• unlimited access with print and download
    $ 37 00
  • read full document, no print or download, expires after 72 hours
    $ 4 99
More info
Unlimited access including download and printing, plus availability for reading and annotating in your in your Udini library.
  • Access to this article in your Udini library for 72 hours from purchase.
  • The article will not be available for download or print.
  • Upgrade to the full version of this document at a reduced price.
  • Your trial access payment is credited when purchasing the full version.
Buy
Continue searching

Who stays and who leaves? Social dynamics surrounding employee turnover

Dissertation
Author: Iryna Shevchuk
Abstract:
My dissertation consists of two essays that examine employee turnover as an independent and a dependent variable. In my first essay I examine the relationship between job-related attitudes, such as job satisfaction and organizational commitment, and employee turnover behavior. In a sample of over 6,000 teachers in 200 public elementary schools, I found significant variability across employee and organizational characteristics in the strength of the relationship between job attitude ratings and turnover. I attribute this variability to two sources: (a) systematic differences in attitude thresholds, i.e. the minimum acceptable level of a job attitude necessary to remain with the organization; and (b) systematic response biases in attitude ratings. I develop a model to determine which characteristics relate to significant differences in attitude thresholds and/or response biasing factors and found that both individual and organizational attributes are distinguishing factors. In my second essay I present findings from three inter-related studies investigating human and social capital as mechanisms that may determine whether and why employee retention is associated with organizational performance. In a sample of public schools I found a linear relationship between employee retention and organizational performance. Further, this positive association is fully mediated by human capital and partially mediated by social capital. In addition, I examined human and social capital at the individual level of employees who remain with the organization versus those who leave. I found that organizational performance suffers most when the employees who leave have high levels of both human and social capital. Finally, I distinguished human and social capital losses based on their specificity: organization-specific versus task-specific. As predicted, the losses of task-specific human and social capital were more deleterious to organizational performance than the losses of organization-specific forms of capital.

TABLE OF CONTENTS ACKNOWLEDGEMENTS ....................................................................................................... XI 1.0 INTRODUCTION ........................................................................................................ 1 2.0 ESSAY 1: JOB ATTITUDES AS PREDICTORS OF EMPLOYEE TURNOVER: SYSTEMATIC THRESHOLD EFFECTS ................................................................................. 9 2.1 THEORY ............................................................................................................ 11 2.1.1 Work Attitudes and Turnover: Attitude Threshold Effects ................... 13 2.1.2 Work Attitudes and Turnover: Response Biases ..................................... 15 2.2 MODEL DEVELOPMENT .............................................................................. 17 2.3 RESEARCH SETTING AND METHODS ..................................................... 20 2.3.1 Sample and Procedure ................................................................................ 20 2.3.2 Measures ...................................................................................................... 22 2.3.3 Analytic Approach ...................................................................................... 26 2.4 RESULTS ........................................................................................................... 27 2.4.1 Descriptive Results ...................................................................................... 27 2.4.2 Threshold Effects in the Relationship between Attitudes and Turnover 28 2.4.3 Response Biases in the Relationship between Attitudes and Turnover . 30 vi

2.4.4 Are Response Biases Eliminated by Constructing a Composite Index of Commitment? ............................................................................................................. 33 2.5 DISCUSSION ..................................................................................................... 34 3.0 ESSAY 2: EMPLOYEE RETENTION AND ORGANIZATIONAL PERFORMANCE: THE MEDIATING ROLE OF ORGANIZATION- AND TASK- SPECIFIC FORMS OF HUMAN AND SOCIAL CAPITAL ................................................ 50 3.1 THEORY AND HYPOTHESES ...................................................................... 54 3.1.1 Retention and Organizational Performance: The Human Capital Perspective .................................................................................................................. 55 3.1.2 Retention and Organizational Performance: Social Capital Perspective 56 3.1.3 Interacting Effect of Human and Social Capital ...................................... 58 3.1.4 Capital Specificity ....................................................................................... 60 3.2 OVERVIEW OF THE RESEARCH ............................................................... 63 3.3 STUDY 1: METHODS ...................................................................................... 65 3.3.1 Sample .......................................................................................................... 65 3.3.2 Dependent Variable .................................................................................... 66 3.3.3 Independent Variable ................................................................................. 66 3.3.4 Control Variables ........................................................................................ 68 3.3.5 Analytic Approach ...................................................................................... 69 3.4 STUDY 1: RESULTS ........................................................................................ 69 3.4.1 Summary ...................................................................................................... 71 3.5 STUDY 2: METHODS ...................................................................................... 71 vii

3.5.1 Sample and Procedure ................................................................................ 71 3.5.2 Measures ...................................................................................................... 72 3.5.3 Mediating Variables.................................................................................... 72 3.5.4 Analytic Approach ...................................................................................... 75 3.6 STUDY 2: RESULTS ........................................................................................ 75 3.6.1 Summary ...................................................................................................... 77 3.7 STUDY 3: METHODS ...................................................................................... 78 3.7.1 Sample .......................................................................................................... 79 3.7.2 Dependent Variable .................................................................................... 79 3.7.3 Independent Variable ................................................................................. 80 3.7.4 Control Variables ........................................................................................ 85 3.7.5 Analytic Approach ...................................................................................... 85 3.8 STUDY 3: RESULTS ........................................................................................ 86 3.9 DISCUSSION ..................................................................................................... 90 4.0 CONCLUSIONS ...................................................................................................... 105 APPENDIX A ............................................................................................................................ 110 APPENDIX B ............................................................................................................................ 111 BIBLIOGRAPHY ..................................................................................................................... 112 viii

LIST OF TABLES Table 2-1. Sample Characteristics ................................................................................................ 39 Table 2-2. Sample Characteristics (cont.) ..................................................................................... 40 Table 2-3. Descriptive Statistics: Attitude Measures .................................................................... 41 Table 2-4. Analysis of Variance for Commitment and Satisfaction Ratings ................................ 42 Table 2-5. Analysis of Variance for Commitment and Satisfaction Ratings (cont.) .................... 43 Table 2-6. Hierarchical Linear Modeling Results for Turnover ................................................... 44 Table 2-7. Hierarchical Linear Modeling Results for Turnover (cont.) ....................................... 45 Table 3-1. Descriptive Statistics: Study 1 ..................................................................................... 96 Table 3-2. Hierarchical Regression Analysis: Test of Hypothesis 1, Study 1 .............................. 97 Table 3-3. Descriptive Statistics: Study 2 ..................................................................................... 98 Table 3-4. Results of Mediation Analyses: Testing Hypotheses 2-3, Study 2 ............................. 99 Table 3-5. Results of Mediation Analyses: Alternative Measure of Organization-Specific Human Capital ......................................................................................................................................... 100 Table 3-6. Descriptive Statistics: Study 3 ................................................................................... 101 Table 3-7. Hierarchical Regression Analysis: Test of Hypotheses 4 and 5, Study 3 ................. 102 Table 3-8. Hierarchical Regression Analysis: Alternative Measure of Organization-Specific Human Capital ............................................................................................................................ 103 ix

LIST OF FIGURES

Figure 2-1. Effect of differences in attitude thresholds on attitude-turnover relationship ............ 46 Figure 2-2. Effect of response biases on true attitude – attitude rating relationship .................... 46 Figure 2-3. Effect of response biases on attitude-turnover relationship ....................................... 47 Figure 2-4. Threshold effect in attitude-turnover relationship ...................................................... 48 Figure 2-5. Response bias in attitude-turnover relationship ......................................................... 49 Figure 3-1. Interaction between task-specific social capital and human capital losses predicting school performance ..................................................................................................................... 104 x

ACKNOWLEDGEMENTS Finally, it is my time to put the last words into my dissertation and move forward to a new stage in my life and career. While only my name appears on the title of this document, it would never become a final product without the support, guidance, and mentorship of many people. I would like to use this moment to express my genuine gratitude and appreciation to all the people who took part in my development as a researcher. My deepest appreciation is owed to my Advisor and Chair, Carrie Leana, who is a remarkable mentor. She gave me both space and support to explore my own intellectual wanderings but was always alert to step in when I was loosing direction and needed her guidance. She provided me with sincere and straightforward advice but never forced her opinion. She worked closely with me on every draft and provided valuable comments and ideas. Thank you, Carrie! The other members of my committee were also highly supportive. I was really lucky to work with Vikas Mittal. He taught me a lot about doing research, especially with respect to research methods. He was also an immeasurable source of motivation and optimism – I always left our meetings feeling encouraged, excited, and confident. I would also like to express my keen appreciation to Marick Masters for all the support and encouragement he gave me throughout my PhD program. I value multiple opportunities to cooperate with him on various research projects. Richard Moreland encouraged me to pay attention to the smallest details and to xi

xii think carefully about every word I was using to express my ideas. His comments always revealed subtle nuances that I might overlook in my reasoning. Last, but certainly not least, Denise Rousseau – she was an immense source of great ideas and inspiration. I am truly grateful for the opportunity to work with her. In addition, I have benefited from the cooperation and interaction with other people beyond those who form my dissertation committee. I would like to thank the Organizational Behavior and Human Resource Management group at the Katz Graduate School of Business. My sincere gratitude also goes to all faculty members that I had seminars with, both at the University of Pittsburgh and at the Carnegie Mellon University. These seminars set the foundation of my thinking as a researcher and provided me with various valuable resources to conduct research. The Katz doctoral program has a terrific Doctoral Office. They do their best to ease the life of doctoral students and to buffer them from administrative issues. Carrie, John, and Gina – thank you for all your efforts! I got lucky to meet wonderful fellow doctoral students – they were enormous source of emotional support and really good friends. Ray and Tom, I enjoyed our trips to the conferences. We had really good time together. Emily and Anushri, you were excellent officemates. I will miss our chitchats. Finally, I would like to thank my family. My parents always encouraged me in my striving for knowledge and excellence. They provided me with all the resources that I needed in pursuit of my goals. They are my best role models by being intelligent, hard-working, and entrepreneurial. Finally, I would like to thank my beloved husband Andriy. Five years ago we started this journey together. Throughout that period he was a patient, loving, and unquestioningly supportive companion for me. Thank you!

1.0 INTRODUCTION Personnel turnover is a phenomenon that all organizations experience at some level. For example, in the 2008’s “100 Best Companies to Work For” (Fortune, 2008) voluntary turnover rates range from as low as 2% (S.C. Johnson & Son) to 29% (eBay). Employee turnover has attracted research attention for more than a hundred years (Dalton & Todor, 1979; Mobley, 1982). The importance of studying this major organizational phenomenon is widely recognized as it is a “relatively clear-cut act of behavior that has potentially critical consequences for both the person and the organization” (Porter & Steers, 1973: 151). Practitioners also acknowledge employee turnover as a major concern. According to a recent survey, more than two-thirds of HR managers stated that retaining and recruiting employees was their highest priority (Express Personnel Services, 2006). Theoretical discussions recognize multiple pathways from employee turnover to organizational performance implying that turnover may have both positive and negative consequences (e.g., Dalton & Todor, 1979; Dess & Shaw, 2001; Mobley, 1982; Staw, 1980). Dalton and colleagues (Dalton, Krackhardt, & Porter, 1981; Dalton & Todor, 1979; Dalton, Todor, & Krackhardt, 1982) distinguish among functional and dysfunctional voluntary turnover. Functional turnover is believed to be beneficial to the organization as it refers to the voluntary separation of individuals who are negatively evaluated by the organization. Conversely, dysfunctional turnover is seen as negative since it refers to the voluntary separation of

1

individuals who are valued by the organization. Other researchers (e.g., Abelson & Baysinger, 1984) search for optimal turnover rates by comparing turnover versus retention costs, reasoning that the lack of turnover may lead to the increased retention costs through such factors as higher compensation based on tenure. Finally, some employee turnover may initiate the inflow of “new blood” (Grusky, 1960), promote creativity in work groups through newcomer innovations (Levine, Choi, & Moreland, 2003), and increase organization’s adaptability to changes in the environment (March, 1992; Staw, 1980). The predominant theoretical approaches to organizational-level consequences of employee turnover, however, emphasize the detrimental effects of turnover as it negatively influences organizational accumulations of intangible assets. Particularly, human capital embedded into employees is generally associated with better organizational performance (Hitt, Bierman, Shimizu, & Kochhar, 2001; Huselid, 1995; Penning, Lee, & van Witteloostuijn, 1998; Pfeffer, 1994; Wright, Smart, & McMahon, 1995).

Moreover, human capital theory (Becker, 1964; Strober, 1990) stresses the importance of firm-specific human capital accumulations. While valuable,

human capital, however, is not appropriable by the organization; employees, when leaving, take their knowledge, skills, and abilities with them (Coff, 1997). Thus, one way in which personnel turnover threatens organizational performance is by diminishing organizational accumulations of human capital (Osterman, 1987). In addition to reducing the available stocks of organization-specific knowledge and skills, turnover induces disruptions to the fabric of social relations within the organization. Social capital, embedded into the connections between organizational members, is credited with a number of potential benefits including the amount and quality of information flows within the organization (Edmondson, 1999; Krackhardt & Hanson, 1993; Zander & Kogut, 1995); increased

2

cooperation and reduced need for formal control mechanisms (Coleman, 1990); and enhanced organizational citizenship behavior and beneficial extra-role behavior (Adler & Kwon, 2002; Shaw, Duffy, Johnson, & Lockhart, 2005). Turnover breaks some of the existing links in the communication network among employees and disrupts existing social capital. As personnel turnover diminishes organizational accumulations of social capital, it weakens the benefits derived from such accumulations. Relatedly, turnover poses a threat to the efficiency of organizational transactive memory systems (Moreland, 1999). Transactive memory systems reflect a shared awareness among organizational members of who knows what. Such shared awareness allows better coordination (Murnighan & Conlon, 1991; Wittenbaum, Vaughan, & Stasser, 1998), quicker and more efficient problem solving (Moreland & Levine, 1992), and overall higher productivity (Moreland & Argote, 2003). Turnover, in turn, involves changes in organizational membership and, consequently, produces disruptions to organizational transactive memory systems since the departure of old members and the arrival of new members change the distribution of task- specific knowledge and responsibilities within a group (Moreland, 1999) and can make it difficult to follow who really knows what. With few exceptions (e.g., Greebbeek & Bax, 2004; Harris, Tang, & Tsend, 2006; Koys, 2001), empirical findings point to the negative association that employee turnover has with indicators of organizational performance such as productivity (Guthrie, 2001), efficiency (Alexander, Bloom, & Nuchols, 1994; Arthur, 1994), safety (Shaw, Gupta, & Delery, 2005), and sales growth (Batt, 2002; Shaw et al., 2005). The idea that employee turnover has predominantly negative consequences to significant organizational outcomes has spurred a vast research interest concerning antecedents of turnover

3

(see Griffeth, Hom, & Gaertner, 2000; Hom & Griffeth, 1995; Maertz & Campion, 1998, for qualitative and quantitative literature reviews). Most research attention is focused at the individual level of analysis and is devoted to determining individual correlates of employee turnover (Greebbeek & Bax, 2004) since it is believed that effective retention strategies require reliable knowledge about prospective leavers (Judge, 1993; Trevor, Gerhart, & Boudreau, 1997). While understanding the determinants of employee turnover is critical to the creation of effective retention strategies, it is also important to understand and assess the potential costs and organizational consequences of turnover (Mobley, 1982). Not all employees possess knowledge, skills or connections that are of equal strategic importance to organizational objectives (Lepak & Snell, 1999). Similarly, not all employees demonstrate equally high performance levels (Dalton et al., 1982). Consequently, retention strategies are more efficient if they are targeted at employees who have the greatest impact on core activities within the organization. This dissertation takes a position that retention strategies need to target employees who are most likely to leave and employees who possess the greatest value to the organization’s objectives. In this dissertation, I treat employee turnover as both an independent and a dependent variable. More specifically, I pursue two major objectives. The first one is to more closely examine the situated nature of the relationship between employee attitudes and employee turnover. While theoretically job attitudes are core factors influencing employee turnover decisions (e.g. March & Simon, 1958; Mobley, 1977; Mobley, Griffeth, Hand, & Meglino, 1979; Steers & Mowday, 1981), quantitative reviews have reported that job attitudes have only modestly predicted actual turnover behaviors (Griffeth et al., 2000; Mathieu & Zajac, 1990; Tett & Meyer, 1993). To explore this issue I investigate a “threshold effect” – a critically low level of job attitude that, once reached, is likely to result in a decision to quit. I argue that there is a

4

systematic variation in threshold levels of job attitudes based on differences among employees in their demographic and work characteristics. In addition I investigate a “response bias” – systematic differences in the observed relationship between attitudes and turnover based on the tendency of certain employee groups to under- or over-report their true attitudes. I argue that these effects partially account for the moderate strength of the relationship between job attitudes and employee turnover. The second goal of this dissertation is to examine the consequences of turnover for the organization. More specifically, I examine one important path that employee retention may take in impacting organizational performance: to preserve and develop intangible organizational resources, such as human and social capital. High retention rates mean that the majority of an organization’s employees stay with the organization and fewer newcomers need to acclimate to the requirements of a new job and organizational culture. As members of a stable community, employees have many opportunities to develop a wide network of connections, build closer relationships, and understand the transactive memory of the organization. These firm-specific human assets (Coff, 1997) can be instrumental to organizational effectiveness, and can lead to a sustained competitive advantage (Barney, 1991). Based on this reasoning, is it correct to conclude that those organizations unable to retain their human and social capital through employee retention will under-perform? If so, what is the relative impact of each form of capital and to what extent are they substitutable resources? The first essay of this dissertation focuses at the relationship between job-related attitudes and actual turnover behavior. Job attitudes are core elements of most turnover theories (March & Simon, 1958; Mobley, 1977; Mobley et al., 1979; Steers & Mowday, 1981). These theories specify a strong and negative relationship between job-related attitudes and employee turnover

5

such that dissatisfied and less committed employees are more likely to leave. However, poor job satisfaction and organizational commitment do not necessarily result in subsequent turnover. In addition, most empirical studies report only a moderate association between work attitudes and turnover (e.g., Griffeth et al., 2000; Mathieu & Zajac, 1990; Tett & Meyer, 1993). Guided by this moderate association, I take a closer look at the differences among employees in the relationship between their attitudes toward their jobs (e.g., job satisfaction, organizational commitment) and their actual behavior (i.e., turnover). The literature on attitudes suggests generally that “the way the attitude is manifested depends upon certain situational pressures” (Wicker, 1969: 44). As such, the situational threshold leading to expressions of negative feeling (e.g., job dissatisfaction) may be lower than the threshold of actually quitting the job (Campbell, 1963). In addition to situational thresholds, employees may systematically differ in their tendency to under- or over-report true job attitudes. Such tendencies would eventually bias the true attitude-turnover links. By identifying systematic attitude thresholds and response biases attributable to observable employee profiles, this essay develops a model that explains systematic variability in the attitude-turnover link. I find that turnover rates for employees having higher attitude thresholds will be systematically greater than for employees having lower thresholds at identical attitude levels. Using hierarchical linear modeling I analyze how observable employee- and organization-related characteristics such as gender, tenure, age, education, ability, school size, school socio-economic status, and location are associated with systematic differences in teacher attitude thresholds and response biases in a sample of over 6,000 teachers in a large urban school district. The findings of the study point to gender, tenure, ability, and school socio-economic status as significant determinants of threshold

6

effects; and to ability, assignment to mandatory testing grades, and school environment as significant determinants of response biases. My second essay examines the role of human and social capital as mechanisms that may explain whether and why employee retention is associated with organizational performance. I argue that organizations with high turnover rates are likely to underperform because of their inability to accumulate substantial stocks of organization-specific human and social capital. In organizations with low turnover rates the majority of employees stay longer, thus having greater opportunities to develop: (a) organization-specific human capital through extended exposure to organizational problems and routines; and (b) stronger and better-informed relationships with their colleagues at work (i.e., social capital). Thus, through retention, organizations can accumulate stocks of human and social capital that comprise an idiosyncratic set of capabilities which, because of their uniqueness, can yield superior returns (Barney, 1991). In addition to examining the accumulations of these forms of intangible capital, I also look at their losses due to turnover. Particularly, I investigate both main and interactive effects of such losses on performance. I ask whether these two forms of capital are at least partially substitutable resources. Finally, I argue that the more closely the form of capital resembles the performance outcome of interest, the greater will be its ability to explain the effects of turnover on performance. Thus, I expect that task-specific forms of capital have more powerful effects on performance than do more general (i.e., organization-specific) forms. To address these issues, I collected archival and survey data to examine retention among public school teachers and conducted three sets of analyses. Studies 1 and 2 show a linear positive relationship between teacher retention and school performance that is fully mediated by organization-specific human capital and partially mediated by organization-specific social

7

capital. Study 3 examines the individual human and social capital of employees who remain with the organization versus those who leave. I find that losses of task-specific human and social capital are more deleterious to organizational performance than are losses of more general (i.e. organizational) forms of capital. Taken together, the findings described in the second essay suggest that employee retention is instrumental to organizational performance because it facilitates the development and maintenance of human and social capital. At the same time, the more specific such capital is to the work itself (i.e., task-specific), the stronger is its effect on performance. The findings from both essays of my dissertation provide important contributions to the two streams of research in the turnover literature: (a) research about the antecedents of turnover; and (b) research about the consequences of turnover. The first essay offers a framework that allows a closer examination of the link between job-related attitudes and actual turnover behavior. The second essay investigates the path from employee retention to organizational performance through organizational intangible assets such as human and social capital. A detailed discussion of specific contributions of each essay to the organizational literature are provided in the corresponding sections. Next, I present my first essay entitled “Job Attitudes as Predictors of Employee Turnover: Systematic Threshold Effects and Response Biases”. Following is the second essay entitled “Employee Retention and Organizational Performance: the Mediating Role of Organization- and Task-Specific Forms of Human and Social Capital”. Finally, I conclude with a discussion of the collective contributions and implications of findings presented in two essays.

8

2.0 ESSAY 1: JOB ATTITUDES AS PREDICTORS OF EMPLOYEE TURNOVER: SYSTEMATIC THRESHOLD EFFECTS AND RESPONSE BIASES Work attitudes, including job satisfaction and organizational commitment, have been recognized as important predictors of employee voluntary turnover in many theoretical models and empirical studies (Lee & Mitchell, 1994; Lee, Mitchell, Holtom, McDaniel, & Hill, 1999; March & Simon, 1958; Mobley, 1977; Mobley, Griffeth, Hand, & Meglino, 1979; Steers & Mowday, 1981). To illustrate, a recent meta-analysis on turnover by Griffeth, Hom and Gaertner (2000) incorporated 67 samples that included measures of job satisfaction, organizational commitment, and turnover. Theoretically, such attitudes should strongly predict employee turnover. Job dissatisfaction signals unhappiness with the status quo and the desirability of movement. According to a long tradition in the turnover literature, it should result in the employee quitting the organization when a reasonable alternative is present (March & Simon, 1958). Organizational commitment is viewed as a binding force that links an individual to the organization (Meyer & Herscovitch, 2001), and should thereby also reduce the likelihood of turnover. Despite these strong theoretical arguments, quantitative reviews of the literature (Griffeth et al., 2000; Hom & Griffeth, 1995; Mathieu & Zajac, 1990; Tett & Meyer, 1993) show that job satisfaction and organizational commitment are only moderately related to actual turnover behavior. In individual studies, the relationship of job satisfaction and organizational

9

commitment with turnover ranges from moderate (Blau & Boal, 1989; Mitchell, Holtom, Lee, Sablynski, & Erez, 2001; van Breukelen, van der Vlist, & Steensma, 2004), to low (Lee, Mitchell, Sablynski, Burton, & Holtom, 2004), to statistically non-significant (Cohen, 2000; Taylor, Audia, & Gupta, 1996). Clearly, there is variability among employees in actual turnover behavior even when they report very similar attitudes toward their jobs and organizations. In this paper I suggest that the strength of the observed link between job attitudes and turnover may systematically vary across employee groups with different characteristics. For some employees, self-reported attitudes are strongly related to subsequent turnover, while for others such attitudes are weakly associated with turnover. I propose two sources of systematic variability in the attitude-turnover link: (1) different attitude thresholds; and (2) systematic response biases. First, attitude measures may fail to account for underlying differences in the threshold beyond which employees actually leave their jobs. Employees may vary in their tolerance of low job satisfaction and organizational commitment. Thus, as explained in more details below, turnover rates for employees having higher thresholds, i.e., less tolerable of low attitudes, will be systematically greater than for employees having lower thresholds, i.e., more tolerable of low job satisfaction and organizational commitment, at identical attitude levels. Second, there may be systematic differences among employees in how accurately the measures of job satisfaction and commitment reflect true, underlying attitudes. Because of these, weaker attitude-turnover links will be observed overall. In this paper I report results from an analysis of attitudes and turnover of over 6,000 teachers in a large urban school district. I focus on observable employee characteristics and how they might be associated with systematic differences in employee attitude thresholds and response biases. My basic premise is that understanding systematic differences among

10

employees in how their attitude ratings translate into actual turnover may explain inconsistencies in previous research findings, and allow researchers to better model the relationship between attitudes and turnover in future studies. At the same time, the ability to identify differences in response patterns is important to management practice. Identifying employee groups whose relevant job attitude will more strongly predict turnover allows managers to focus on targeted rather than generalized interventions, which should be more effective. 2.1 THEORY Decision making about turnover is an inherently complex process (Lee & Mitchell, 1994), and the ultimate choice to stay or to leave is an interplay of many factors in addition to job attitudes. These include the future expected utility of turnover, normative pressures, moral and continuous attachment, psychological contracts, and the number and quality of perceived alternative opportunities (Maertz & Campion, 1998). Given this complexity, research has naturally sought to identify conditions under which the link between work attitudes and turnover is systematically enhanced or mitigated. For example, previous research has reported that job dissatisfaction predicts actual turnover more strongly in periods of economic prosperity when unemployment rates are low (Hom, Caranikas-Walker, Prussia, & Griffeth, 1992). Similarly, job dissatisfaction is a better predictor of quitting for employees with higher human capital (i.e. more educated both generally and occupation-specific) and higher cognitive ability (Trevor, 2001). Organizational commitment, in turn, has been found to be a stronger predictor of actual retention among younger rather than older employees (Cohen 1991, 1993).

11

Apart from factors like these that logically affect the strength of the relationship between attitudes and turnover, some methodological artifacts may also influence the observed magnitude of the relationship. For instance, organizational commitment was found to be a stronger predictor of turnover in studies with smaller samples and when the time that elapsed between measurement points was shorter (Griffeth et al., 2000). Similarly, job attitude-turnover correlations are sensitive to the choice of the attitude scale used (Cohen 1993; Griffeth et al., 2000; Tett & Meyer, 1993). When attitude data are collected through surveys, the respondents themselves can be an important source of methodological bias (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Specifically, respondents’ ratings may not fully and accurately capture their underlying attitudes. As summarized by Fazio and Zanna (1978, p. 399), “the underlying attitudes of two individuals with identical scale scores may differ in many other respects that may affect the relation of the attitude score to the behavior manifested by those individuals.” Two individuals with identical job attitude scores may have different rates of actual turnover behavior for two reasons. First, both employees may report their true job attitudes on the scale. However, for one of them, the reported level of the attitude is unacceptably low and he leaves the organization, while for the other, the same reported level of job attitude is quite acceptable and she stays. Thus, employees may accurately use the scale point to describe themselves as “dissatisfied” with their jobs, but not all of them will act on this dissatisfaction by quitting. This situation results from a difference in what I call job attitude thresholds, i.e. the minimum level of job satisfaction or organizational commitment that is acceptable to an employee in order to stay with the organization.

12

Second, attitude scores may imperfectly reflect true underlying attitudes. The actual job behavior – turnover – should be based on true attitudes (and not necessarily those reported). However, employees with identical attitude scores may have different rates of turnover because of systematic response bias, i.e., systematic over- or under-reporting of the underlying job attitude. For example, Arnold, Feldman and Purbhoo (1985) observed that social desirability bias attenuated the strength of the observed relationship between job attitudes and turnover. In the case of a single individual, it would be difficult to determine which of the two factors just described above may be at play: threshold differences or response biases in reported scores. However, if employees with similar characteristics (e.g., gender, work experience) demonstrate similar patterns in thresholds and/or in response bias, then in aggregate one can infer the expected strength of the job attitude-turnover relationship for particular employee groups. For example, if women tend to provide more biased job satisfaction scores and are more tolerant of job dissatisfaction compared to men, then we would observe a weaker relationship between satisfaction and turnover for them (because of response biases), and lower turnover rates overall (because of threshold effects). I elaborate on each of these phenomena, i.e. thresholds and response biases, below. 2.1.1 Work Attitudes and Turnover: Attitude Threshold Effects At its core, turnover is observed and modeled as an individual binary choice to stay with or to quit the current organization. Using job attitudes as predictors of turnover implicitly assumes the existence of individual threshold levels that, once reached, result in the employee leaving the organization. In other words, when the organizational commitment or job satisfaction falls below some threshold level, then an individual considers quitting. For example, modeling

Full document contains 136 pages
Abstract: My dissertation consists of two essays that examine employee turnover as an independent and a dependent variable. In my first essay I examine the relationship between job-related attitudes, such as job satisfaction and organizational commitment, and employee turnover behavior. In a sample of over 6,000 teachers in 200 public elementary schools, I found significant variability across employee and organizational characteristics in the strength of the relationship between job attitude ratings and turnover. I attribute this variability to two sources: (a) systematic differences in attitude thresholds, i.e. the minimum acceptable level of a job attitude necessary to remain with the organization; and (b) systematic response biases in attitude ratings. I develop a model to determine which characteristics relate to significant differences in attitude thresholds and/or response biasing factors and found that both individual and organizational attributes are distinguishing factors. In my second essay I present findings from three inter-related studies investigating human and social capital as mechanisms that may determine whether and why employee retention is associated with organizational performance. In a sample of public schools I found a linear relationship between employee retention and organizational performance. Further, this positive association is fully mediated by human capital and partially mediated by social capital. In addition, I examined human and social capital at the individual level of employees who remain with the organization versus those who leave. I found that organizational performance suffers most when the employees who leave have high levels of both human and social capital. Finally, I distinguished human and social capital losses based on their specificity: organization-specific versus task-specific. As predicted, the losses of task-specific human and social capital were more deleterious to organizational performance than the losses of organization-specific forms of capital.