social work intention

Upload: gadowl

Post on 13-Apr-2018

221 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/27/2019 Social Work Intention

    1/23

    This article was downloaded by: [76.98.75.159]On: 28 September 2013, At: 16:07Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

    Social Work in Health CarePublication details, including instructions for authors andsubscription information:

    http://www.tandfonline.com/loi/wshc20

    Supervisory Communication, Burnout,

    and Turnover Intention Among SocialWorkers in Health Care SettingsHansung Kim Phd

    a& Sun Young Lee PhD

    b

    a

    Department of Social Work, California State University, Fullerton,California, USAbSchool of Journalism and Communication, Kyung Hee University,

    Seoul, Republic of Korea

    Published online: 24 Apr 2009.

    To cite this article:Hansung Kim Phd & Sun Young Lee PhD (2009) Supervisory Communication,

    Burnout, and Turnover Intention Among Social Workers in Health Care Settings, Social Work in Health

    Care, 48:4, 364-385, DOI: 10.1080/00981380802598499

    To link to this article: http://dx.doi.org/10.1080/00981380802598499

    PLEASE SCROLL DOWN FOR ARTICLE

    Taylor & Francis makes every effort to ensure the accuracy of all the information (theContent) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

    This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &

    Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

    http://www.tandfonline.com/page/terms-and-conditionshttp://dx.doi.org/10.1080/00981380802598499http://www.tandfonline.com/page/terms-and-conditionshttp://www.tandfonline.com/page/terms-and-conditionshttp://dx.doi.org/10.1080/00981380802598499http://www.tandfonline.com/action/showCitFormats?doi=10.1080/00981380802598499http://www.tandfonline.com/loi/wshc20
  • 7/27/2019 Social Work Intention

    2/23

    364

    Social Work in Health Care, 48:364385, 2009Copyright Taylor & Francis Group, LLCISSN: 0098-1389 print/1541-034X onlineDOI: 10.1080/00981380802598499

    WSHC0098-13891541-034XSocial Work in Health Care, Vol. 48, No. 4, Mar 2009: pp. 00Social Work in Health Care

    Supervisory Communication, Burnout,and Turnover Intention Among Social

    Workers in Health Care Settings

    Supervisory CommunicationH. Kim and S. Y. Lee

    HANSUNG KIM, PhDDepartment of Social Work, California State University, Fullerton, California, USA

    SUN YOUNG LEE, PhDSchool of Journalism and Communication, Kyung Hee University, Seoul, Republic of Korea

    The current study tests the effects of different types of supervisory com-

    munication on burnout and turnover intention among health socialworkers. The study proposed a conceptual model of supervisory com-munication and tested it empirically using structural equation mod-

    eling (SEM) techniques with a random sample of 211 Californiastate-registered social workers working in health or mental health caresettings. The results of the present study provide empirical evidence of

    the unique roles that different types of supervisory communication playas antecedents of burnout and turnover intention. Specifically, sup-portive relationship communication had an indirect effect on burn-

    out and turnover intention through its effect on perceived stress,whereas job-relevant communication had not only an indirect effecton burnout and turnover intention through its effect on stress, but also

    a direct effect on turnover intention. In addition, the results showedthat upward communication moderated the relationship between

    stress and burnout. Implications for social work administration and

    possible elaboration of the theoretical framework are discussed.

    KEYWORDS supervisory communication, burnout, turnoverintention, social workers in health care

    Received January 29, 2007; accepted September 25, 2007.This research was supported in part by a research award from the Hamovitch Center for

    Science in the Human Services and by an Albert and Frances Feldman Endowed Fellowshipfrom the USC School of Social Work.

    Address correspondence to Sun Young Lee, School of Journalism and Communication,Kyung Hee University, Hoegi-Dong, Dongdaemun-Gu, Seoul 130-701, Republic of Korea.

    E-mail: [email protected]

  • 7/27/2019 Social Work Intention

    3/23

    Supervisory Communication 365

    BACKGROUND

    Social work delivers vital health services to a vast array of populations inneed. Areas of practice within health social work include health promotion,

    general health care, HIV/AIDS (Hall, 2007), disease management (Claiborne &Vandenburgh, 2001), mental health care (Cohen, 2003), and health care withthe elderly (Berkman, Gardner, Zodikoff, & Harootyan, 2005; Chong, 2007).In these fields, social workers play unique and essential roles that includedischarging planning (Lechman & Duder, 2006) and resolving ethical dilem-mas in hospice (Csikai, 2004). Given the current emphasis on health carecost control, productivity, and accountability, social workers job attitudesand job performance in health care settings has become an importantresearch topic.

    In a recent report on the difficulties of the social work profession in gen-

    eral, examples of job demands included increasing paperwork and caseloadsas well as staff shortages and reduced availability of adequate supervision(Center for Workforce Studies, 2006). Social workers in health care settingsface even more demanding situations because of the increased role of man-aged care (Egan & Kadushin, 2004). Previous literature on burnout has sug-gested that demanding job conditions are significant antecedents of socialworker burnout (Sderfeldt, Sderfeldt, & Warg, 1995). Social worker burnoutis a critical managerial issue because social worker burnout negatively affectsthe quality, consistency, and quality of client services. A review of burnout lit-erature suggests that worker burnout increases levels of depression (Glass &McKnight, 1996), sick-leave absences (Toppinen-Tanner, Kalimo, & Mutanen,2005), turnover intention (Huang, Chuang, & Lin, 2003), actual turnover(De Croon, Sluiter, Blonk, Broersen, & Frings-Dresen, 2004), and self-reportedhealth problems (Schaufeli & Enzmann, 1998). Although it is generally agreedthat preventing burnout is important, only a few empirical studies have exam-ined what alleviates burnout among social workers in health care settings.

    Considering that social work is done in the context of supervisory rela-tionships, effective supervision is essential to preventing burnout and turn-over. Social work supervisors play the role of teacher, enabler, consultant,

    and manger for frontline social workers (Kadushin, 2002). An effectivesupervisor must provide a frontline social worker with essential guidelines,professional skills, and knowledge related to services while also communi-cating with social workers about opinions, feelings, and decision makingwith mutual trust and respect. This notion has been supported by studiesreporting that perceived supervisory support, which is considered an out-come of supervisory communication (Ray & Miller, 1991), is associated withlevels of burnout and turnover intention among those being supervised (Um &Harrison, 1998; Mor Barak, Nissly, & Levin, 2001).

    Several explanations have been offered as to why supervisory commu-

    nication helps prevent burnout and turnover. First, it reduces workers role

  • 7/27/2019 Social Work Intention

    4/23

    366 H. Kim and S. Y. Lee

    stress (i.e., role ambiguity and role conflict) (Miller, Ellis, Zook, & Lyles,1990; Ray & Miller, 1991). Because role stress often leads to burnout orturnover (Cordes & Dougherty, 1993; Mor Barak, Nissly, & Levin, 2001),supervisory communication indirectly influences worker burnout and turn-

    over through its effect on perceived role stress. Second, when an individualis experiencing a high level of role stress, supervisory communication is animportant coping resource. Several studies have suggested that supervisorysupport buffers the effect of role stress on burnout (Bakker, Demerouti, &Euwema, 2005; Kickul & Posig, 2001). Despite the significant contributionof many studies, however, few have examined the different effects ofsupervisory communication on burnout and turnover intention amonghealth care social workers in particular. The current study examines (1) theindirect effects of supervisory communication on burnout and turnoverintention resulting from perceived role stress and (2) the moderating effects

    of supervisory communication on the relationship between role stress andburnout.

    Supervisory communication can be described using certain characteris-tics, including the content of communication and the direction of communi-cation flow (Krone, Jablin, & Putnam, 1987). This study focuses on threespecific types of supervisory communication: job-relevant communication,upward communication, and positive relationship communication. A con-ceptual model hypothesizing the relationships among the three types ofsupervisory communication, role stress, burnout, and turnover intention isproposed and empirically examined through Structural Equation Modeling(SEM) techniques using data from a sample of California-registered socialworkers in health care settings.

    LITERATURE REVIEW

    Burnout and Turnover Intention

    When it was first introduced, the term burnout referred to a phenomenonthat was observed among human service workers who had to deal with

    emotionally demanding individuals. Since then, the majority of burnoutstudies have been based on Maslach and Jacksons (1986) conceptualizationof burnout as having three components: emotional exhaustion (feelings ofbeing overextended and depleted of emotional and physical resources),depersonalization or cynicism (negative or excessively detached responsesto various aspects of the job), and diminished personal accomplishment(feelings of incompetence and a lack of achievement at work).

    Theoretical frameworks of burnout (e.g., Cordes & Dougherty, 1993;Demerouti, Bakker, Nachreiner, & Schaufeli, 2001) have placed burnout asa key mediator of the relationship between chronic job stress and various

    strain outcomes. Research has consistently shown that a workers level of

  • 7/27/2019 Social Work Intention

    5/23

    Supervisory Communication 367

    emotional exhaustion is greatly affected by the nature and intensity of stressin the work environment (Cordes & Dougherty, 1993). A worker who isexperiencing higher levels of emotional exhaustion is more likely to havedepersonalized attitudes toward his or her clients and lack a sense of per-

    sonal accomplishment at work (Cordes & Dougherty, 1993; Leiter &Maslach, 1988). This view of how burnout develops explains how job stressresults in higher levels of not only emotional exhaustion but also deperson-alization and diminished personal accomplishment. Comprehensive reviewsof the burnout literature suggest that social workers are more likely to feelburned out when they perceive higher levels of job stress characterized ashigh role conflict, role ambiguity, and role overload (Sderfeldt, Sderfeldt, &Warg, 1995). Among strain outcomes is turnover intention, which severalstudies have empirically supported as a key outcome of burnout (Harrington,Bean, Pintello, & Mathews, 2001; Huang, Chuang, & Lin, 2003). Therefore, it

    is expected that role-related job stress (i.e., role stress), characterized ashigh levels of role conflict, role ambiguity, and role overload, is positivelyassociated with burnout, and that burnout is positively associated with turn-over intention among social workers.

    Types of Supervisory Communication

    Based on Miles, Patrick, and Kings (1996) discussion, this study focuses onthree major types of supervisory communication: job-relevant communica-tion, upward communication, and positive relationship communication.

    JOB-RELEVANTCOMMUNICATION

    Social work uses technologies that are not highly determinate and rely onthe discretion of frontline workers (Patti, 2000). The general characteristicsof social work amplify the importance of the supervisors roles as a teacheror instructor for frontline social workers. Supervisory communicationregarding job-related matters (i.e., job relevant communication) embracesperformance feedback, information about rules and policies, work sched-

    ules and assignments, task-specific instructions, and goals (Miles, Patrick, &King, 1996). Previous studies have suggested that effective job-relevantcommunication between supervisees and their supervisors can increaseprofessional knowledge or skill levels (Albrecht & Adelman, 1987),improve role definition, and raise feelings of accomplishment at work (Ellis& Miller, 1994) and job performance among supervisees (Kluger & DeNisi,1996). In his national study of social work supervision, Kadushin (1992)also found that the most useful function of social work supervision waseducational. Job-relevant communication with supervisors has even beenfound to be the key predictor of social workers performance (York &

    Denton, 1990).

  • 7/27/2019 Social Work Intention

    6/23

    368 H. Kim and S. Y. Lee

    UPWARDCOMMUNICATION

    Upward communication is defined as information exchange in upwardtransmission (from lower to higher members in organizational hierarchy)(Roberts & OReilly, 1974). The literature on social work supervision has

    emphasized that the relationship between supervisor and social workersshould be less hierarchical and more one of mutual influence (Hbert, 1992).When supervisors and frontline workers have mutual trust, respect, and rap-port, the workers can more readily communicate their opinions and feelingsupward. Upward communication from social workers to their supervisors isconsidered a key characteristic of an open communication relationship inwhich both parties perceive the other to be a willing and receptive listener(Jablin & Krone, 1994). For example, Bolon (1995) argued that successfulsupervisors rely on upward communication by creating a climate of trustand respect with supervisees, and Glauser (1984) observed that upwardcommunication can help mangers and supervisors to recognize areas inneed of change and adjustment. At the same time, upward communicationhelps workers to have higher levels of job satisfaction and work perfor-mance, so social workers who enjoy open communication relationshipswith supervisors will be more likely to perceive their supervisory relation-ships as participatory and trustworthy (Eisenberg & Witten, 1987).

    POSITIVERELATIONSHIPCOMMUNICATION

    Positive relationship communication between a supervisor and a socialworker occurs when supervisors interact with social workers in an informaland supportive manner. Supportive personal relationships lead workers tobelieve that they are valued by others, and having a supportive supervisorhas been identified as an important condition for lowering levels of jobstress. For example, evidence suggests that having a supportive supervisoris significantly related to higher levels of job satisfaction (Newsome & Pillari,1991; Rauktis & Koeske, 1994) and that perceived supervisory support inthe work place decreases the likelihood of worker burnout (Houkes,Janssen, de Jonge, & Bakker, 2003) and turnover intention (Mor Barak et al.,

    2001).

    The Role of Supervisory Communication in InformationProcessing and Reducing Uncertainty

    Supervisory communication serves a critical role in models of health socialworker burnout in terms of reducing the perceived levels of job stress.Miller et al. (1990) discussed social information processing theory (Salancik &Pfeffer, 1978) and uncertainty reduction theory (Berger & Calabrese, 1975)

    as theoretical foundations with which to explain the role of supervisory

  • 7/27/2019 Social Work Intention

    7/23

    Supervisory Communication 369

    communication in the process of burnout. First, the social information pro-cessing theory considers perceptions of the workplace to be a function ofthe communication environment in which workers are embedded, ratherthan a function of the objective characteristics of jobs and needs of workers

    (Salancik & Pfeffer, 1978). Beehr (1995) suggested that supervisor contact islikely to consist of instrumental support and affect the conditions that leadto job stress. Similarly, Leiter and Maslach (1988) argued that a major part ofrole stress comes from the extent to which workers perceive supervisors tobe making irreconcilable demands on their time and resources. Thus, job-relevant communication between a supervisor and social worker is associ-ated with social workers perceived levels of role stress.

    Upward communication and positive relationship communication isalso related to social workers role stress. Based on uncertainty reductiontheory (Berger & Calabrese, 1975), Albrecht and Adelman (1987) discussed

    the role of social support in reducing uncertainty in organizational contexts.Schuler and Jackson (1986) stated that the more participation allowedemployees, the more likely they will be able to get a clear understanding ofwhat is expected and what is rewarded. Consequently, the less uncertaintyand stress there is for the individual (as cited in Miller et al., 1990, 215).Miller et al. (1990) found that supervisory support reduced perceived rolestress and the perceptions of workloads among permanent staff in hospitals.Therefore, the current study hypothesizes that social workers attitudes andperceptions of the role-related stress, such as role ambiguity, role conflict,and role overload, will be a function of communication with the immediatesupervisor.

    Supervisory Communication as Empowermentand as a Job Resource

    Models of job demands and resources (JD-R) (Demerouti et al., 2001) pro-vide theoretical frameworks concerning why certain resource factors arecritical in explaining the relationship between role stress and burnout. Thecentral theme of the JD-R model of burnout is that burnout develops when

    job demands are high and when job resources are limited because suchnegative working conditions lead to energy depletion and undermineemployees motivation, respectively (Demerouti et al., 2001, 499). Jobresources are defined as physiological, psychological, social, or organiza-tional aspects of the job that (a) are functional in achieving work goals,(b) reduce job demands and the associated physiological and psychologicalcosts, or (c) stimulate personal growth and development (Demerouti et al.,2001, 501). According to the JD-R model, when the working environmentlacks resources, social workers cannot cope with the negative influences ofrole stress. Recently, Bakker et al. (2005) reported that job demands (e.g.,

    work overloads, emotional demands, unfavorable working conditions, and

  • 7/27/2019 Social Work Intention

    8/23

    370 H. Kim and S. Y. Lee

    workhome interference) had a weaker or no relationship with burnout incases when job resources (e.g., social support, high quality of relationshipwith the supervisor, and performance feedback) were available in theworking environment. Based on the JD-R model of burnout, this study con-

    sidered effective supervisory communication as job condition that not onlyreduces perceived role stress, but is also helpful in coping with perceivedrole stress.

    Empowerment theory is another theoretical framework that explainsthe moderating effects of supervisory communication on the relationshipbetween role stress and burnout. Conger and Kanungo (1988) suggestedthat supervisory practices that include expressing confidence in subordi-nates, fostering opportunities for subordinates to participate in decisionmaking, and providing autonomy from bureaucratic constraint are morelikely to be empowering. This study considers intrinsically rewarding job

    characteristics such as open communication between social workers andtheir supervisors as key elements of empowerment strategies and practicesfor social workers. Open communication with immediate supervisors allowsocial workers to access resources, information, and support in the organi-zation to get the job done successfully; in other words, enhancing the moti-vating potential of jobs by allowing frontline social workers to participate indecision making empowers social workers and intensifies affective reactionstoward the job. Upward communication, job-relevant communication, andsupportive relationship communication can be also empowerment factorsfor social workers. For example, Eby, Freeman, Rush, and Lance (1999)used SEM based on meta-analytic correlations to identify performance feed-back and supervisory satisfaction as key intrinsic motivators. Therefore, thecurrent study hypothesizes that social workers with high quality of supervi-sory communication in terms of job-relevant communication, upward com-munication, and positive relationship communication will be less likely toburn out, compared to social workers who do not.

    A Proposed Hypothetical Model

    Based on the literature review, the following relationships between supervisorycommunication, role stress, burnout, and turnover intention are hypothesized:

    Hypothesis 1: Role stress will be positively associated with burnout in socialworkers.

    Hypothesis 2: Burnout will be positively associated with turnover intentionin social workers.

    Hypothesis 3: Supervisory communication (i.e., job-relevant communication:H3-a, upward communication: H3-b, positive relationshipcommunication: H3-c) will be negatively associated with role

    stress in social workers.

  • 7/27/2019 Social Work Intention

    9/23

    Supervisory Communication 371

    Hypothesis 4: Supervisory communication (i.e., job-relevant communication:

    H4-a, upward communication: H4-b, positive relationshipcommunication: H4-c) will moderate the relationshipbetween role stress and burnout in social workers.

    To test these hypotheses, a hypothetical model was developed. Figure 1illustrates the hypothesized relationships between key constructs.

    Several studies have identified the importance of demographic factorsin explaining worker burnout and turnover intention. For example, Brewerand Shapards (2004) meta-analytic review of burnout studies reports age asnegatively correlated to worker burnout, and Jackson (1993) found that

    gender is a significant demographic factor of burnout (Brewer & Shapard,2004). Other studies have suggested that workers with shorter organiza-tional tenure (Somers, 1996) and higher dissatisfaction with salary (Currall,Towler, Judge, & Kohn, 2005) are more likely to quit their jobs. Therefore,in order to develop a more valid and complete model of turnover intention,age, gender, organizational tenure, and annual salary have been included inthe hypothetical model as control variables.

    METHODOLOGYSample and Procedure

    For the studys cross-sectional survey design, 1500 registered social workerswere randomly selected from the total population of state-registered socialworkers working in California (n = 21,518) in 2005. Each of these 1500social workers was mailed a survey instrument, an introductory letter, aninformation sheet, and a return postage-paid envelope. Of the 529 question-naires returned, 51 were ineligible because respondents were currentlyretired or working in other fields (43), registered after the data-collection

    period (5), or failed to answer a significant number of questions (3). From

    FIGURE 1 A proposed conceptual model of supervisory communication, role stress, burnout,and turnover intention.

    Supervisory

    Communication

    Role Stress Burnout Turnover

    Intention

  • 7/27/2019 Social Work Intention

    10/23

    372 H. Kim and S. Y. Lee

    the original sample of 1500, then, 478 questionnaires were eligible, yieldinga survey response rate of 32%. Among eligible participants, 405 social work-ers were working full time or part time in organizational settings. A sub-sample of 211 full-time social workers in health care or mental health care

    settings was also used.

    Measures

    SUPERVISORYCOMMUNICATION

    Communication with ones supervisor was measured by a short form (24-itemversion) of Huseman, Hatfield, Boulton, and Gatewoods (1980) original56-item instrument representing eight message types: feedback, rationale,information, direction, negative expression, positive expression, participa-

    tion, and upward openness. Each question was measured with a five-pointLikert-type scale. Miles, Patrick, and King (1996) performed exploratory fac-tor analysis for this instrument and identified four types of supervisory com-munication: job-relevant communication, upward communication, positiverelationship communication, and negative relationship communication. Inthe present study, three items of negative relationship communication wereexcluded because social workers (n= 48) who participated in a pilot studysuggested that these items were not appropriate to research in social work.These three items are My supervisor ridicules or makes fun of me, Mysupervisor criticizes my work in front of others, and My supervisor is criti-

    cal of me as a person. After omitting these items, a principal component fac-toring analysis with varimax rotation was performed for the modified 21-itemversion of Huseman et al.s measure. Results suggested three sub-factors (i.e.,job-relevant communication, upward communication, positive relationshipcommunication) structure of the measure, which were consistent with Mileset al.s 1996 study. Job-relevant communication consisted of ten items (e.g.,My supervisor gives clear instructions to me and My supervisor lets meknow why changes are made in work assignments). Positive relationshipcommunication consisted of seven items (e.g., My supervisor strikes upcasual conversations with me). Upward communication consisted of three

    items (e.g., I question my supervisors instructions when I dont understandthem and I tell my supervisor when I think things are being done wrong).Cronbachs alphas for this study sample were 0.95 for the job-relevant com-munication scale, 0.92 for the positive relationship communication scale, and0.77 for the upward communication scale.

    ROLESTRESS

    Role stress was assessed by measuring three role-related stressors: role conflict,

    role ambiguity, and role overload. Role conflict (RC) and role ambiguity (RA)

  • 7/27/2019 Social Work Intention

    11/23

    Supervisory Communication 373

    were measured using a shortened form of the Rizzo, House, and Lirtzman(1970) role conflict/role ambiguity questionnaire. The role conflict scaleconsisted of eight items (e.g., I receive an assignment without the man-power to complete it) and included questions regarding: (1) conflict

    between the respondents internal standards and the defined role behavior;(2) conflict between the time, resources, or capabilities of the respondentand the defined role behavior; (3) conflict between several roles for thesame person that require different behaviors; and (4) conflict betweenexpectations and organizational demands in the form of conflicting requestswith incompatible standards of evaluation. Role ambiguity consisted of sixitems (e.g., I know what my responsibilities are). Participants were asked torespond to the six-item RA and eight-item RC scales by indicating on a 7-pointscale ranging from very false (1) to very true (7) the degree to which thecondition applied to them. Cronbachs alpha for the internal consistency

    and reliability for this sample was 0.88 for the RC scale and 0.84 for the RAscale. Role overload was measured by three items from the five-item scaleof work overload, which were specifically designed to measure workloadamong human service workers (Lait & Wallace, 2002): (1) I have to attendtoo many meetings in this job, (2) My job involves a lot of paperwork,and (3) I have to work very fast to get everything done in my job. Thisscale also employed a 7-point Likert-type scale ranging from stronglydisagree (1) to strongly agree (7). The Cronbachs alpha for its internalconsistency and reliability was 0.60.

    BURNOUT

    Burnout among social workers was assessed using Maslachs Burnout Inven-toryHuman Service Survey (MBIHSS; Maslach & Jackson, 1986), whichmeasures burnout components by asking about the frequency with whichworkers experience feelings related to each aspect of the burnout syndrome.The survey included nine questions on perceived emotional exhaustion, fivequestions on depersonalization, and eight questions on personal accomplish-ment (Maslach & Jackson, 1986). Each statement was rated on a 7-point con-

    tinuum from never experienced (0) to experienced every day (6). For asample of social service workers, the MBI showed internal reliability (Cronbachsalpha ranging from 0.71 to 0.90) and testretest reliability (2- to 4-week intervalsfor all scales, ranging from 0.60 to 0.82) (Maslach & Jackson, 1986).Cronbachs alpha for the current study was 0.91 for emotional exhaustion,0.75 for depersonalization, and 0.79 for personal accomplishment.

    TURNOVERINTENTION

    Organizational turnover intention was measured by three items from the

    four-item scale of intention to leave (Nissly, Mor Barak, & Levin, 2005):

  • 7/27/2019 Social Work Intention

    12/23

    374 H. Kim and S. Y. Lee

    (1) In the next few months, I intend to leave this organization, (2) In thenext few years, I intend to leave this organization, and (3) I occasionallythink about leaving this organization. These items were rated on a 7-pointLikert-type scale ranging from strongly disagree (1) to strongly agree (7),

    for which the Cronbachs alpha was 0.76.

    CONTROLVARIABLES

    The questionnaire included questions on respondents age, gender, organi-zational tenure, and annual salary. Organizational tenure was determinedby the number of years the respondent has worked for his or her currentemployer, and annual salary was based on respondents total yearly incomefrom their current jobs.

    Analysis

    Missing values for all items in the standardized scales (e.g., MBIHSS) werereplaced by plausible values based on the multiple imputation methodusing the Windows freeware, NORM (Schafer, 1999). Following the methoddescribed by Olsen and Schafer (1998), we created three imputed data setsand obtained a single-point estimate by averaging across the estimates fromthe data sets. However, demographic variables still have missing values:1.6% of the samples were missing an age, 1.9% were missing organizationaltenure, and 3.1% were missing annual wage. In order to handle these miss-ing values with no loss of information in the available dataset, we used theFull Information Maximum Likelihood (FIML) Method (Arbuckle, 1996). Inorder to use the FIML method, the data must be multivariate normal. Skew-ness ranged between 0.30 and 1.46, and kurtosis ranged between 0.96and 1.77, so practically normal distributions were assumed for all variables.

    After the preliminary step, structural equation analyses using the Mplus4.0 program (Muthn & Muthn, 2006) were conducted with a FIMLmethod. First, composite scores for three types of supervisory communica-tion, three kinds of role stressors, three dimensions of burnout, and turn-

    over intention were calculated. Second, as suggested by Anderson andGerbing (1988), a confirmatory factor analysis (CFA) was performed to test ameasurement model that included two latent variables, role stress and burn-out. Once the measurement model was estimated and convergent, andproper solution was derived, a third step assessed a series of structuralequation models based on the proposed conceptual model. To evaluatemodel fit, the c2/df, comparative fit index (CFI), and the Root-Mean-SquareError of Approximation (RMSEA) were examined. A value of c2/dfless than3 indicated a reasonable fit (Kline, 1998). CFI can range from 0 to 1, where0 represents the goodness of fit associated with the null model and 1 repre-

    sents the goodness of fit associated with a saturated model; a value above

  • 7/27/2019 Social Work Intention

    13/23

    Supervisory Communication 375

    0.90 suggests an acceptable fit between the model and data (Kline, 1998).Finally, the RMSEA is the most sensitive index for models with misspecifiedfactor pattern coefficients and the least sensitive to the sample size com-pared to other model fit indexes. An RMSEA of 0.05 or less indicates a good

    fit and an RMSEA of 0.08 or less indicates a reasonable error of approxima-tion (Browne & Cudeck, 1993).In the final step, multi-group analyses were performed to test the mod-

    erating effects of supervisory communication in this model. The main ques-tion of Hypothesis 4 was whether supervisory communication moderatesthe relationship between role stress and burnout in social workers. Anotherway of expressing the question in terms of an interaction effect is whethergroup membership (i.e., high supervisory communication vs. low supervi-sory communication) moderates the relationship between role stress andburnout. For this analysis, the sample was divided into two sub-samples

    based on whether the respondent was above or below the median on eachtype of supervisory communication. Three estimated models were used totest the moderating effect on each path (between role stress and burnout).In the first model, Model 1, we freely estimated the models for high and lowon each type of supervisory communication. In Model 2, all of the factorloadings were constrained to be equal. In Model 3, a path between role stressand burnout was constrained to be equal across groups. A non-significantchi-square difference between Model 1 and Model 2 would indicate mea-surement invariance across groups. If factor loadings were equivalent acrossgroups, the moderating effects could be examined by examining chi-squaredifference between Model 2 and Model 3. A significant chi-square differencebetween Model 2 and Model 3 would suggest that the equality constraintsare not consistent with the data and, thus, that an interaction effect exists(Rigdon, Schumacker, & Wothke, 1998).

    RESULTS

    Table 1 presents the means, standard deviations, and correlation coefficients

    for 14 observed variables in the final structural equation model. The resultsof the confirmatory factor analysis, including two latent variables, did notsupport the measurement model. Overall, there was a c2(8) value of 27.48,with c2/df = 3.44, CFI = 0.93, and RMSEA = 0.107. A standardized factorloading for personal accomplishment was small (b= 0.24), and the burn-out factor only marginally explained variance in personal accomplishment(R2= 0.06). It is consistent with previous findings that emotional exhaustionand depersonalization may be key components reflecting a high degree ofburnout among social workers (Walkey & Green, 1992). Therefore, personalaccomplishment was excluded from the measurement model. Because the

    desired number of indicators for a latent variable is at least three (Kline, 1998),

  • 7/27/2019 Social Work Intention

    14/23

    376

    TABLE1

    Means,StandardDeviations,andCorrelationCoefficientsf

    orObservedVariables

    Var

    iable

    1

    2

    3

    4

    5

    6

    7

    8

    9

    10

    11

    12

    13

    14

    1.

    JC

    1.0

    0

    2.

    PC

    0.7

    5*

    1.0

    0

    3.

    UC

    0.3

    5*

    0.4

    8*

    1.0

    0

    4.

    RC

    0.3

    1*

    0.2

    9*

    0.0

    1

    1.0

    0

    5.

    RA

    0.4

    0*

    0.2

    9*

    0.0

    8

    0.4

    2*

    1.0

    0

    6.

    RO

    0.27*

    0.2

    4*

    0.0

    6

    0.5

    2*

    0.2

    9*

    1.0

    0

    7.

    EE1

    0.2

    6*

    0.3

    1*

    0.1

    1

    0.47*

    0.37*

    0.4

    9*

    1.0

    0

    8.

    EE2

    0.2

    5*

    0.2

    9*

    0.07

    0.5

    1*

    0.3

    3*

    0.5

    4*

    0.9

    0*

    1.0

    0

    9.

    DP

    0.0

    5

    0.0

    8

    0.0

    3

    0.3

    4*

    0.2

    1*

    0.2

    3*

    0.5

    1*

    0.4

    9*

    1.0

    0

    10.

    TI

    0.37*

    0.3

    6*

    0.07

    0.27*

    0.2

    8*

    0.2

    4*

    0.3

    9*

    0.3

    9*

    0.0

    6

    1.0

    0

    11.

    Age

    0.0

    8

    0.1

    2

    0.0

    6

    0.0

    9

    0.0

    0

    0.0

    0

    0.0

    2

    0.0

    1

    0.0

    8

    0.1

    0

    1.0

    0

    12.

    Gender

    0.0

    5

    0.07

    0.0

    8

    0.1

    5*

    0.0

    8

    0.1

    0

    0.0

    4

    0.0

    4

    0.1

    3

    0.0

    2

    0.2

    5*

    1.0

    0

    13.

    Tenure

    0.0

    5

    0.07

    0.0

    2

    0.0

    2

    0.1

    2

    0.0

    1

    0.0

    1

    0.0

    0

    0.0

    9

    0.0

    8

    0.57*

    0.1

    8*

    1.0

    0

    14.

    Salary

    0.0

    8

    0.0

    5

    0.0

    6

    0.0

    8

    0.1

    4*

    0.0

    4

    0.0

    2

    0.0

    5

    0.0

    6

    0.0

    9

    0.3

    3*

    0.2

    9*

    0.4

    2*

    1.0

    0

    Mean

    3.4

    8

    3.4

    6

    4.07

    4.3

    5

    2.77

    5.1

    1

    3.1

    1

    2.9

    0

    1.47

    3.8

    3

    45.1

    1

    0.7

    8

    7.9

    9

    57.47

    SD

    1.0

    6

    1.0

    9

    0.6

    8

    1.4

    5

    1.0

    5

    1.1

    9

    1.3

    6

    1.2

    8

    1.17

    1.7

    3

    12.0

    0

    0.4

    1

    7.5

    3

    17.9

    1

    JC:jobrelevantcommunication;PC:positiverelationshipcommunication;U

    C:upwardcommunication;RC:role

    conflict;RA:roleambiguity;RO:roleoverload;EE1:

    emo

    tionalexhaustion1;EE2:emotional

    exhaustion2;DP:depersonalization

    ;TI:turnoverintention;Tenure:orga

    nizationaltenure;Salary:annualsala

    ry.

    *p