Turnover

Boomerang Employees

"Boomerang employees" who quit but later return (Shipp, Furst-Holloway, Harris, & Rosen, 2014).

Future Directions in Turnover

1. Theorize and study change in turnover antecedents and consequences. 2. Investigate postturnover implications for employees and organizations. 3. Study distinct forms of—and motivations for—leaving and staying mindsets. 4. Expand turnover studies to better capture context. 5. Examine turnover management strategies and practices.

Realistic Job Previews and Retention

A third line of inquiry stemmed from rising awareness that effective recruitment and new hire assimilation can improve retention. Weitz (1956) furnished new hires with a booklet about insurance agent work and showed that this "realistic job preview" (RJP) boosted retention, a pioneering finding later replicated by Farr, O'Leary, and Bartlett (1973), who used work samples to reduce quits among sewing machine operators. These initial tests motivated a vast literature on RJP media, mechanisms, and moderators (Earnest, Allen, & Landis, 2011; Griffeth & Hom, 2001 Wanous, 1973). Though less influential, other articles demonstrated how orienting newcomers (Rosen & Turner, 1971) and recruiting them from certain sources (e.g., employee referrals; Gannon, 1971) curbed attrition.

Turnover Intentions is an Important Predictor of Turnover

Although Kraut (1975) first showed that quit intentions can foreshadow leaving, Mobley's theorizing firmly implanted this construct into turnover theory, claiming that such intentions represent the most proximal— and strongest—turnover antecedent (realizing that its predictive efficacy depends on time lag and measurement specificity). Given its predictive superiority (Griffeth et al., 2000; Rubenstein et al., 2015), turnover intentions have served as a surrogate or proxy for turnover when quit data are unavailable (Jiang, Liu, McKay, Lee, & Mitchell, 2012).

Steers and Mowday (1981) Model of Turnover

Critiquing the Mobley and Price- Mueller models, Steers and Mowday (1981) formulated a more comprehensive turnover process that (a) added new antecedents (notably, performance, other job attitudes) (b) identified moderators (e.g., nonwork causes, job search success) (c) explicated other ways to manage dissatisfaction besides quitting (e.g., change the situation, withdraw in other ways, cognitively reevaluate the job more favorably) (d) outlined feedback loops (e.g., dissatisfaction may prompt attempts to improve the job and if successful, upgrade one's attitudes) (e) specified multiple turnover routes (e.g., some employees quit without job offers, while others follow a "conventional path" by acquiring job offers before leaving).

Employee turnover

Employee turnover— employees' voluntary severance of employment ties (Hom & Griffeth, 1995)

Practical Suggestions for Managing Turnover

Employers can use validated selection procedures (e.g., biodata, personality, person-organizational fit) to screen out job applicants who might become prospective leavers. Employers should also pay special attention to on-boarding practices (including RJPs) as longstanding research has shown that most turnover occurs among new hires who face difficulty adjusting to the job. Organizations might monitor prominent causes underlying turnover (via surveys or personnel records), such as attitudinal trajectories (Liu et al., 2012) to foreshadow turnover or learn what (deteriorating) work conditions must be ameliorated to lessen potential turnover. Firms might also track turnover rate trajectories to project impending performance decrements (Call et al., 2015), use dashboards or scorecards to assess turnover costs, and capture data about who leaves (e.g., high performers, central actors in networks) and where they go (e.g., exit workforce, join competitors). Moreover, organizations might identify and assess the extent of reluctant staying and reluctant leaving (notably those due to external forces). While many firms assess job engagement (a symptom of reluctant stayers), they might also assess this mindset directly as other reasons besides poor job fit (e.g., poor organizational fit, abusive supervision) may occasion this state. Further, assessing reluctant leaving and its etiology (e.g., spousal relocation, unsolicited job offer) would help employers better prepare for future turnover (i.e., identify replacements beforehand) or how to counteract external forces for leaving (e.g., counter family pressures to leave by offering family benefits or decreasing work-family conflict by demanding less out-of-town travel).

Research on the Unfolding Model

Equally important, Lee and his colleagues (1996, 1999) pioneered qualitative methodology for validating turnover models. Based on interviews with leavers, they classified turnover cases into one of their turnover paths based on pattern matching. They determined that the majority of leavers followed one of four theorized paths, a finding often borne out by later investigations (Holtom et al., 2008). Shocks Drive Turnover More than Job Satisfaction (Holtom et al., 2008) Job embeddedness may buffer against shocks (Burton, Holtom, Sablynski, Mitchell, & Lee, 2010).

Family Embeddeness

Extending this theory cross-culturally, Ramesh and Gelfand (2010) thus validated the basic model in India but also advanced "family embeddedness," comprising a family's pride in a family member's employment in a company, the benefits a family derives from the company (e.g., health insurance), and family ties to company personnel. Unlike individualists who stay to fulfill self-interests, they claimed that Indian collectivists often join and remain in organizations to satisfy family needs, status, or obligations. In support, they found that family embeddedness explains unique variance in turnover in India but also in America.

Occupational Embeddedness

Feldman and Ng (2007) conceived "occupational embeddedness," identifying specific forces relevant to occupations, such as industry contacts, involvement in professional societies, compatibility with occupational demands and rewards, human capital investments, and occupational status. This embeddedness form does not necessarily promote loyalty to organizations as people embedded in professional fields may quit to practice or hone their professional skills elsewhere.

Collective Turnover and Turnover Rates

HRM practices reduce attrition via collective commitment (Gardner, Wright, & Moynihan, 2011), differentially affect quit and fire rates (Batt & Colvin, 2011), and lessen the effects of prior layoffs on quits (via embedding HRM practices; Trevor & Nyberg, 2008). Finally, the most thorough meta-analysis on antecedents of collective turnover to date identified many predictors besides HRM practices, such as climate, supervisory relations, and diversity (Heavey et al., 2013). A spate of recent primary studies and meta-analytic tests reveal stable negative associations between turnover rates and various dimensions of organizational performance (Hancock, Allen, Bosco, McDaniel, & Pierce, 2013; Heavey et al., 2013; Park & Shaw, 2013). Nonetheless, many of these correlations were derived from studies lacking a theoretical focus on turnover rates or collective turnover, which leaves ample opportunity for investigations that develop new theory (e.g., "turnover capacity," Hausknecht & Holwerda, 2013; "context-emergent turnover theory," Nyberg & Ployhart, 2013) and/or test existing or emerging models.

Boundary Conditions Around Effects of Collective Turnover

High supervisory and crew turnover at fast-food restaurants lower store sales and profits by prolonging customer wait time (verifying mediation via worsening customer service). Shaw and colleagues (2005) examined a different type of dysfunctional turnover—those occupying central niches in workplace communication networks—and revealed how turnover severs coworkers' relationships, which thereby disrupts communication networks, undermines social capital, and ultimately reduces productivity. According to their findings, these patterns are most harmful when stable relationships and social exchange among employees exist—when store attrition is low.

Thinking big and small about turnover

How does it affect things in startups? Declining industries? Does it work differently across different ranges of employee populations

Work by Hulin et al (1985)

Hulin et al. (1985) further argued that employees do not quit because they surmise job availability from local unemployment statistics. Rather, employees leave when they actually secure job offers. This astute observation coincided with later findings that many leavers do not seek jobs before leaving because they instead receive unsolicited job offers (a "shock," Lee et al., 1996; Lee et al., 1999) or are highly confident about obtaining jobs (after leaving) due to bountiful job opportunities in their field Hulin et al. (1985) additionally clarified that dissatisfaction does not inevitably culminate in leaving by noting that dissatisfied incumbents may lower job inputs (leading to psychological withdrawal) or improve their circumstances (via promotion or unionization) rather than leave (with or without job offers in hand). Later authors expanded the response taxonomy to include work withdrawal and (scarce) organizational citizenship as turnover alternatives or predictors (Chen, Hui, & Sego, 1998; Hulin, 1991), which may allow time for dissatisfying working conditions to ameliorate (e.g., promotions or transfers; Mobley et al., 1979).

Hobos and Spontaneous Turnover

Hulin et al. (1985) sought to explain why unemployment rates more accurately predict turnover than do perceived alternatives (Steel & Griffeth, 1989). They identified a workforce segment peripherally attached to the labor market and whose quit behaviors are poorly explained by conventional models. Calling them "hobos," these individuals freely drift from job to job, and may, when dissatisfied or bored, exit the labor market periodically to pursue more pleasurable or less stressful avocations. For them, the complex cognitive processes envisioned in standard turnover models (e.g., systematic search and rational analysis of jobs) are irrelevant; rather dissatisfaction (or wanderlust) translates directly into quits. Later researchers began identifying hobos (Judge & Watanabe, 1995; Woo, 2011) or spontaneous turnover paths that do not involve deliberate SEU calculations of the job or alternatives (e.g., script-based leaving, impulsive quits, or labor market exits; Lee et al., 1996; Lee et al., 1999; Maertz & Campion, 2004).

Mobley (1977) Process Model of Turnover

In the most influential single paper on turnover, Mobley (1977) elaborated a process model of how dissatisfaction evolves into turnover. He theorized a linear sequence: dissatisfaction ¡ thoughts of quitting ¡ evaluation of subjective expected utility (SEU) of job search and costs of quitting ¡ search intentions ¡ evaluation of alternatives ¡ comparison of alternatives and present job ¡ quit intentions ¡ quits.

Social Networks and Turnover

Krackhardt and Porter (1985) observed a "snowball effect" in which occupants of similar structural positions in communication networks often quit in clusters. Assessing strength of network ties, Krackhardt and Porter (1986) next showed that employees whose close contacts quit tend to form positive job attitudes, presumably to rationalize why they remain when friends exit. these findings foreshadowed mounting demonstrations that employees remain when they have strong or interconnected network ties (Feeley et al., 2008; Hom & Xiao, 2011; Mossholder, Settoon, & Henagan, 2005) or leave when their ties end (Felps et al., 2009).

Unfolding Model of Turnover

Lee and Mitchell (1994) put forth a radically new turnover theory known as the "unfolding model," challenging the prevailing paradigm. Departing from March and Simon (1958), they disputed three assumptions underlying their view—notably, (a) job dissatisfaction is a paramount turnover cause, (b) dissatisfied employees seek and leave for alternative (better) jobs, and (c) prospective leavers compare alternatives to their current job based on a rational calculation of their SEUs. To formulate a more valid and encompassing theory, they introduced various novel constructs, notably, "shocks" or jarring events (including external events) that prompt thoughts about leaving and drive alternative paths to turnover. Their model specifies four distinct turnover paths a conventional affect-initiated path (No. 4) in which dissatisfied employees quit after procuring job offers (e.g., Hom & Griffeth, 1991). (No. 1), some shocks activate a preexisting plan for leaving (matching script), inducing turnover (e.g., a woman quits once she becomes pregnant [the shock] because of preexisting plans to raise a child full time). (No. 2), negative job shocks violate employees' values, goals, or goal strategies (image violations, such as a boss pressuring a subordinate to commit a crime) and thus prompt them to reconsider their attachment to the company. Unsolicited job offers (a shock) induces a third path (No. 3), whereby employees compare offers to their current job and even seek additional jobs for further comparisons. In this path, one first quickly judges alternative jobs (unsolicited offers and those from a search) for compatibility with personal values or goals (image compatibility), screens out incompatible jobs, and then calculates SEUs for the feasible set of job offers (and present job). Echoing Hulin et al. (1985), Lee and Mitchell also upended traditional viewpoints by realizing that leavers do not always quit for other jobs. Rather, some Path 1 leavers exit the workforce for full-time schooling or stay-at-home parenting.

Maertz and Campion (2004) Turnover Framework

Maertz and Campion (2004) conceived an integrative framework outlining both how and why people quit. They identified different processes for four leaver types ("decision types") based on different motivational forces for leaving (impetuses for leaving, such as negative affect, perceived alternatives, or normative pressures). Example process types are "impulsive quit quitters" (those leaving without jobs in hand) and "preplanned quitters" (those leaving with a definite plan). Their decision types correspond to Lee and Mitchell's (1994) turnover paths but are not identical. Maertz and Campion (2004) differentiate between preplanned quitters (quitting when a specific time or event occurs) and conditional quitters (quitting if an uncertain event happens in the future)

March and Simon's (1958) theory of voluntary turnover

March and Simon's (1958) inaugural theory of voluntary turnover was a paradigmatic shift away from atheoretical research. Yet this revolution was delayed until publications by Mobley (1977; Mobley, Horner, & Hollingsworth, 1978) and Price (1977; Price & Mueller, 1981) who adopted March and Simon's (1958) central constructs—movement desirability and ease (defining them as job satisfaction and perceived job opportunities, respectively)—as cornerstones for more complex turnover models.

Job Embeddedness (Mitchell et al., 2001)

Mitchell et al. envisioned a causal indicator construct (or formative measurement model) comprising on-the-job forces for staying—namely, job fit, links, and sacrifices— as well as corresponding off-the-job forces (i.e., community fit, links, and sacrifices). Although some on-the-job forces (e.g., job sacrifices; Meyer & Allen, 1997) resemble prior constructs (e.g., costs of turnover; Mobley, 1977), community embeddedness captures turnover deterrents long neglected by prevailing thought (nonwork influences). Job embeddedness can attenuate shocks' deleterious consequences (e.g., higher quit intentions; Burton et al., 2010; Mitchell & Lee, 2001), while showing that employees whose colleagues or superiors are embedded are less quit-prone (Felps et al., 2009; Ng & Feldman, 2012). Apart from loyalty effects, Lee, Mitchell, Sablynski, Burton, and Holtom (2004) revealed that job embeddedness enhances job performance and organizational citizenship. Ng and Feldman (2010) noted declining social capital development among embedded incumbents, presumably because they had already amassed social contacts and felt less need to cultivate new ones. Ng and Feldman (2012) further documented that rising job embeddedness over time escalates workfamily conflicts. Finally, Huysse-Gaytandjieva, Groot, and Pavlova (2013) described how the experience of being trapped in a dissatisfying job ("job lock") impairs employees' mental health.

Mobley et al.'s (1979) Content Model of Turnover

Mobley et al.'s (1979) ground-breaking content model specified a large array of distal causes to clarify why people quit (e.g., disagreeable job features underlying job dissatisfaction, desirable attributes of alternative jobs). They introduced SEUs of the present job and alternatives which, along with job satisfaction, constitute proximal antecedents of search and quit intentions and mediate the impact of distal causes. Like prior scholars (Mitchell & Albright, 1972), expectancy theory was central to Mobley et al.'s (1979) theorizing. They argued that employees may stay in\ bad jobs because they expect eventual positive utility (e.g., promotions, desirable transfers), whereas employees may leave good jobs because they expect higher utility from other employment (performing a rational cost-benefit analysis to compare their job to alternatives). They further recognized that nonwork values and consequences of leaving moderate how job satisfaction and SEUs of the current job and alternatives underpin turnover.

Work Family and Turnover

Mobley et al.'s (1979) provisional ideas about "nonwork" influences resurfaced as more specific constructs as work-family conflict (i.e., employees opt out of paid employment to care for children; Hom & Kinicki, 2001) and "family embeddedness" (i.e., employees stay to avoid uprooting children or depriving families of corporate benefits; Feldman, Ng, & Vogel, 2012; Ramesh & Gelfand, 2010). Price and Mueller's "kinship responsibilities" construct advanced turnover understanding, which historically downplayed or neglected family influences on decisions to stay or leave. Standard theory (March & Simon, 1958) cannot readily account for family causes given the prominence accorded to job satisfaction and job alternatives (Abelson, 1987; Barrick & Zimmerman 2005). Price and Mueller (1981, 1986) thus conceived how kinship ties can deter turnover, which they captured with questions about number of children, marital status, number of relatives residing nearby, and the like (Blegen, Mueller, & Price, 1988). This construct foretold—if not directly shaped—subsequent inquiries into how families can initiate or impede quits, such as exiting for full-time elder care (Hom & Kinicki, 2001) or remaining to avoid loss of health benefits or first-rate schools for children (Feldman et al., 2012; Ramesh & Gelfand, 2010).

Methodological Improvements on Predicting Turnover

OLS regression not preferred SEM and Cox regression are better SEM explains covariance among explanatory concepts rather than variance in turnover

Effects of Turnover

Organizational researchers have shown that turnover disrupts various productivity-related outcomes (Hausknecht, Trevor, & Howard, 2009; Shaw, Gupta, & Delery, 2005) and reduces financial performance (Heavey et al., 2013; Park & Shaw, 2013). Other investigations documented how employees defecting to competitors can undermine their former employer's competitive advantage (via human or social capital losses or trade secret theft) or survival (Agarwal, Ganco, & Ziedonis, 2009). Finally, turnover has other side effects, such as hindering workforce diversity when women of color exit (Hom, Roberson, & Ellis, 2008) or spreading via turnover contagion (Felps et al., 2009).

Community Embeddeness

Pointing out the narrow scope of community embeddedness, Feldman et al. (2012) similarly conceptualize that family embeddedness in the community also matters—even to Americans—who may stay in a job or community they dislike because relocating would disrupt spousal careers or children's education. Mitchell et al.'s (2001) original view of community embeddedness thus underrepresents how families can embed employees (though their community embeddedness index taps employees' marital status and number of relatives living nearby) when families too are embedded in the organization or community (which Feldman et al. [2012] term "embeddedness by proxy").

Met Expectations Theory

Porter and Steers' (1973) met expectations theory asserts that job satisfaction and retention hinge on how closely a job fulfills employees' initial job expectations.

Organizational Commitment and Turnover

Porter and associates also conceived a new attitude—namely, organizational commitment—that can explain unique—if not more—turnover variance than do job satisfaction (Porter, Crampon, & Smith, 1976; Porter, Steers, Mowday, & Boulian, 1974). They argued that turnover implies the repudiation of organizational membership, not necessarily job duties that can be assumed elsewhere. Commitment is clearly inversely related to turnover and explains different portions of turnover variance than do job satisfaction (Hom & Griffeth, 1995; Klein, Cooper, Molloy, & Swanson, 2014).

Functional Turnover

Porter and his protégés (Dalton, Krackhardt, & Porter, 1981) introduced "functional turnover"—whereby the loss of surplus, low-quality, or costly labor can enhance organizational effectiveness. Whereas scholars and practitioners historically focused on turnover rates, Porter urged scrutiny of who quits given that high talent or performer turnover most harms firms. This conceptualization challenged the assumption that turnover was always dysfunctional

Turnover Theories by Price 1977; Price & Mueller, 1981, 1986, 2001

Price (1977; Price & Mueller, 1981, 1986) His theories emphasized key environmental drivers (revealed by his 1977 review) rather than attitudinal causes (which are not isomorphic; Weitz & Nuchols, 1955), yielding practical models identifying what managers can leverage to reduce turnover. His promulgation of objective environmental attributes (though he often used perceptual indices) also foreshadowed modern inquiry into external influences such as social cues (Felps et al., 2009), social networks (Feeley, Hwang, & Barnett, 2008), and community or family embeddedness (Mitchell & Lee, 2001; Ramesh & Gelfand, 2010).

Longitudinal Designs Help with explaining the attitudes turnover casual direction

Random coefficient modeling (RCM) find that job satisfaction's change trajectory explains additional variance in quit propensity beyond static satisfaction scores upholding the time-honored claim that attitudinal shifts predate leaving (Hom & Griffeth, 1991; Hulin, 1966; Porter et al., 1976). Liu et al. increased explained variance from 5% to 43% by moving from static measures of satisfaction (i.e., individual and work group scores) to dynamic measures (i.e., changes in satisfaction among individuals and work groups). Using RCM, Sturman and Trevor (2001) similarly established that performance velocity explains additional turnover variance beyond static performance scores. Deploying latent growth modeling (controlling measurement errors and instability), Bentein, Vandenberghe, Vandenberg, and Stinglhamer (2005) showed that a declining trajectory for affective organizational commitment predicts ascending quit intentions.

Research on some of the elements of the Steers Mowday Model

Researchers later established that job opportunity moderates how attitudes and quit intentions affect turnover (Carsten & Spector, 1987; Hom et al., 1992), amplifying their effects when employees can easily change jobs (Steers & Mowday, 1981). Subsequently, other scholars came to realize that dissatisfied incumbents can respond in other ways, such as avoiding work or reducing organizational contributions, before or besides leaving (Hom & Kinicki, 2001; Hulin, Roznowski, & Hachiya, 1985). Finally, contemporary investigations increasingly acknowledge alternative turnover paths other than the standard job-search ¡ job offers ¡ turnover route and impulsive quits (e.g., Lee, Gerhart, Weller, & Trevor, 2008; Lee & Mitchell, 1994; Maertz & Campion, 2004).

Multistage job search process Steel 2002

Steel (2002) put forth a multistage process through which employees move from passive scanning of the labor market to active solicitation of employers. His cybernetic theory described how job seekers progressively acquire more particularistic labor market information by selectively attending to certain information levels or sources and gaining feedback about job prospects and thus their employability. In support, he marshaled evidence that leavers' labor market perceptions better match labor market statistics (e.g., unemployment rates) than do stayers' perceptions, presumably because leavers actively pursue jobs and thus gather more valid labor market data. Steel (2002) also explained that individuals can exit without a job search when they have other income sources or receive unsolicited job offers, whereas others search to upgrade their current circumstances with counteroffers (not because they want to leave; Bretz, Boudreau, & Judge, 1994). Not fully tested

Turnover in the 1960-1970's research focused on Weight Application Blanks

These studies report predictive test validation for weighted application blanks (WAB; Buel, 1964; Cascio, 1976; Federico, Federico, & Lundquist, 1976; Schuh, 1967; Schwab & Oliver, 1974) and other selection tests (e.g., vocational interests, achievement motivation; Hines, 1973). During this renewal period, Schuh (1967) reviewed the accuracy of selection tests in predicting job tenure and concluded that WABs are most predictive because 19 of 21 studies showed that "some items in an applicant's personal history can be found to relate to tenure in most jobs" (p. 145). Given this endorsement, test validation research during this era largely focused on WABs (Federico et al., 1976). Whereas Schwab and Oliver (1974) disputed Schuh's validity conclusions, Cascio (1976) documented that WABs can have similar (moderate) predictive validity for Whites and minorities as well as mitigate adverse impact. Later work further attested to WABs' superior predictive efficacy over other selection tests (Hom & Griffeth, 1995). Yet narrative and quantitative reviews of early WAB tests overstated validity because findings were rarely cross-validated (Schwab & Oliver, 1974) and WAB studies often inflated turnover variance by creating equal-sized high and low-tenure comparison subsamples (i.e., generating artifactual 50% quit rate; e.g., Minor, 1958).

Early Research Findings on Turnover

Using a quasi-experiment, Hulin (1968, p.125) later concluded that a "company program initiated in 1964 brought about an increase in job satisfaction . . . and that this increase led to a reduction in turnover in 1966." Early investigations further reported that leavers more negatively perceive leaders (e.g., authoritarian, inconsiderate; Fleishman & Harris, 1962; Ley, 1966) and proximal environmental conditions (e.g., pay, shift work, performance reviews, underutilized capacity and talents; Hellriegel & White, 1973) than do stayers. Inspired by growing beliefs that dissatisfying work features induce leaving (Hulin, 1966, 1968), several scholars applied broader theories of work motivation or job attitudes—notably, motivator hygiene (e.g., Karp & Nickson, 1973), motivational needs (e.g., Hines, 1973), equity (e.g., Dittrich & Carrell, 1979), expectancy (e.g., Mitchell & Albright, 1972), and reasoned action (e.g., Newman, 1974)—to explain leaving.