reliablealliance) to .83 (guidance). The internal consistency of the scale forsupervisor relationships was .91 with reliabilities of the supervisorsubscales ranging from .74 (reliable alliance) to .83 (attachment).

2.7. Supervisors' quality of help provided

Trainees were asked at the pretest to rate the quality of helpreceived from their supervisor in relation to assessments, case plans,dealing with families and navigating policies and procedures. Itemswere rated on a scale from 1 (not at all) to 5 (very high quality). Theinternal consistency of these four items was .95.

2.8. Workers' view of workplace environment

Trainees were also asked to complete a 27 item multi-dimensionjob work place environment inventory based on Herzberg (1966)'sprominent, motivator-hygiene theory. This index assessed intrinsicand extrinsic job satisfaction factors by measuring the employee'ssense of achievement, responsibility, the work itself, recognition, co-worker cooperation, competent supervision, relationship with super-visor, salary, security, department policies, working conditions,autonomy, and status. Trainees were asked to rate their level ofagreement with the statements on a scale from 1 (strongly disagree)to 5 (strongly agree). The Cronbach alphas of the 12 subscales rangedfrom .12 to .79. It was decided that only the top 3 subscales would beused given the poor reliability of the other 9 subscales. The top 3subscales included competent supervision (Alpha=.77), relationshipwith supervisor (Alpha=.79) and salary (Alpha=.79).

2.9. Workers' perception of supervisor's attitude toward training

On the post-training questionnaire, trainees were also asked toindicate their supervisor's attitude toward training. Six items werecreated by the authors to assess the supervisor's attitude includingitems that addressed whether they thought their supervisor held ageneral positive impression of training, a positive attitude towardtheir attendance, thought training was time well spent, thoughttraining provided was practical to their work, thought training wouldimprove their work performance and increase their knowledge. Eachitem was rated on a scale from 1 (not at all) to 7 (very much so). TheCronbach alpha for these six items was .94.

2.10. Perceptions of training

On the post-training questionnaire, trainees were asked to indicatethe degree to which they thought their trainer exhibited 10 character-istics. Examples of the characteristics include if the trainer showedinterest in each trainee, was effective in keeping the training interactiveand well paced, used examples that were relevant to theworkers, had agood understanding of the trainees' work environment, and hadsufficient experience in the area they taught. The Cronbach alpha forthese 10 items was .94. Additionally, trainees were asked to rate thetraining on 8 items addressing the degree of practicality, importance,increased knowledge and skill, changed confidence in dealing withfamilies, in writing case assessments and in writing case plans. TheCronbach alpha for these 8 items was also 93.

3. Results

3.1. Analyses of demographic characteristics: stayers versus leavers

Separate Chi-Square analyses were conducted to examine thedifference between the stayers and the leavers on region (urban,

Page 5: Individual and organizational factors in job retention in Kentucky's child welfare agency

551P.A. Yankeelov et al. / Children and Youth Services Review 31 (2009) 547–554

rural), gender, race, and education. Employees were more likely to stayif they were employed in rural regions than if they were employed inurban regions, χ2 (1)=15.62, pb .001. Males and females were equallylikely to stay and leave the agency, χ2 (1)=2.22, ns. There was norelationship between race (Caucasian, People of Color) and employ-ment status, χ2 (1)=1.82, ns. Lastly, there was an association betweendegree (BSW, BA in psychology, BA in other majors, MSW) andemployment status, χ2 (1)=9.80, pb .02, Employees with a MSSWwere more likely to leave than expected by chance and employeeswith an undergraduate degree of any kind (excluding psychology andsocial work majors) were more likely to stay. Thus, there was nosupport for our first hypothesis, BSWs did not stay longer than thosewith other undergraduate degrees, but this study adds to the literaturein showing greater turnover for the MSWs. As a result of theassociation between employment status and region, all forthcominganalyses will include both of these factors.

3.2. Personality differences across groups

Personality differences between groups were examined using amultivariate analysis of variance (MANOVA) with two between groupfactors, Employment Status, with two levels (stayers, leavers) andRegion, with two levels (urban, rural). Dependent variables for themultivariate analysis included scores on the Big Five, specificallyconscientiousness, neuroticism, extraversion, agreeableness, andopenness to experience. There was no main effect for EmploymentStatus, F(5,635)=.81, ns, and partial η2=.01, power=.29. There wasno main effect for Region, F(5, 635)=1.88, ns, and partial η2=.02,power=.64. Lastly, there was no Employment Status×Region interac-tion, F(5, 635)=1.01, ns, and partial η2=.01, power=.37. Thus, therewas no support for the second hypothesis that the personality traits ofconscientiousness, emotional stability and agreeableness wouldpredict retention.

3.3. Workers' perceptions of the social provisions offered by co-workers

Differences in perceptions of the social provisions offered by co-workers were examined using a multivariate analysis of variance(MANOVA)with two between group factors, Employment Status, withtwo levels (stayers, leavers) and Region, with two levels (urban, rural).Dependent variables for the multivariate analysis included scores onthe 5 subscores of Social Provision Scale (Cutrona & Russell, 1987),specifically attachment to co-workers, social integration with co-workers, reassurance of worth by co-workers, reliable alliance withco-workers and guidance offered by co-workers. There was a no maineffect for Employment Status, F(5,624)=.61, ns, and partial η2=.005,power=.22. There was no main effect for Region, F(5,624)=1.31, ns,and partial η2=.01, power=.47. There also was no EmploymentStatus×Region interaction, F(5,624)=1.04, ns, partial η2=.008,power=.37.

3.4. Workers' perceptions of the social provisions offered by supervisors

Differences in perceptions of the social provisions offered bysupervisors were examined using a multivariate analysis of variance(MANOVA)with two between group factors, Employment Status, withtwo levels (stayers, leavers) and Region, with two levels (urban, rural).Dependent variables for the multivariate analysis included scores onthe 5 subscores of Social Provision Scale (Cutrona & Russell, 1987),specifically attachment to supervisors, social integration with super-visors, reassurance of worth by supervisors, reliable alliance withsupervisors and guidance offered by supervisors. There was a maineffect for Employment Status, F(5,602)=2.79, p=.02, and partialη2=.02, power=.83. Prior to examining any of the univariate ANOVAresults, the alpha level was adjusted to α=.01 since five dependentvariables were analyzed. Univariate F-tests showed that employment

status differed on attachment to supervisors, F(1, 411)=10.35, p=.001,and partial η2=.03, power=.89 in that stayers were more attached tosupervisors than leavers. Univariate F-tests showed that employmentstatus differed on guidance provided by supervisors, F(1, 606)=8.68,p=.003, and partial η2=.02, power=.84. Stayers tended to reporthigher levels of guidance offered by supervisors than leavers.

There was a main effect for Region, F(5,602)=4.85, p=.001, andpartial η2=.04, power=.98. Prior to examining any of the univariateANOVA results, the alpha level was adjusted to α=.01 since fivedependent variables were analyzed. Univariate F-tests revealed thatregion differed on attachment to supervisors, F(1, 606)=10.50,p=.001, and partial η2=.02, power=.90. Rural hires tended to reporthigher levels of attachment to supervisors than urban hires. UnivariateF-tests also showed that region differed on reliance on supervisor andguidance from supervisor, F(1, 606)=15.92, p=.001, and partialη2=.03, power=.98; F(1,602)=7.46, p=.006, and partial η2=.01,power=.78; respectively. Rural hires reported higher level of relianceon supervisor than did urban hires. Rural hires also reported receivingmore guidance from their supervisors than urban hires. There also wasno Employment Status×Region interaction, F(5,602)=1.05, ns, partialη2=.009, power=.38. Thus, there was some support for the thirdhypothesis in that supervisor support in the forms of attachment,guidance and reliable alliance did predict retention.

3.5. Workers' perceptions of supervisor quality of work assistance

Differences in workers' perceptions of supervisors' quality of workassistance were explored using a univariate analysis of variance(ANOVA) with two between group factors, Employment Status, withtwo levels (stayers, leavers) and Region, with two levels (urban, rural).The dependent variables of this scale include ratings on four itemsused to measure the supervisors' quality of assistance offered for caseassessments, treatment planning, direct practice with families andunderstanding policies and procedures of the agency. There was noeffect for Employment Status, F(1,601)=3.29, ns, and partial η2=.005,power=.44. There was a main effect for Region, F(1,601)=8.50,p=.004, and partial η2=.02, power=.83. Lastly, therewas no Employ-ment Status×Region interaction, F(1,601)=1.80, ns, and partialη2=.003, power=.27. Results showed that rural hires provided higherratings to the level of assistance offered by their supervisors.

3.6. Workers' view of the workplace environment

Differences in views of the workplace environment perceptionswere explored using a multivariate analysis of variance with twobetween group factors, Employment Status, with two levels (stayers,leavers) and Region, with two levels (urban, rural). Dependentvariables for the multivariate analysis included scores on threehighest reliable subscales of the Herzberg (1966) multi-dimensionjob satisfaction inventory, specifically competent supervision, rela-tionship with supervisor, and salary. There was no main effect forEmployment Status, F(3,620)=2.31, ns, and partial η2= .01,power=.58. There was a main effect for Region, F(3,620)=12.54,p=.001, and partial η2=.06, power=1.00. Lastly, there was noEmployment Status×Region interaction, F(3,620)=1.79, ns, and partialη2=.009, power=.46. Prior to examining any of the univariate ANOVAresults, the alpha level was adjusted to α=.02 since three dependentvariables were analyzed.

Univariate F-tests showed that region differed on competentsupervision, F(1, 622)=29.31, pb .001, and partial η2= .05,power=1.00. Rural hires tended to report higher levels of satisfactionwith the competency of the supervision they received than did urbanhires. Univariate F-tests showed that region differed in relationshipwith supervisor, F(1, 622)=16.51, pb .001, and partial η2=.03,power=.98. Rural hires reported greater satisfaction with theirrelationships with their supervisors than did urban hires. Thus there

Page 6: Individual and organizational factors in job retention in Kentucky's child welfare agency

Table 2Means and standard deviations of the individual and organizational factors for the employment status and region comparisons.

Variables StayersMean (S.D.)

LeaversMean (S.D.)

Employmentstatus

UrbanMean (S.D.)

RuralMean (S.D.)

Region

Personalitya ns nsConscientious 2.81 (.26) 2.83 (.24) ns 2.81(.24) 2.82 (.25) nsNeuroticism 2.80 (.32) 2.82 (.30) ns 2.82 (.29) 2.81 (.31) nsExtraversion 3.76 (.47) 3.76 (.49) ns 3.77 (.47) 3.75 (.49) nsAgreeableness 2.72 (.24) 2.75 (.25) ns 2.76 (.24) 2.71 (.25) nsOpenness to Experience 2.72 (.24) 2.75 (.25) ns 2.76 (.24) 2.71 (.25) ns

Co-worker support a ns nsAttachment 2.79 (.40) 2.74 (.42) ns 2.73 (.43) 2.80 (.40) nsSocial integration 2.18 (.32) 2.23 (.38) ns 2.22 (.36) 2.18 (.34) nsReassurance of worth 2.17 (.32) 2.18 (.36) ns 2.17 (.33) 2.18 (.34) nsReliable alliance 2.53 (.27) 2.55 (.32) ns 2.53 (.30) 2.55 (.28) nsGuidance 2.43 (.29) 2.43 (.29) ns 2.41 (.29) 2.45 (.29) ns

Supervisor supporta ⁎ ⁎⁎

Attachment 2.69 (.45) 2.60 (.45) ⁎ 2.58 (.44) 2.71 (.45) ⁎⁎

Social integration 2.27 (.38) 2.27 (.41) ns 2.28 (.36) 2.26 (.41) nsReassurance of worth 2.12 (.33) 2.15 (35) ns 2.15 (.37) 2.12 (.31) nsReliable alliance 3.03 (.42) 2.97 (.40) ns 2.93 (.39) 3.06 (.42) ⁎⁎

Guidance 2.56 (.34) 2.41 (.29) ⁎⁎ 2.42 (.34) 2.50 (.32) ⁎

Quality of supervisor b 4.19 (.86) 4.04 (.99) ns 4.00 (.94) 4.22 (.87) ⁎⁎

Workplace evaluationa ns ⁎⁎

Competent supervision 4.16 (.97) 3.94 (1.07) ns 3.82 (1.01) 4.26 (.97) ⁎⁎

Relationship with supervisor 4.03 (.90) 3.80 (.96) ns 3.76 (.95) 4.07 (.89) ⁎⁎

Salary 2.54 (1.10) 2.56 (1.12) ns 2.43 (1.07) 2.63 (1.12) nsWorkers' perception of supervisors' training attitudesb 5.91 (1.23) 5.85 (1.25) ns 5.75 (1.19) 5.97 (1.26) nsWorkers' perception of traininga ns nsTraining quality 6.05 (.83) 5.87 (.91) ns 5.88 (.87) 6.05 (.86) nsTrainer quality 5.64 (1.14) 5.53 (1.13) ns 5.47 (1.13) 5.69 (1.13) ns

Bold font indicates a significant finding.⁎Significant at the 0.05 level; ⁎⁎Significant at the 0.005 level.

a Based on repeated measures MANOVA.b Based on repeated measures ANOVA.

552 P.A. Yankeelov et al. / Children and Youth Services Review 31 (2009) 547–554

was no support for the fourth hypothesis that a positive view of thework environment would be positively correlated with retention.

3.7. Workers' perception of supervisors' training attitudes

Differences inworkers' perceptions of supervisors' training attitudeswere explored using a univariate analysis of variance (ANOVA) with twobetween group factors, Employment Status, with two levels (stayers,leavers) and Region, with two levels (urban, rural). Dependent variablesfor this scale included the ratings on 6 items used to measure thesupervisor attitude toward training including general impressionsexpressed related to the practicality, and the potential impact ofwork performance and increased knowledge. There was no maineffect for Employment Status, F(1,572)=.02, ns, and partial η2=.001,power=.05. There was no main effect for Region, F(1, 572)=3.69,ns and partial η2=.006, power=.48. Lastly, there was no Employ-ment Status×Region interaction, F(1,572)=.15, ns, and partialη2=.001, power=.07. Thus, there was no support for the hypothesisthat supervisory support for training predicted retention.

3.8. Workers' perceptions of training

Differences in workers' perceptions of training were examinedusing a multivariate analysis of variance (MANOVA) with two betweengroup factors, Employment Status, with two levels (stayers, leavers)and Region, with two levels (urban, rural). Dependent variables for themultivariate analysis included scores on the measures of trainingquality and trainer ability. There was no main effect for EmploymentStatus, F(2,603)=2.41, ns, and partial η2=.008, power=.49. Therewas no main effect for Region, F(2,603)=3.48, ns, and partial η2=.01,power=.65. Lastly, there was no Employment Status×Region interac-tion, F(2,603)=.81, ns, and partial η2=.003, power=.18. Stayers andleavers and urban and rural workers did not differ in their perceptions

of the quality of training. Table 2 provides a summary of theemployment status and region comparisons for all study variables.

4. Discussion

This study found a high turnover rate among the child welfareworkers in this sample. Approximately 14% left by the end of their firstyear of employment, another 28% left during their second year, and36% left during their third year. This data suggest that workers in childwelfare are at highest risk of leaving during their third year. We havefound similar results in other research in our state as we have trackedthe PCWCP graduates (Barbee et al., in press), although the PCWCPgraduates were retained at higher rates. The PCWCP graduates left at arate of 7% the first year, 14% the second year, 27% the third year and35% the fourth year. Collectively, these studies suggest that interven-tions for retention need to begin early. Intensifying the attention paidto workers early in their tenure in the form of guidance and socialsupport from their supervisor may prevent departure. The PCWCPresults found that after the fourth year, the rate of departure remainedstable suggesting that retention inventions in the first several years ofemployment could be crucial to curb the loss of employees.

The results of this study show that stayers and leavers in this childwelfare agency did not differ in their gender, race, or cumulative G.P.A.MSWs were more likely to leave the agency and employees withmajors other than psychology or social work were more likely to stay.Both stayers and leavers gave similar ratings to the value of thetraining, their trainers' expertise and to their supervisors' attitudetoward training. Those who remained in the agency did not differ intheir personality traits when compared to those who left the agency.This finding is consistent with other studies demonstrating thatpersonality is not predictive of training transfer or commitment to thepublic child welfare agency (Antle, Barbee, & van Zyl, 2008; Barbeeet al., in press).

Page 7: Individual and organizational factors in job retention in Kentucky's child welfare agency

553P.A. Yankeelov et al. / Children and Youth Services Review 31 (2009) 547–554

Perceptions of co-worker support also did not differ among thestayers and the leavers. Previous research by the authors demon-strated a significant positive relationship between co-worker supportand commitment to remain with the agency for up to five years(Barbee et al., in press). This suggests that although co-worker supportpromotes feelings of commitment to the agency, it does not predictactual retention.

A major finding of this study is that those who stayed were moreattached to their supervisors, suggesting that employees felt a sense ofsecurity in their relationships with the supervisors. Research on theeffects of attachment in high stress jobs (such as firefighters, Israelisoldiers) indicates that a supervisor can serve as a secure base,buffering their employees from the trauma of the work they areengaged in (e.g., Davidovitz, Mikulincer, Shaver, Izsak, & Popper,2007). Perhaps, the supervisors in this study buffered their newworkers from the stresses of the job to such an extent that the workerstayed longer in the job. These employees also receivedmore guidancein the form of advice and information from their supervisor,suggesting that their supervisor is helping their employee approachthe problems they encounter on the jobwith a clear sense of direction,information and tangible aid, when necessary. Lower emotionalexhaustion of child welfare workers has been associated with higherretention rates (Drake & Yadama, 1996). Furthermore, Reagh (1994)suggests that a better balance between working with people andcompleting all of the necessary paperwork may lead to higherretention rates. Supervisors can definitely assist their front lineworkers learn effective ways to manage the competing values (asdiscussed by Guterman & Jayaratne, 1994) so that a balance can beachieved between paperwork and their clients for the benefit ofreduced worker emotional exhaustion.

The overall findings about the importance of the supervisoryrelationship supports prior literature describing the importance of theinteraction betweenworker and supervisor (Azar, 2000; CWLA, 2002;Davis-Sacks et al., 1985; Rycraft, 1994; Samantrai, 1992), however, thisstudy measured both the technical and emotional aspects of super-vision and lends support to the need to promote both types ofsupervision for employee retention. These more specific measures ofsupervisory support also lend ideas for the types of peoplewho shouldbe promoted to be a supervisor and to the kinds of training thatsupervisors should receive once they move into this position.

One practice that has been identified in the field to improve bothrecruitment and the retention rate of child welfare workers isimplementing leadership/mentoring programs. (USGAO, 2003). Intheory, leadership/mentoring programs could also provide someknowledge and skill training relief to the supervisor by increasingthe likelihood for balance between technical and emotional super-vision. For example in 1997, the Kentucky's Cabinet for Health andFamily Services (formerly called the Cabinet for Families andChildren), developed a program to train “Field Training Specialists.”These Field Training Specialists provided on-the-job training andongoing reinforcement of a range of casework skills for new employ-ees. This programwas replaced by a coaching andmentoring initiativethat emerged from Kentucky's Program Improvement Plan after itsChild and Family Service Review of 2003. The coaching andmentoringinitiative provided several days of classroom training to all childwelfare supervisors across the state, so that they could provide theneeded casework and relational support to workers in the agency.Research on this project found that supervisors who were trainedspecifically in how to work with their team members on clientengagement, assessment, case planning, interventions, case closureand relational support had workers who were better at these skillsand clients who were less likely to hurt their children again when thechildren were returned to the home and children who were better offpsychologically and physically (Antle et al., 2008). Putting practicessuch as these into place may provide supervisors with the necessaryskills to build relationships and transfer knowledge to their workers.

Partnerships between university schools of social work and statechild welfare agencies is another strategy that has helped in therecruitment and eventual retention of qualified staff (USGAO, 2003;Zlotnik, 2002). Many states that have implemented training programsutilizing Title IV-E funds have evaluated the retention rates of thegraduates of those programs, and found them to positively impactretention (Barbee et al., in press; Brown, Chavkin, & Peterson, 2002;Dickinson & Perry, 2002; Gansle & Ellett, 2002; Jones & Okamura,2000; Robin & Hollister, 2002). These programs may be helpful inretaining workers because workers are more prepared for the job andtheir supervisors can perhaps provide more emotional support andguidance, because they do not have to teach all the technical skillsneeded for the job to the new employee. They havemore time to focuson other ways of support, rather than only teaching them about thejob. The supervisors can be more balanced in their approach of howthey help the new employee because they know more about how todo the job when they come in to the agency. Non-graduates of theseprograms have to learn more about the job up front.

Thefinal groupoffindingsnoted that ruralworkers perceivedgreatersupport from their supervisors than did their urban counterparts. Ruralworkers scored higher than urban workers regarding their attachmentto their supervisor, in their ratings of their ability to rely on theirsupervisor, and rural workers reported received more guidance onassessments, treatment plans and more assistance interpreting policiesthan their urban counterparts. In a similar vein, Sundet and Cowger(1990) found that rural workers, as compared to urban workers,generally did not rate their setting as highly problematic with theexception of issue of professional isolation. It could be speculated thatchildwelfareworkers in rural communities report beingmore attached,report being able to rely on their supervisor to a greater extent andreportmore guidance thanurbanworkers due to the nature of living in asmall town where the probability of having a personal, as well as, aprofessional relationship with one's supervisor is likely. The multi-facetednature of these relationshipsmay inflate thepositive appraisal ofthe rural worker.

The strengths of this study lie in the breadth and depth of theindividual and organizational predictors explored. This is but thefourth study that has used prospective data to explore retention rates.The measures were collected in the second month of the front lineworker's employment to predict their later decision to stay with orleave the agency. Albeit the measures were taken one time and fairlyearly in the employees' employment history, it can be argued that twomonths of employment should provide the worker with a fairlyaccurate perception of their job responsibilities. Furthermore, workersparticipated in a number of training programs in these first twomonths and carried a caseload from their start date. Each of thesefactors suggests that workers would have had a fairly accurateunderstanding of the degree of emotional and technical supportavailable for their casework. Future research could evaluate thesevariables following a longer period of employment to assess whetherthese perceptions change over time.

5. Conclusion

This study has shown that the relationship between supervisor andworker is extremely importantwhen dealingwith the issue of retentionin child welfare. Having a supportive supervisor who is believed to becompetent is a key factor in staying on the job. Longitudinal studies areneeded in the field of child welfare to more adequately assess theretention rates of child welfare workers by state and nationally. Thiswould give amore adequate picture of the field and how it is affected byturnover. Also, more in-depth exit interviews, that give insight intoreasons why people leave, would be helpful to gain insight into issuesaffecting turnover, and to help in the prevention of workers leaving. Inaddition, more data are needed on the impact of supervisors and teamsonworker retention. In particular, future research should use measures

Page 8: Individual and organizational factors in job retention in Kentucky's child welfare agency

554 P.A. Yankeelov et al. / Children and Youth Services Review 31 (2009) 547–554

that pick up on very specific behaviors of supervisors so as to betterunderstand exactly how supervisors can behave in away that facilitatesgrowth and satisfaction in workers. Other methods of inquiry are alsoneeded. Observations of the interactions between supervisors and eachworker on their team could lead to a greater understanding of thetransactions that take place that are helpful to workers as they navigatesuch a difficult job. These types of research studies will lead tointerventions that will create a more experienced child welfareworkforce that feels more supported in their environment, which willin turn lead to better outcomes for the children and families beingserved by public child welfare agencies.

Acknowledgments

The authors wish to thank Kentucky's Cabinet for Health andFamily Services for their support in implementing this research.

References

Antle, B. F., Barbee, A. P., & van Zyl, M. A. (2008). A comprehensive model for childwelfare training evaluation. Child and Youth Services Review, 30(9), 1063−1080.

Azar, S. T. (2000). Preventing burnout in professionals and paraprofessionals who workwith child abuse and neglect cases: A cognitive behavioral approach to supervision.Journal of Clinical Psychology/In Session: Psychotherapy in Practice, 56(5), 643−663.

Balfour, D. L., & Neff, D. M. (1993). Predicting and managing turnover in human serviceagencies: A case study of an organizational crisis. Public Personnel Management, 22(3), 473−486.

Barbaranelli, C., Caprara, G. V., & Maslach, C. (1997). Individuation and the Five FactorModel of personality traits. European Journal of Psychological Assessment, 13(2),75−84.

Barbee, A.P., Sullivan, D.J., Antle, B.F., Moran, E.B., Hall, J.C. & Fox, S. (in press). The PublicChild Welfare Certification Program: Worker retention and impact on practice.Journal of Social Work Education.

Brown, J. K., Chavkin, N. F., & Peterson, V. (2002). Tracking process and outcomes resultsof BSW students' preparation for public child welfare practice: Lessons learned.Journal of Health and Social Policy, 15(3/4), 105−116.

Cahalane, H., & Sites, E.W. (2004). Is it hot or cold? The climate of child welfareemployee retention. Unpublished manuscript, University of Pittsburgh.

Child Welfare League of America (September 2002). Child welfare workforce. Research toPractice : Author.

Conrad, D., & Kellar-Guenther, Y. (2006). Compassion fatigue, burnout and compassionsatisfaction among Colorado child protection workers. Child Abuse and Neglect, 30,1071−1080.

Curry, D., McCarragher, T., & Dellman-Jenkins, M. (2005). Training, transfer andturnover: Exploring the relationship among transfer of learning factors and staffretention in child welfare. Children and Youth Services Review, 27, 931−948.

Cutrona, C. E., & Russell, D. W. (1987). The provisions of social relationships andadaptation to stress. In W. H. Jones, & D. Perlman (Eds.), Advances in PersonalRelationships, vol. (37–67). (pp. )Greenwich, CT: JAI Press.

DePanfilis, D. (2006). Compassion fatigue, burnout, and compassion satisfaction:Implications for retention of workers. Child Abuse and Neglect, 30, 1067−1069.

Davidovitz, R., Mikulincer, M., Shaver, P., Izsak, R., & Popper, M. (2007). Leaders asattachment figures: Leaders' attachment orientations predict leadership-relatedmental representations and followers' performance and mental health. Journal ofPersonality and Social Psychology, 93, 632−650.

Davis-Sacks, M. L., Jayaratne, S., & Chess, W. A. (1985). A comparison of the effects ofsocial support on the incidence of burnout. Social Work, 30(3), 240−244.

Drake, B., & Yadama, G. N. (1996). A structural equation model of burnout and job exitamong child protective service workers. Social Work Research, 20(3), 179−187.

Dickinson,N. S., & Perry, R. E. (2002). Factors influencing the retentionof specially educatedpublic child welfare workers. Journal of Health and Social Policy, 15(3/4), 89−103.

Dhooper, S. S., Royse, D. D., & Wolfe, L. C. (1990). Does social work education make adifference? Social Work, 35(1), 57−61.

Ellett, A. J. (2000). Human caring, self-efficacy beliefs, and professional organizationalculture correlates of employee retention in child welfare. Baton Rouge, LA: LouisianaState University and Agricultural and Mechanical College.

Ellett, A. J., Ellett, C. D., & Rugutt, J. K. (2003). A study of personal and organizationalfactors contributing to employee retention and turnover in child welfare in Georgia:Executive summary and final project report. Athens, GA: University of Georgia Schoolof Social Work.

Ellett, A., Collins-Carmargo, C., & Ellett, C. (2006). Personal and organization correlatesof outcomes in child welfare: Implications for supervision and continuingprofessional development. Professional Development, 9(2/3), 44−53.

Gansle, K. A., & Ellett, A. J. (2002). Child welfare knowledge transmission, practitionerretention, and university–community impact: A study of Title IV-E child welfaretraining. Journal of Health and Social Policy, 15(3/4), 69−88.

Goldberg, L. R. (1992). The development of markers for the Big-Five factor structure.Psychological Assessment, 4, 26−42.

Graef, M. I., & Hill, E. L. (2000). Costing child protective services staff turnover. ChildWelfare, 79(5), 517−533.

Guterman, N. B., & Jayaratne, S. (1994). Responsibility at-risk: Perceptions of stress,control and professional effectiveness in child welfare direct practitioners. Journalof Social Service Research, 20(1/2), 99−120.

Harris, N., Middleton, S., Byrnes, E., Tollefson, D., Sahami, S., & Berry-Johnson, S. (2000).DCFS turnover study 2000. Salt Lake City, UT: University of Utah Graduate School ofSocial Work.

Herzberg, F. (1966). Work and the nature of man. Cleveland, Ohio: The World ofPublishing, Co.

Jayaratne, S., & Chess, W. A. (1984). Job satisfaction, burnout, and turnover: A nationalstudy. Social Work, 29(5), 448−453.

Jones, L. P., & Okamura, A. (2000). Reprofessionalizing child welfare services: Anevaluation of a Title IV-E training program. Research on Social Work Practice, 10(5),607−621.

Koeske, G. F., & Kirk, S. A. (1995). Direct and buffering effects of internal locus of controlamong mental health professionals. Journal of Social Service Research, 20, 1−28.

Landsman, M. J. (2001). Commitment in public child welfare. Social Service Review, 75(3),386−420.

Lieberman, A. A., Hornby, H., & Russell, M. (1988). Analyzing the educationalbackgrounds and work experience of child welfare personnel: A national study.Social Work, 33(6), 485−489.

Mor Barak, M. E., Nissly, J. A., & Levin, A. (2001). Antecedents to retention and turnoveramong child welfare, social work, and other human service employees: What canwe learn from past research? A review and metanalysis. Social Service Review, 75(4), 625−662.

Nissly, J. A., Mor Barak, M. E., & Levin, A. (2005). Stress, support, and workers' intentionsto leave their jobs in public child welfare. Administration in Social Work, 29(1),79−100.

Pew Commission on Children in Foster Care (2004). Fostering the future: Safety,permanence and well-being for children in foster care. : Author.

Reagh, R. (1994). Public child welfare professional—those who stay. Journal of Sociologyand Social Welfare, 21(3), 69−79.

Robin, S. C., & Hollister, C. D. (2002). Career paths and contributions of four cohorts of IV-Efunded MSW child welfare graduates. Journal of Health and Social Policy, 15(3/4),53−67.

Rosenthal, J. A., McDowell, E., & White, T. L. (1998). Retention of child welfare workers inOklahoma. Norman, OK: University of Oklahoma School of Social Work.

Rosenthal, J. A., & Waters, E. (2006). Predictors of child welfare worker retention andperformance: Focus on Title IV-E-funded social work education. Journal of SocialService Research, 32(3), 67−85.

Rycraft, J. (1994). The party isn't over: The agency role in the retention of public ofpublic child welfare caseworkers. Social Work, 39(1), 75−80.

Salgado, J. F. (2002). The big five personality dimensions and counterproductivebehaviors. International Journal of Selection and Assessment, 10(1/2), 117−125.

Samantrai, K. (1992). Factors in the decision to leave: Retaining social workers withMSWs in public child welfare. Social Work, 37(5), 454−459.

Smith, B. (2005). Job retention in child welfare: Effects of perceived organizationalsupport, supervisor support, and intrinsic job value. Children and Youth ServicesReview, 27, 153−169.

Sundet, P. A., & Cowger, C. D. (1990). The rural community environment as a stressfactor for rural child welfare workers. Administration in Social Work, 14(3), 97−110.

United States General Accounting Office (2003). Child welfare: HHS could play a greaterrole in helping child welfare agencies recruit and retain staff. Report GAO 03-357:Washington, DC.

University of Arkansas at Little Rock School of Social Work (2002). MS South Division ofChildren and Family Services recruitment and retention study: FSW survey Little Rock,AR: Author.

University of Arkansas at Little Rock School of Social Work (2002). MS South Division ofChildren and Family Services recruitment and retention study: Supervisor survey.Little Rock, AR: Author.

Viswesvaran, C., & Oanes, D. S. (2000). Perspectives on models of job performance.International Journal of Selection and Assessment, 8(4), 216−226.

Weaver, D., Chang, J., Clark, S., & Rhee, S. (2007). Keeping public child welfare workerson the job. Administration in Social Work, 31(2), 5−25.

Wehrmann, K. C., Shin, H., & Poertner, J. (2002). Transfer of training: An evaluationstudy. Journal of Health and Social Policy, 15(3/4), 23−37.

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

Zlotnik, J. L. (2002). Preparing social workers for child welfare practice: Lessons froman historical review of the literature. Journal of Health and Social Policy, 15(3/4),5−21.

Zlotnik, J. L., DePanfilis, D., Daining, C., & Lane, M. (2005). Factors Influencing Retention ofChild Welfare Staff: A Systematic Review of Research. Washington, DC: Institute forthe Advancement of Social Work Research.