Abstract
This study extended prior research on staff turnover in adult corrections to juvenile corrections by analyzing direct and indirect interrelationships among personal, work environment, job attitude, and turnover variables. Data came from a mail survey of youth worker staff as well as from agency archives. Race, age, tenure, input into decisions, and job stress had significant direct effects on job satisfaction, while organizational commitment was directly affected by gender, stress, and satisfaction. Tenure, satisfaction, and commitment directly affected intent, while only race and age directly affected actual turnover. Satisfaction and commitment performed significant mediating functions. Results suggest that staff turnover intent can be reduced by promoting job satisfaction and organizational commitment and, further, that these attitudes can be improved by providing staff greater input into decisions and reducing job stress. Future research must specify conditions under which intent predicts behavior, as the two were not significantly related in this study.
Keywords
The magnitude and negative consequences of staff turnover are well documented in literature on correctional organizations. The average annual turnover rate in adult corrections is around 20% and has been reported to sometimes exceed twice that (Lambert, 2001; Lambert & Hogan, 2009b; Patenaude, 2001; K. Wright, 1994). The situation is no better in juvenile corrections. An earlier study (T. A. Wright, 1993) reported annual turnover to be almost 20% among detention workers, and a more recent one (Minor, Wells, Angel, & Matz, 2011) found that approximately a quarter of newly hired staff resigned from state-operated juvenile correctional facilities within the first year of being hired and trained. 1 The problem is not limited to actual turnover but also includes employees wanting or planning to quit their jobs, a cognitive process known as turnover intention. A study found that 11% of staff working in one state’s juvenile correctional facilities intended to leave the agency over the next year, other than for purposes of retirement (Matz, Wells, Minor, & Angel, 2012), while another found that a third of correctional officers at a state prison reported being likely to leave their jobs in the coming 3 years (Minor, Dawson-Edwards, Wells, Griffith, & Angel, 2009).
Staff turnover has deleterious effects for correctional agencies. Crews and Bonham (2007) cited an annual turnover cost of over $21 million for one state department of corrections. Minor et al. (2011) reported costs of turnover among new hires to exceed $4.5 million based on estimated replacement costs of $31,000 per exiting employee. Lambert and Hogan (2009b) estimated that combined costs of starting a new employee on the job in adult corrections can be as much as $20,000. Cost figures like these assume added importance as agency administrators grapple with pressures to prioritize scarce funds and contain expenses amid fiscal shortfalls.
The costs of staff turnover in corrections are not restricted to the monetary realm. Correctional organizations rely heavily on staff to accomplish their goals, which is why Lambert (2001) pointed to staff turnover levels as a good barometer of correctional agency effectiveness. Turnover can undermine effectiveness by creating personnel shortages, causing existing staff to be overworked, hampering staff morale, destabilizing daily operations, and hurting the agency’s public image. For its part, turnover intent can be indicative of discontented and non-invested workers who view their current job as a stopgap until better employment materializes and who contribute minimally to the organization in the meantime—possibly while expending time and energy seeking alternative employment. When issues like these are combined with the drain on monetary resources that turnover can exact, it is not difficult to see how high turnover, once embedded in a relatively ineffectual organization, can beget more of the same (Minor et al., 2011; Mitchell, Mackenzie, Styve, & Gover, 2000). After all, resources expended in response to turnover (e.g., processing separations, paying overtime to cover vacancies, training new staff, etc.) and turnover intent (e.g., picking up slack for employees who want or plan to leave) are resources that cannot be devoted to either (a) programming and services to promote offender betterment or (b) improvements in pay, benefits, and work conditions that might better control staff turnover.
Of course, correctional agencies must occasionally separate select staff from their positions (e.g., those who abuse prisoners or coworkers), and this is why researchers distinguish between voluntary and involuntary turnover (Price & Mueller, 1986). With the former, the employee initiates cessation of his or her job (e.g., through resignation, retirement, or voluntary transfer), while with the latter, the agency does so (e.g., through layoff, firing, or forced transfer). It turns out that voluntary turnover is more common in corrections than involuntary turnover, and it is associated with more cost and disruption to the organization because, disproportionately, it is likely to claim higher performing employees (Blakely & Bumphus, 2004; Price, 1977; T. A. Wright, 1993). The good news is that an organization can usually exert greater control over voluntary turnover because factors prompting involuntary turnover (e.g., budget cuts that result in layoffs and employee behavior that results in termination) are less amenable to agency control.
For correctional administrators to best position themselves to control voluntary turnover and turnover intention among their staff, knowledge is required of underlying dynamics. Administrators need to know which personal characteristics of staff, which work environment factors, and what kind of job attitudes among workers are most related to staff leaving (or wanting to leave) their jobs, as well as how these sets of variables relate to one another. Without such information, turnover control efforts are likely to be fragmentary and even misguided. Hence, this study builds on a limited volume of past research, especially the work of Lambert and Hogan (2009b) in adult corrections which is most comparable to the current study, to present and test a theoretical model of turnover intent and voluntary turnover among staff working in correctional facilities administered by the Kentucky Department of Juvenile Justice (DJJ). Based on prior research, the antecedents of turnover intent and behavior posited by the model include staff personal characteristics (education, gender, race, age, and tenure), work environment variables (staff perceptions of role ambiguity, input into decision-making processes, job stress, organizational fairness, and job dangerousness), as well as the job attitudes of job satisfaction and organizational commitment.
Previous Research
Given the consequences of voluntary staff turnover in correctional organizations, remarkably little research has addressed the factors underlying the problem. Most research has focused on staff working in adult rather than juvenile corrections (Byrd, Cochran, Silverman, & Blount, 2000; Camp, 1994; Ferdik, Smith, & Applegate, 2014; Jacobs & Grear, 1977; Jurik & Winn, 1987; Kiekbusch, Price, & Thesis, 2003; Lambert, 2006; Lambert & Hogan, 2009b; Lambert & Paoline, 2010; Leip & Stinchcomb, 2013; Minor et al., 2009; Simmons, Cochran, & Blount, 1997; Slate & Vogel, 1997; Stohr, Self, & Lovrich, 1992). Only a handful of studies have examined juvenile corrections (Liou, 1998; Matz et al., 2012; Minor et al., 2011; Mitchell et al., 2000; Tipton, 2002; T. A. Wright, 1993). This deficiency needs to be addressed because there are differences between working in juvenile corrections and working in adult prisons. In juvenile corrections, even staff who fill what are conventionally regarded as custodial roles usually also have much responsibility for promoting rehabilitation and the best interests of youth; the role is not exclusively (or even primarily) custodial (Liou, 1998). Also, juvenile correctional facility work environments differ from their adult counterparts in usually being smaller, less crowded, less preoccupied with security, and less violent (Wells, Minor, Angel, Matz, & Amato, 2009). Such differences could well affect turnover dynamics.
In addition, much correctional staff turnover research (e.g., Byrd et al., 2000; Griffin, Hogan, & Lambert, 2014; Kiekbusch et al., 2003; Leip & Stinchcomb, 2013; Mitchell et al., 2000) has examined turnover intention rather than behavior, something understandable considering the impediments to gathering valid turnover data often presented by agency records, as well as the relative ease with which intent data can be obtained from surveys (Lambert, 2006). Nonetheless, turnover intent and behavior are not synonymous concepts. One is a cognitive process (sometimes with an attendant behavioral component of actively seeking another job), while the other involves acting on a decision to quit a job. There is disagreement over the extent to which actual turnover is predicted by turnover intent (Camp, 1994; Kirschenbaum & Weisberg, 1990; Minor et al., 2009; Steele & Ovalle, 1984; Udechukwu, Harrington, Manyak, Segal, & Graham, 2007). A staff member may, for instance, want or plan to leave an organization without actually doing so, or he or she may resign on short notice in response to another opportunity without much desire or planning preceding resignation. Even if one assumes that turnover intent is not a stable predictor of actual turnover, it remains true, for reasons stated above, that turnover intent among employees is not a good sign for an organization. Optimally, then, studies ought to include both concepts, but they seldom have. The present study incorporates both variables.
Turnover Modeling
Lambert (2001) developed a theoretical model of turnover among staff working in adult corrections in which voluntary turnover is directly affected by turnover intention. In turn, intent is directly affected by alternative employment prospects, the job attitude variables of job satisfaction and organizational commitment, as well as employee personal characteristics. In Lambert’s model, personal characteristics directly affect both job attitude variables, as do work environment factors. Job satisfaction directly affects organizational commitment. Finally, commitment mediates the effects of satisfaction on intent, and both job attitude variables mediate the effects of personal characteristics and work environment factors on turnover intent. In researching this model with staff from one prison, Lambert (2006) reported that turnover intent was significantly greater among women staff, those with college educations, those with less tenure, as well as among individuals displaying less job satisfaction and organizational commitment; the strongest correlate was job satisfaction. Indirect effects were not estimated in this study. Lambert and Hogan (2009b) later tested a model of turnover intent among adult corrections workers in which intent was conceptualized as a direct effect of personal characteristics, job satisfaction, organizational commitment, and alternative employment options. The effects of personal characteristics and work environment variables were theorized to be mediated by the job attitude variables, and organizational commitment was posited to mediate the relationship of job satisfaction to intent (see also Lambert, Barton, & Hogan, 1999). Lambert and Hogan’s findings were generally supportive of this model; sizable portions of variability were explained in turnover intent, job satisfaction, and organizational commitment. This study, however, did not include a measure of turnover behavior.
Other than the Matz et al. (2012) study referenced previously, we are aware of no published research that has tested theoretical models designed to account for turnover among staff working in juvenile corrections, and Matz et al. estimated only direct effects on turnover intent. We are also aware of no study of juvenile corrections staff (and very few studies of adult correctional staff) that have incorporated measures of both turnover intention and turnover behavior.
Recent research on correctional staff (Lambert & Hogan, 2009b; Lambert & Paoline, 2010; Matz et al., 2012) has grouped variables conceived as possible antecedents of turnover into three categories: (a) staff personal characteristics (e.g., gender or race), (b) work environment factors (e.g., perceptions of dangerousness or input into workplace decisions), and (c) job attitudes (e.g., job satisfaction or organizational commitment). The present study included all three groupings of possible antecedents.
Personal Characteristics
Research is inconclusive about the relationship of staff personal characteristics to turnover measures (Matz, Woo, & Kim, 2014). Studies of staff in juvenile corrections have produced contradictory findings concerning the roles of gender, age, and tenure with agency (Liou, 1998; Mitchell et al., 2000; Tipton, 2002; T. A. Wright, 1993). Findings can depend on whether investigators use turnover intent or behavior as the dependent variable. For example, Wright used a behavioral measure and found an inverse relationship with tenure, whereas Tipton reported that tenure was positively related to intent. Similarly, while Mitchell et al. (2000) reported that minority race and higher education levels were related to increased intent, Matz et al. (2012) found no effects of these variables. Despite there being much more research on staff in adult corrections, the same pattern of inconsistent findings exists. There is some trend across studies for turnover intent and behavior to be more probable among women, racial minorities, younger staff, more educated staff, and those with less tenure, but on the whole the literature remains inconclusive (Lambert & Hogan, 2009b). We examined both direct and indirect effects of these variables in hopes of helping better understand the nature and extent of their role.
Work Environment Variables
The literature from the adult and juvenile levels is generally conclusive that work environment and job attitude variables are more strongly related to turnover measures than personal variables (Leip & Stinchcomb, 2013; Matz et al., 2014). Work environment variables can be assessed from multiple sources (e.g., using agency records of assaults on staff as a proxy for job dangerousness), but most often in correctional staff research, these variables are measured as self-reported perceptions of the context in which staff work. Self-reported perceptions were used in the present study.
Role ambiguity is the extent to which an employee perceives his or her job duties as uncertain or obscure instead of being clear (Rizzo, House, & Lirtzman, 1970). This variable did not directly affect turnover intent in Lambert and Hogan’s (2009b) study but was indirectly related to intent through employee job satisfaction. Because rehabilitative and custodial responsibilities tend to be more fused in juvenile corrections, there is reason to believe ambiguity over work roles may be related to staff turnover there. While Liou (1998) reported that juvenile correctional staff with a greater punitive (vs. treatment) orientation displayed greater turnover intention, the role ambiguity variable has not been directly examined in prior studies of turnover among juvenile corrections staff.
Input into decision making is the extent to which employees see themselves as having a say and being involved in organizational decisions (Lambert, Minor, Wells, & Hogan, 2014). Like role ambiguity, this variable has not been examined in prior studies of turnover in juvenile corrections. However, it has been linked to turnover intention in adult corrections (Jurik & Winn, 1987; Matz et al., 2014; Minor et al., 2009; Slate & Vogel, 1997). Lambert and Hogan (2009b) reported that input into decisions negatively affected turnover intent through organizational commitment.
As defined here, job stress is a work environment–induced perception of experiencing strain, drain, and hardening in a job (Minor, Wells, Lambert, & Keller, 2014). The only study of juvenile corrections staff that has included this variable (Mitchell et al., 2000) reported job stress to be negatively related to turnover intent. Job stress has been linked to turnover in some studies of adult corrections (Matz et al., 2014; Minor et al., 2009; Slate & Vogel, 1997) but not in others (Camp, 1994; Lambert & Paoline, 2010). Byrd et al. (2000) reported that job stress indirectly affected turnover intent by promoting job dissatisfaction.
Organizational fairness refers to whether staff believe they are treated in non-biased and just manner at work (Lambert & Hogan, 2009b). This variable has not been directly included in studies of turnover in juvenile corrections, but Tipton (2002) reported positive relationships of two allied variables (satisfaction with supervision and promotion prospects) to turnover intent. Nor has the variable received much attention in turnover research in adult corrections. Patenaude’s (2001) study of Arkansas correctional officers linked turnover with perceptions of unfair treatment from supervisors and organizational management. In Lambert and Hogan’s (2009b) study, organizational fairness indirectly affected turnover intent through both job satisfaction and organizational commitment.
Dangerousness refers to the perceived likelihood of injury or assault occurring on the job (Cullen, Link, Wolfe, & Frank, 1985). This variable has been included in past studies of juvenile corrections staff, but results have been mixed. While Tipton (2002) reported an inverse relationship between staff perceptions of safety and turnover intent, other studies have found no relationship (Matz et al., 2012; Mitchell et al., 2000). Mixed findings have also been reported in the literature on adult correctional staff (Lambert, 2006; Lambert & Paoline, 2010; Matz et al., 2014; Minor et al., 2009; Patenaude, 2001). Lambert and Hogan (2009b) reported no significant direct or indirect effects of this variable on turnover intent among prison staff.
Job Attitudes
Research on correctional staff turnover has also examined various attitudes held by workers concerning their employment. The most often studied attitudes are job satisfaction and organizational commitment. Job satisfaction is the extent to which people like their jobs and find their work enjoyable and meaningful, while organizational commitment is the bond of identification and loyalty an employee has to the organization for which he or she works (Lambert et al., 1999; Porter, Steers, Mowday, & Boulian, 1974). Of the different personal, work environment, and job attitude variables that have been included in studies of correctional staff turnover, job satisfaction and organizational commitment have been most consistently linked to measures of turnover. There are exceptions (Camp, 1994; Minor et al., 2011), but studies have usually uncovered direct effects of these variables on turnover (Griffin et al., 2014; Jurik & Winn, 1987; Lambert & Paoline, 2010; Leip & Stinchcomb, 2013; Matz et al., 2014; Udechukwu et al., 2007). Furthermore, these variables appear to be important in mediating the relationship of both personal and work environment variables to turnover intention (Byrd et al., 2000; Lambert & Hogan, 2009b). Studies of staff in juvenile corrections have linked job satisfaction to turnover (Liou, 1998; Mitchell et al., 2000; T. A. Wright, 1993) but have seldom examined organizational commitment. Two exceptions are works by Minor et al. (2011), which found no relationship of organizational commitment to turnover behavior, and by Matz et al. (2012), which reported a positive relationship of commitment to intent. The present study includes these key variables and estimates direct and indirect effects on both outcomes.
Current Study
We used path analysis to examine the model of turnover intent and turnover behavior depicted in Figure 1. The model is comprised of exogenous and endogenous variables. The former are variables with no antecedents or causes specified and include the five personal characteristics and the five work environment variables. The endogenous variables have antecedents posited in the model and include the two job attitude variables (job satisfaction and organizational commitment), along with the two turnover measures (intent and voluntary turnover behavior). The model derives from the one specified by Lambert and Hogan (2009b) but has the following differences: (a) Lambert and Hogan included staff position as a personal characteristic, while as specified below, position is a constant in this study; (b) Lambert and Hogan included role conflict and role overload among the work environment variables, while this study did not contain measures of those variables and included job stress instead; (c) this study lacked a measure of staff views toward alternative employment options; and (d) this study incorporated a measure of turnover behavior.
Hypothesized Model of Turnover Intent and Turnover Behavior
In Figure 1, personal characteristics are posited to have direct effects on job satisfaction, which in turn is posited to mediate the relationship of personal characteristics and directly affect organizational commitment. Personal characteristics are also posited to directly affect organizational commitment, turnover intent, and turnover behavior. Work environment variables are expected to have direct effects on job satisfaction, which in turn is expected to mediate the relationship of work environment variables with turnover intent and turnover. Likewise, work environment variables are expected to have direct effects on organizational commitment and turnover as well as indirect effects on turnover intent and turnover mediated by organizational commitment. Job satisfaction is postulated to have direct effects on turnover intent and turnover and indirect effects through organizational commitment. Organizational commitment was predicted to have direct effects on both turnover intent and turnover. Finally, turnover intent is hypothesized to directly affect turnover behavior and mediate the relationships of personal characteristics and job attitudes to turnover behavior.
In terms of work environment, the model postulates that employees will display lower job satisfaction and organizational commitment and also be more likely to quit their jobs if they see their work roles as ambiguous, see themselves as lacking input into decisions, are experiencing relatively high job stress, perceive the organization for which they work as being less fair, and/or perceive their job as relatively dangerous. Job satisfaction and organizational commitment are hypothesized to mediate the effect of these variables on turnover intention and behavior. For example, staff members who experience role ambiguity and stress may still find their jobs enjoyable, feel bonded to the organization, and be less likely to want to leave as a result.
Finally, the present study treated position as a constant and focused on youth worker staff. These workers have the most frequent and direct daily contact with youth in DJJ facilities, and they are the most numerous type of staff member employed in these facilities. They perform a hybrid of rehabilitative, human service, and custodial roles. It was turnover among these staff that was of most concern to the DJJ administration.
Method
Participants and Data Collection
The participants for this study were 437 youth workers employed at correctional facilities administered by the Kentucky DJJ. These facilities included less secure day treatment facilities and group homes as well as more secure short-term detention centers and longer term youth development centers. 2 As Table 1 indicates, the majority of staff worked in youth development and detention centers. Participants’ personal characteristics are also summarized in Table 1. It can be seen that the majority were male, White, and had not completed college degrees. Their average age was just over 39 years, and the average tenure in corrections was almost 8 years. 3
Descriptive Statistics for Variables
Note. α = Cronbach’s alpha value. The central tendency measures presented for the seven scales (i.e., role ambiguity through organizational commitment) are item measures.
Participants were assured of the voluntary and confidential nature of their participation in the study. The mail survey instrument was pilot tested with 113 staff spread across all four facility types. This testing resulted in an 85.8% return rate and a few minor wording clarifications. For the main administration, the instrument was mailed to all remaining DJJ facility staff, and a standard survey procedure was used involving pre-notice letters, the initial mailing, postcard reminders, and two follow-up mailings (Leeuw, Hox, & Dillman, 2008). This procedure yielded a 70.8% response rate, inclusive of surveys from the pilot test.
Variables
Except for facility type and turnover behavior data, which were retrieved from DJJ records, all variables in this study came from the survey described below. Table 1 provides descriptive information on the five work environment variables, the job attitude measures of job satisfaction and organizational commitment, and the two turnover variables.
Facility type was dummy coded into three dichotomous variables including youth development center, group home, and detention center; day treatment center was designated the reference group. The work environment and job attitude measures were derived through confirmatory factor analyses employing software developed by Muthén and Muthén (2010). The items loaded on the factors as predicted, and factor loadings are presented in the Appendix. As indicated in Table 1, all scales displayed high internal reliability.
Role ambiguity was measured using two items from the Quality of Supervision Scale (QSS) of Saylor and Wright’s (1992) Prison Social Climate Survey (PSCS), plus two items from Royal and Rossi (1996). Input into decision making was assessed with three QSS items and two items adapted from Downs and Hazen’s (1977) Communication Satisfaction Questionnaire. The measures of job stress, organizational fairness, dangerousness, and job satisfaction came from the PSCS. Organizational commitment was measured using items from Mowday, Steers, and Porter’s (1979) Organizational Commitment Questionnaire. The measures for role ambiguity, input into decisions, organizational fairness, and the two job attitudes were scored such that higher scores are indicative of more favorable staff perceptions (e.g., higher scores indicate greater perceived clarity of roles and lower ones more ambiguity). Lower scores on the job stress and dangerousness measures indicate more favorable perceptions.
Staff were coded as displaying turnover intent if they indicated on the survey they were likely to leave DJJ within the next year; those who indicated plans of staying longer than 12 months were coded as not displaying turnover intent. 4 Similarly, turnover behavior was measured using agency records and was coded in terms of whether a given staff member was still employed with the organization or had resigned 1 year following survey administration. 5
Results
The cases to variables ratio of 27.3:1 was verified as sufficient. Data were screened for skewness or kurtosis, as well as outliers. Given that our instrument was designed to measure several different perceptions of the work environment and job attitudes, some of which may be considered problematic by only a small number of staff, it was expected that some data would be skewed. Two variables were positively skewed and seven negatively skewed, but these were determined to be legitimate and to contain valuable information related to the study. As precaution, the maximum likelihood robust estimator, which is robust to non-normal data and non-independence of observations, and can include missing data, was used to accommodate any lack of multivariate normality (Tabachnick & Fidell, 2013). There were no outliers. Inspection of correlation matrices, condition indexes, and variance proportions revealed no issues with multicollinearity. The correlation between the measures of job satisfaction and organizational commitment was rather high (.74), but according to Tabachnick and Fidell (2013), “the statistical problems created by singularity and multicollinearity occur at much higher correlations (.90 and higher)” (p. 90). In addition, the correlation between the two was expected to be high because job satisfaction is seen as one of the best predictors of organizational commitment (Lambert, 2004; Lambert & Hogan, 2009b).
Given that survey participants were clustered within different facilities, and that staff perceptions within a particular facility may display similar patterns reflecting the shared environment, multi-level modeling was considered. However, diagnostic procedures revealed that the average cluster size was only 12.85 for the 34 clusters, and the majority of intraclass correlation coefficients were single digits (on the percentage scale). Hence, single-level analysis was deemed most appropriate (Tabachnick & Fidell, 2013).
The bivariate correlation results are presented in Table 2. As can be seen, increased tenure and age were significantly correlated with higher job satisfaction, and White staff had significantly higher organizational commitment. In addition to being significantly related to one another, job satisfaction and organizational commitment scores were significantly related to more favorable staff perceptions on the five work environment measures. Working in a youth development center was related to less job satisfaction and less organizational commitment, while working in detention was associated with higher commitment. In addition, tenure and age were inversely correlated with turnover intent, and less turnover intent was related to more favorable work environment perceptions as well as greater job satisfaction and organizational commitment. Turnover behavior was inversely related to both job satisfaction and organizational commitment as well as input in decisions; it was also significantly more likely among non-Whites.
Correlation Matrix
Note. YDC = youth development center; DT = detention center; GH = group home.
p < .05. **p < 01.
Direct Effects
Table 3 presents direct effects data for the model in Figure 1. 6 The data revealed significant direct effects of race, age, tenure, and two work environment variables on job satisfaction. Specifically, job satisfaction was significantly greater among White staff, older staff, and staff with more tenure; it was also greater among staff who perceived more input into decision making and less job stress. The relationship of less role ambiguity and greater job satisfaction approached but did not attain statistical significance (p = .055). Organizational commitment levels were significantly greater among males as well as among staff with lower job stress and higher job satisfaction. Turnover intent was significantly more likely among staff having less tenure, as well as among those with less job satisfaction and lower organizational commitment. By contrast, only personal variables had direct effects on turnover behavior, with turnover significantly more likely among non-Whites and younger staff. The proportions of variance explained were 43% for job satisfaction, 62% for organizational commitment, 51% for turnover intent, and 15% for turnover behavior.
Direct Effects
Note. B = unstandardized beta coefficient; YDC = youth development center; DT = detention center; GH = group home; N/A = not applicable; AIC = Akaike information criterion; BIC = Bayesian information criterion.
p ≤ .05. **p ≤ .01.
Indirect Effects
A number of variables had significant indirect effects on the dependent variables through a mediating variable. Job satisfaction mediated the relationships of race (0.193, p < .01), input into decisions (0.089, p < .01), and job stress (−0.155, p < .01) to organizational commitment. Similarly, job satisfaction mediated the relationship of race (−0.303, p < .05) to turnover intention, as well as the relationships of input into decisions (−0.141, p < .05) and job stress (0.244, p < .01) to intent.
Organizational commitment also served important mediating functions. First, it was a significant mediator of the relationships of both gender (−0.185, p < .05) and stress (0.121, p < .01) to turnover intent. Second, the relationship between job satisfaction and turnover intent was mediated by organizational commitment (0.334, p < .01).
As mentioned above, consistent with the model hypothesized in Figure 1, race and age exerted significant direct effects on turnover behavior. Contrary to predicted, however, turnover behavior was not directly or indirectly affected by work environment variables. Nor was it affected in either direct or mediational fashion by job satisfaction, organizational commitment, or turnover intention.
Discussion and Conclusion
This study aimed to contribute to the correctional staff literature by focusing on interrelationships between personal, work environment, job attitude, and turnover variables. The study addressed three voids in current knowledge. First, it is largely unknown to what extent findings from the adult corrections context hold for the juvenile one. Most previous research has been conducted with staff in adult corrections, and we tested a model informed by this research with staff employed in juvenile facilities. We examined several variables that have either been under- or un-researched in work on staff turnover in juvenile corrections (i.e., role ambiguity, input into decisions, job stress, organizational fairness, perceived danger, and organizational commitment). Second, turnover intention and behavior are not synonymous concepts, and most research has addressed intent only. This study is one of few to have included both variables. Third, most studies of correctional staff turnover have examined only direct effects of variables and excluded consideration of indirect effects and mediating variables. In testing the model in Figure 1, this study examined the mediational roles of job satisfaction, organizational commitment, and turnover intent. Hence, our study expanded on prior research to build understanding of dynamics surrounding turnover intention and behavior.
Our results partially support the turnover model in Figure 1. While there has been less study of staff job satisfaction in juvenile corrections, consistent with research in adult corrections (Lambert, Hogan, & Barton, 2002), increases in age and tenure were associated with higher satisfaction. As staff age and remain with an organization, they may find positions that fit their needs or are more rewarding, resulting in higher satisfaction. Also, White staff were generally more satisfied with their jobs. The relationship between race and satisfaction has varied by the adult correctional facility under study (Lambert et al., 2002). The majority of DJJ facilities are in rural areas without large percentages of non-White residents. Because most DJJ staff are White, non-White staff may experience tokenism and thus lower satisfaction (Minor et al., 2014). Consistent with past work (Lambert & Paoline, 2008; Slate, Wells, & Johnson, 2003), input into decision making had a positive association to satisfaction. Input allows staff to feel valued and provides a sense of belonging and control. As expected, job stress was related to lower satisfaction. Being stressed is unpleasant and can have a variety of deleterious effects on staff (Keinan & Maslach-Pines, 2007), including lessened job satisfaction (Lambert, 2004).
Gender was the only personal characteristic to have a significant association with organizational commitment, a variable missing from most research on juvenile corrections staff. Male staff displayed higher affective commitment. Given that males comprised three-quarters of the respondents, they could feel more connected with the organization than women. Women could face greater job obstacles (e.g., pay and promotion), greater tokenism, and greater perceived risk of victimization in a field long dominated by men employees (Gordon, Proulx, & Grant, 2013). Job stress was also a significant correlate of organizational commitment. This finding is consistent with studies of staff in adult corrections (Lambert, 2004; Lambert & Hogan, 2009b). Staff who experience prolonged job stress can come to blame the organization, thereby hampering organizational commitment (Lambert, Hogan, & Jiang, 2008). Also consistent with previous studies (Lambert, 2004; Minor, Wells, & Angel, 2006), job satisfaction had a significant positive relationship with organizational commitment. Staff who are satisfied with their jobs probably see the organization in favorable terms, and this is likely to promote a stronger bond with it.
While prior research has produced mixed findings concerning the tenure variable (Matz et al., 2012; Mitchell et al., 2000; Tipton, 2002; T. A. Wright, 1993), we found a negative association of tenure to turnover intent. This makes sense because, over time, investments accumulated in a correctional agency can bind staff to want to remain (Becker, 1960; Lambert et al., 2008). Both job attitudes had direct negative effects on turnover intent, something consistent with previous studies (Lambert & Hogan, 2009b; Leip & Stinchcomb, 2013; Matz et al., 2012). Staff who find their jobs satisfying will be less likely to want to leave and actively seek another job (Lambert & Hogan, 2009b). Likewise, those who have strong affective bonds with the organization have little compelling reason to want to leave (Lambert, 2001).
Race and age were the only personal characteristics to have significant direct relationships to actual turnover. Compared with White staff, non-White were more likely to leave their jobs. As described earlier, relatively few studies have included behavioral measures of turnover, and in most such studies, race was either not included as a variable (Kiekbusch et al., 2003) or found to have insignificant effects on turnover (Minor et al., 2011). Two studies (Jacobs & Grear, 1977; Jurik & Winn, 1987) reported significantly greater turnover among minority staff, and both interpreted this in terms of conflict between predominantly White administrators and non-White line officers. The same interpretation could be valid in this study, especially because non-White staff also showed lower job satisfaction. Younger staff were also more likely to leave employment. Older staff were more satisfied with their jobs and were probably more entrenched and involved with their employment. They may have had greater financial and familial obligations and possibly faced greater age barriers in obtaining new jobs.
Contrary to the hypothesized model, work environment variables did not have significant direct effects on actual turnover, and nor did they indirectly affect it through job attitudes. Furthermore, neither the job attitudes nor turnover intent affected turnover behavior. As posited though, and consistent with the limited prior research (Lambert 2001, 2006; Lambert & Hogan, 2009b), work environment variables had significant indirect effects on turnover intent through job attitudes. Udechukwu et al. (2007) reported significant bivariate associations of job satisfaction and organizational commitment to intent, but voluntary quitting was unrelated to both job attitudes. The well-established relationship of these job attitudes to turnover intent notwithstanding (Matz et al., 2014), on the whole studies have produced mixed findings about the relationship of the attitudinal variables and turnover intent to actual turnover in corrections (Camp, 1994; Minor et al., 2011; Stohr et al., 1992; T. A. Wright, 1993). At least in some cases, actual quitting may be more spontaneous or situational than job satisfaction, organizational commitment, and contemplation of quitting. Turnover behavior could be conditioned by variables extraneous to the work environment and attendant job attitudes (e.g., other immediate job openings in relation to the employee’s overall economic situation at the particular time).
Turnover intent is often viewed as being the last stage before, and a good predictor of, voluntary turnover (Lambert, 2006; Matz et al., 2014). Kiekbusch et al.’s. (2003) study of jail staff turnover supports this logic, but other findings do not (Camp, 1994; Udechukwu et al., 2007). There are plausible explanations for the lack of a significant linkage between turnover intent and turnover in the present study, other than intent simply being un-predictive of behavior. First, the finding could be due to random error, and the relationship could actually exist in the population. Second, voluntary and involuntary turnover many have been partly confounded because it was not possible to control for select staff who resigned in lieu of being fired. Third, the relationship between turnover intent with turnover is mediated by economic conditions not measured in this study. Lambert (2001) contended that turnover intent would lead to voluntary turnover if the person had economic resources sufficient to leave the job or easily could find another job at the same or higher pay level. DJJ facilities tend to be in rural areas with fewer alternative jobs offering similar pay and benefits. The poverty rate (23.4% vs. 15.6%) and the unemployment rate (9.4% vs. 7.5%) are higher in rural areas of Kentucky than urban areas, and both are higher than the U.S. average (Rural Health Information Center, 2016). Some staff who wanted to quit may have been unable to because of a lack of viable alternatives. Similarly, unmeasured variables (e.g., staff performance) could mediate the turnover intent—behavior link. Fourth, we measured turnover behavior across 1 year following survey administration. It could take more time for some employees to move from wanting to quit to actually leaving. Alternatively, intent might be better linked to turnover behavior if the latter is measured in closer proximity to the measurement of intent (e.g., 30 days); the behavioral ramifications of intent could wane if intent goes unacted upon too long.
Irrespective of how well turnover intent predicts actual turnover, intent is a negative outcome because it is often linked with lower productivity and psychological withdrawal from the job (Lambert, 2001; Matz et al., 2012). Moreover, in cases where intent does precipitate quitting, focusing on turnover behavior is an inherently reactive approach. Once voluntary turnover transpires, it is too late for correctional administrators to initiate efforts to keep employees they wished to retain (Udechukwu et al., 2007). Focus on intent facilitates a more proactive approach.
Besides those alluded to above, our study has limitations that future research can work to overcome. First, given restriction to one state, our data may not generalize across different jurisdictions and agencies. A wider range of contexts need to be examined. Second, future studies should incorporate personal, workplace, and job attitude variables not included here to determine how these fit into turnover models. Our model accounted for sizable portions of variance in job satisfaction, organizational commitment, and turnover intent, but it did not do so for turnover behavior. The amount of variance explained may be increased by incorporating such variables as alternative employment prospects, role conflict, role overload, work-family conflict, formalization, instrumental communication, quality of supervision, job involvement, and job desirability (Ferdik et al., 2014; Matz et al., 2014). Third, while the current study was cross-sectional, longitudinal research is required to learn more about the direction of causal relationships. Fourth, future studies should use alternative measures of key variables. For example, our measure of affective organizational commitment can be broadened to encompass continuance and normative commitment, and the measure of turnover intent can be expanded to incorporate multiple scale items (Lambert & Hogan, 2009b; Meyer & Allen, 1991). Fifth, future studies need to contact those who quit to determine why they left employment with a correctional agency. Similarly, research is needed to determine why people who express turnover intent actually leave or stay. This information would shed light on the relationship of turnover intent to voluntary turnover among correctional staff. Finally, we demonstrated the importance of studying mediating variables and offered tenable explanations for our significant findings, but these explanations need to be tested further. For example, further study is warranted of the race and gender effects uncovered in this study. If future research demonstrates similar effects, it will be vital to develop empirically supported explanations. Future investigations need to further examine the relationship between job attitudes, turnover intent, and actual turnover, including attention to the temporal patterning of these variables. Additional testing could also benefit by expanding our emphasis on mediating variables to include moderating ones; the former explain relationships between variables, whereas the latter involve interaction that shapes the strength and direction of relationships (Baron & Kenny, 1986). Clearly, then, much work remains to be accomplished in this important topic area.
The current study contributes to a growing literature indicating that correctional administrators who want to curtail turnover intention should focus on improving job satisfaction and organizational commitment levels of juvenile correctional staff (Lambert & Hogan, 2009b; Leip & Stinchcomb, 2013; Matz et al., 2012). To respond to the problem of turnover, administrators need to understand its causal process (Lambert & Hogan, 2009b). Our findings demonstrate that besides directly affecting turnover intent, these job attitudes are essential in mediating the effects of personal and work environment characteristics on intent and, more specifically, that satisfaction mediates the effects of these characteristics on commitment, while commitment in turn mediates the effects of satisfaction on intent (Lambert et al., 1999). Our findings imply that job attitudes can be improved by allowing staff greater input into workplace decision making and taking measures to reduce job stress. Input into decisions can be enhanced through the use of staff advisory groups or decentralized decision-making bodies that utilize the rich knowledge staff generally have of problems and issues at their facilities (Minor et al., 2009). Research indicates that increasing staff input and autonomy is a viable way to reduce job stress (Auerbach, Quick, & Pegg, 2003; Dowden & Tellier, 2004; K. Wright, Saylor, Gilman, & Camp, 1997). This means giving staff a voice in the organization (Lambert, Hogan, & Griffin, 2007). In addition, past research indicates that job stress can be reduced by improving communication and training to reduce role conflict, role ambiguity, and role overload. Moreover, reducing the occurrence of these stressors can also improve job satisfaction, and organizational commitment (Lambert & Hogan, 2009b; Lambert, Hogan, Paoline, & Clarke, 2005). In addition, the deleterious effects of job stress on job satisfaction and organizational commitment can be countered through training initiatives to teach staff to better anticipate, preempt, and cope with workplace stressors.
Supervisors need to be made aware of the role that they play in shaping the satisfaction and commitment of staff. Quality and supportive supervision can help increase satisfaction from the job and commitment to the organization (Lambert, 2004). Administrative and supervisor support has been reported to increase the job satisfaction and organizational commitment of juvenile correctional staff, while also reducing stress from the job (Lambert, Minor, Wells, & Hogan, 2016). To increase both forms of organizational support, interpersonal communication is required by listening to staff concerns. Furthermore, acknowledging positive work by staff and providing feedback can improve the perceptions of both forms of support (Lambert & Hogan, 2009a). Not only do administrators need to be aware of the issue of organizational support, they need to ensure that managers and supervisors are aware of it and regularly practice it in a genuine manner. Eisenberger, Huntington, Hutchison, and Sowa (1986) noted that that “indiscriminate praise given to all employees or other easily penetrable facades of disingenuous approval by agents of the organization would reduce perceived organizational support” (p. 504). In addition, supervisors, managers, and administrators need to engage in transactional justice. Transactional justice means being honest and forthcoming to employees and treating them with respect and dignity (Lambert et al., 2007).
All of the suggested methods to deal with juvenile correctional staff involve changing workplace factors. This is good news for correctional administrators because changing workplace factors is more in their control than manipulating personal characteristics, if such was even legal. It may not be easy to make changes to the workplace, but the payoffs to both the staff and the organization in terms of lower job stress, increased satisfaction, heightened commitment, and ultimately lower turnover intent and turnover should be substantial. It will take genuine efforts and time for these and other changes to be made, but it can be done if there is determination and long-term planning. Doing nothing and ignoring the situation is a recipe for continued valuable resources being lost to turnover.
To the extent our results can be replicated, they suggest the need to proactively monitor and address factors promoting voluntary turnover among racial minority and younger staff, as adequate retention and representation of these staff is essential in correctional agencies. Factors rendering such staff at greater risk for turnover should be identified and constructively resolved before culminating in turnover. For example, if interpersonal work conflicts elevate turnover risk as suggested by some studies (Jacobs & Grear, 1977; Jurik & Winn, 1987), then workplace supports in the form of supervisor and peer mentoring, as well as improved communication strategies, could be put in place to counter this risk (Lambert et al., 2016). The point here is that there are creative, data-driven ways to identify turnover risks and take preventative measures before employees whom an organization wants to retain decide to leave.
Staff are vital resources for correctional organizations. They are responsible for a myriad of tasks, duties, and responsibilities that are critical for the operation of juvenile institutions and play a pivotal role in organizational success or failure. However, there has been little research on the dynamics underlying turnover intent and behavior among juvenile correctional staff. The current study adds to the literature by analyzing how personal, work environment, and job attitude variables are directly and indirectly linked to turnover intent and actual turnover among such staff. Turnover intent and behavior are common issues in correctional agencies that exact considerable tolls on resources and functioning. The positive lesson from this research is that these issues are addressable. The alternatives are for policy makers and administrators to ignore the turnover problem, or for efforts aimed at countering it to be unguided by a solid base of academic research. These need not be the choices made.
Footnotes
Appendix
Except for the job stress and dangerousness scales, all items shown below had a 7-point Likert format ranging from strongly disagree (coded 1) to strongly agree (coded 7). Job stress had a 7-point Likert format ranging from never (coded 1) to always (coded 7), and the dangerousness scale had a similar format ranging from not at all likely (coded 1) to very likely (coded 7). The numbers in the parentheses are factor loadings.
This work was supported by funding provided through the Kentucky Department of Juvenile Justice (DJJ) via contract PON2 523 0800009119. Other than funding, DJJ’s involvement was limited to providing feedback on the survey instrument, with all other research activities being the sole responsibility of the authors. The authors have no financial interest or benefit arising from direct applications of this research.