The Effects Of Family Structure On Organizational Commitment, Intention To Leave And Voluntary Turnover [*].
In the following sections, the traditional psychological theory and research on the effects of family on voluntary turnover are reviewed. In particular ambiguities in this literature concerning the substantive content of family are highlighted. To suggest resolutions to these ambiguities, specific characteristics of family and possible processes by which these characteristics might operate on turnover are identified from sociological theory and research. Then, in turn, hypotheses are deduced, empirically tested, and results are reported. Finally, speculations, limitations and implications for future research are offered.
Traditional Psychological Theory and Research
Based on ideas summarized by Fishbein and Ajzen (1975), many models of voluntary turnover adopt the following as their core psychological linkage: (a) job attitudes [right arrow] (b) intention to leave (or stay) [right arrow] (c) actual leaving (c.f., Stroh et al., 1996). Moreover, most turnover theories and much of the corresponding empirical research can be differentiated by their varying emphasis, inclusion and placement within this core linkage of other specific turnover-relevant variables. For example, Gerhart (1990) investigated the direct and interactive effects of perceived ease of movement, unemployment rate and tenure on the linkages from job satisfaction to intention to stay to voluntary turnover. He found that perceived ease of movement directly and interactively (with job satisfaction) affected intention to stay, unemployment rate directly and interactively (with intention to stay) affected turnover, and tenure had direct effects only on intention to stay and turnover. Similarly, Stroh et al. (1 996) studied the direct and interactive effects of sex, family structure (i.e., dual earner status and children at home) and the "glass ceiling" on intention to leave and subsequent leaving. They found that sex, children at home, and the interaction of sex and glass ceiling were significant predictors of intention to leave.
Although some family variables are included in virtually all psychologically-oriented turnover models, theorists have not agreed upon which family characteristics are most relevant to quitting and how they might operate. For instance, Mobley's (1977) intermediate linkages model shows that "nonjob factors" (with "transfer of spouse" as his specific example) operate between job dissatisfaction and intention to leave via an effect on a sequential process of intention to search, search for alternatives and evaluation of alternatives. In their expanded version, Mobley et al. (1979) assert that "family responsibilities" affect individual values, which in turn affect intentions to search and to quit. Steers and Mowday (1981) theorize that "nonwork influences" interact with job attitudes to affect intention to leave. Price and Mueller (1986) argue that "kinship responsibilities" directly affect both organizational commitment and intention to leave. In Hom and Griffeth's (1995) recent integrative model, "conflicts wi th work" and "extraorganizational loyalties" are theorized to affect organizational commitment, which is antecedent to withdrawal cognitions and expected utility of withdrawal. In part, these theoretical differences across turnover models have lead researchers to study different family characteristics and different sets of antecedent and consequent variables. As a result, it is difficult to combine the empirical evidence into a coherent body of knowledge.
This Research Report builds upon the traditional theory and research by investigating the effects of family structure on the psychological linkages among organizational commitment, intention to leave and voluntary turnover. Organizational commitment was adopted as the focal job attitude because of its consistent correlation with marital status, intention to leave and leaving, and its theorized connection to other family variables in many turnover models. Further, this Research Report extends traditional psychological theory and research by drawing from sociological approaches to suggest specific family characteristics and possible social processes by which these family characteristics might operate on voluntary turnover.
In the sociological literature, there has been a long history of archival-based research on the effects of family structure on a host of behavioral outcomes. For example, Wu and Martinson (1993) operationalized family structure as "intact," "mother-only," "one biological parent only" and "all others," and studied its effects on the risk of premarital births with data from the 1987 National Longitudinal Survey of Youth. Downey (1995) operationalized structure as the number of siblings and investigated its effect on educational performance with data from the 1988 National Educational Longitudinal Study. Biblarz and Raftery (1993) operationalized family as "intact" versus "disrupted" (e.g., divorced, blended) and researched its effects on intergenerational social mobility with data from the 1973 Occupational Changes in a Generation Survey. Although variously operationalized, whether there is a spouse (e.g., Biblarz and Raftery, 1993) and children at home (e.g., Stroh et al., 1996), and whether the spouse is empl oyed (e.g., Wu, 1996) may be the most commonly studied and well understood characteristics of family structure. Because turnover theorists and researchers have not agreed upon which specific family characteristics are most important to the leaving process, this Research Report proposes that these three well researched characteristics are reasonable starting points.
Broadly speaking, family structure may affect voluntary turnover via two related processes: First, families vary in the social control over its members. It has also been shown, for example, that two-parent households can more easily supervise children than single-parent situations. Second, family structure directs the members' allocation of financial and other human resources (i.e., time and effort) of its members. It is the theory and research on social control and allocation processes that most directly informed this study's hypotheses. The empirical evidence indicates that family structure can affect individual behavior via its social control of members (Thornton, 1991) and its direction of members' time and energy (Biblarz and Raftery, 1993; Downey, 1995). In addition, Becker's (1985, 1991) human capital theory specifically argues that, because of limitations to one's time and energy, employees must economize between work and family. That is, employees make overt and volitional choices about the time and energy spent in work versus family roles. Based on the empirical evidence and human capital theory, family structure is suggested to affect voluntary turnover by increasing social pressures (controls) in the allocation of time and energy devoted toward (or away from) the job (or family). More specifically, increasing the social pressures in allocation decisions should systematically strengthen the connection between family structure to the linkages among organizational commitment, intention to leave and voluntary turnover. Thus, this study's basic argument is that family structure increases social pressures on its members' allocation decisions of the time and energy devoted to the job (or family).
Hypothesis 1. Having a spouse (H1a), having an employed spouse (H1b) and an increasing number of children at home (H1c) strengthen the effect of intention to leave an subsequent and actual leaving.
Organizational commitment is often defined as an individual's acceptance of the company's goals and values, willingness to exert considerable effort on the company's behalf and a desire to maintain membership (Mowday et al., 1982). Organizational commitment enters into a motivational and decision-making process that produces an intention to leave (Home and Griffeth, 1995). Conceptually, for instance, higher organizational commitment implies weaker desire (or motivation) to leave the company and results in lower intention to leave. Indeed, the empirical evidence speaks clearly on this point: meta-analytic research reveals a corrected mean correlation of --.28 between organizational commitment and intention to leave (Mathieu and Zajac, 1990). Although not offered as a formal hypothesis, a negative correlation between organizational commitment and intention to leave is expected from this study's data. At issue here is whether having a spouse, having an employed spouse or an increasing number of children at home create conditions that strengthen or weaken the negative effect of organizational commitment on intention to leave. As argued, these structural characteristics increase the social pressures that signal more time and energy must be devoted to (or away from) the family or the job. By such signaling (or directing) the allocation of one's time and energy, family structure serves to create external, contextual conditions that should weaken the negative effect of organizational commitment on intention to leave.
Hypothesis 2. Having a spouse (H2a), having an employed spouse (H2b) and an increasing number of children at home (H2c) weaken the negative effect of organizational commitment on intention to leave.
Sample and Data. Archival data for three large samples of US Navy (USN) officers were drawn from official military records. Seven variables were originally obtained as part of the USN's ongoing survey research programs that routinely monitored officers' cognitions. Four other variables, which included this study's criterion and two of its control variables, were taken directly from official USN personnel records. These data were made available to the authors by civilian researchers at USN's Personnel Research and Development Center in San Diego, California. The Surface Warfare Officers (SWO) are the organization's "front-line personnel" and perform the Navy's primary sea-based fighting mission. Informally, they are often considered to be the USN's core officer group. The sample consisted of 3,129 surface warfare officers, who had an average age of 33 years (SD=6) and tenure of 10 years (SD=6); 3.5% (or 110) were nonwhite or female, and all others were white males. Their survey data were collected in August, 1981.
The Aviation Warfare Officers (AWO) are the organization's specialists in aviation and perform the Navy's airpower-line function. Informally, members of the AWO community are often described as having very high self-confidence, pushing rules, machines and equipment to their limits of performance, and enjoying a strong esprit de corps. The sample consisted of 5,051 aviation warfare officers, who had an average age of 33 years (SD=5) and tenure of 9 years (SD=5); 1.7% (or 89) were nonwhite or female, and all others were white males. Their survey data were collected in February, 1981.
The General Unrestricted Officers (GUO) are the organization's staff personnel and perform such services as human resource management, accounting or purchasing. Informally, members of the GUO community are said to be equivalent to the staff personnel of any large organization. The sample consisted of 1,201 general unrestricted officers, who had an average age of 30 years (SD=5) and tenure of 5 years (SD=5); 23% (or 278) were nonwhite or male, and all others were white females. Their survey data were collected in May, 1982.
Analysis. Methodologists who specialize in turnover research suggest that survival analysis is often more appropriate for turnover data than many other common methods (Morita et al., 1993). As such, the Cox proportional hazards model was applied to test Hypothesis 1. In contrast, Hypothesis 2 was tested with OLS regression. For these OLS regressions, involuntary leavers and retirees were omitted from the analysis. To convey a sense of effect sizes and to ease interpretations for the predicted interactions, a three-step subgrouping process was followed when the regression coefficient for a hypothesized moderator variable was statistically significant. First, the samples were split along lines indicated by the moderator. For example, when having a spouse was a significant moderator, the samples were separated into "have a spouse" and "no spouse" groups. Second, the regression equations, which included the same control variables and main effects, were recalculated for each sub-sample. Third, effect sizes can be inferred by the comparison of the differences in corresponding regression coefficients (and correlations in hypothesis 2 only).
The Measurement Window. The time between collection of survey and departure data spanned 70, 76 and 61 months for the SWO, AWO and GUO, respectively. Although long, such a "measurement window" was necessary because many of the study's participants were recently commissioned officers and had a 4-6 year mandatory service obligation. It is after this initial service obligation that voluntary turnover can readily occur. Whereas the USN was not required to allow an officer's voluntary resignation during the mandatory service period, the USN did typically fulfill such requests among its officers. (Despite this long measurement window, substantial predictive effects were found for intention to leave. Thus, intention to leave appears to be a very robust predictor of subsequent leaving.)
Control Variables. Although variance on race and sex was limited, they represented "noise to the system" and were statistically controlled. Because of the military's generous retirement plan, tenure may have a particularly strong effect; in addition, recent meta-analytic research shows a modest correlation between tenure and turnover (Hom and Griffeth, 1995). Thus, tenure was also statistically controlled in the OLS regressions. With survival analysis, tenure was incorporated into the outcome variable. Furthermore, the -survival models tested in hypothesis I may be underspecified. Because of pervasive influence in turnover research, it can be argued that March and Simon's (1958) perceived ease and desirability of movement should be included as control variables (e.g., Trevor et al., 1997). Moreover, it can be argued that organizational commitment, alone, may be a too narrow representation of Fishbein and Ajzen's (1975) notion of "job attitudes." As part of the post hoc analyses, under-specification was teste d. Measures for the perceived ease of movement and for naval career satisfaction, which represented a surrogate for both the perceived desirability of movement and expanded coverage of the job attitude domain, were added as control variables. Finally, the line versus staff distinction was mechanically controlled by separately analyzing data from each sample.
Voluntary Turnover and Survival Duration. From official USN personnel records, an individual's entry date and employment status (e.g., stayer, leaver, retiree) as of June, 1987, were obtained. For those individuals who had left the USN, their personnel records included whether their departure was voluntary or involuntary, as well as their date of departure. In the descriptive statistics, turnover was dummy coded, with 0 representing stayers and 1 representing voluntary leavers. For the substantive survival analysis, an individual's employment duration was the time between one's official entry and departure from the USN. These durations allowed estimation of the conditional probability of leaving (i.e., the hazard function), which is the model's outcome variable (Morita et al., 1993). Information from retirees and involuntary leavers were incorporated into the analysis with the Cox model as censored data.
Intention to Leave. The intention to leave was measured by the question, "How certain are you that you will continue an active Navy career at least until you are eligible for retirement?" The response options had lower values indicating an intention to stay and higher values indicating an intention to leave, and they were as follows.
1.) 99.9-100% I am virtually certain that I will not leave the Navy voluntarily prior to becoming eligible for retirement.
2.) 90.0-99.8% I am almost certain I will continue my military career if possible.
3.) 75.0-89.9% I am confident that I will continue my Navy career until I can retire.
4.) 50.0-74.9% I probably will remain in the Navy until I am eligible for retirement.
5.) 25.049.9% I probably will not continue in the Navy until I am eligible for retirement.
6.) 10.0-24.9% I am confident that I will not continue my Navy career until I can retire.
7.) 0.2-9.9% I am almost certain that I will leave the Navy as soon as possible.
8.) 0-0.1% I am virtually certain that I will not voluntarily continue in the Navy until I am eligible for retirement.
Organizational Commitment. Organizational commitment was measured with averaged composite scores from the Organizational Commitment Questionnaire (Mowday et al., 1979). Alpha reliabilities for the SWO, AWO and GUO were .88, .87 and .91, respectively.
Having a Spouse. Via a check-list, respondents were asked to indicate from the following: (a) "Never married," (b) "Married," (c) "Widow(er)," (d) "(widowed &) Re-married," (e) "Divorced," and (f) "(divorced &) Remarried." Although not included in the present study, respondents also wrote-in the year that the checked event occurred. These dates helped identify questionable data. For example, no date should appear with "never married," and only one date should appear with "married." Because the study's arguments involve being single or married, these categories were collapsed into a dummy variable. Zero represented being single, and 1 represented having a spouse.
Having an Employed Spouse. Via a check list, respondents were asked, "How is your spouse primarily employed?" The listed options were as follows.
1.) Full-Time Homemaker
6.) Other Professional
9.) Navy Officer
10.) Navy Enlisted
11.) Other Military
12.) Other ___
Respondents who were not married were instructed to omit this item. Because the study's arguments involve having an employed spouse, these categories were collapsed into a dummy variable. Zero represented having no spouse or having checked "1.) Full Time-Homemaker," and 1 represented having an employed spouse, which resulted from checking options 2-12.
Children at Home. Via fill-in, respondents indicated the number of children
living at home. Although not included in the present study, respondents had also provided the ages of these children. These data also helped to identify questionable data. For instance, the reported number of children should be equal to the number of ages provided.
Controls Variables. Race was dummy coded, with 0 representing whites and 1 representing nonwhites. Sex was dummy coded, with 0 representing males and 1 representing females. Tenure was operationalized as the number of years since official entry into the USN. Perceived ease of movement was measured by the following item, "If you were to seek civilian employment, how prepared are you to do so?" This was answered on a seven-point scale anchored by 1-essentially unprepared to 7-essentially prepared. Naval career satisfaction was measured by an averaged composite score for the following six items, which were developed by USN personnel researchers. These items were: "I would be very dissatisfied if I had to change my career."; "The morel think about it, the more I feel I made a bad move in entering my career." (reverse scored); "I thoroughly enjoy my career."; "I take great pride in my career."; "I would definitely like to change careers." (reverse scored); and "I feel I could be much more satisfied in a different career." (reverse scored). These items were answered on a seven-point scale anchored by 1-strongly disagree to 7-strongly agree. Factor analysis on these items indicated a unidimensional structure, and the alpha reliabilities for the SWO, AWO and GUO, respectively, were .87, .75 and .78.
Means, standard deviations and correlations for all variables are reported in Tables 1-3 for the SWO, AWO and GUO, respectively. The correlations between intention to leave and subsequent turnover for the SWO, AWO and GUO were, respectively, .50, .46 and .39 (all p[less than].001). These values exceed the corrected mean correlation between intention to leave and turnover reported in a recent meta analysis but were within one standard deviation (Hom and Griffeth, 1995). The correlations between organizational commitment and turnover for the SWO, AWO and GUO were, respectively, -.30, -.26 and -.29 (all p[less than].OOl). These values are quite close to the corrected mean correlation between organizational commitment and intention to leave reported in another meta analysis (Mathieu and Zajac, 1990). Table 4 shows the survival equations when the hazard function is regressed onto the specified model across the 3 samples. With other effects held constant, intention to leave had a positive and predictive effect (ov er time) on turnover in all 3 samples, which replicates their well documented relationship. Moreover, organizational commitment had no main effect on survival duration when the effect of intention to leave was held constant, which is also consistent with many turnover theories. Table 5 shows the OLS regressions when intention to leave is regressed onto the specified model across the 3 samples. With other effects held constant, organizational commitment was negatively associated with intention to leave in all 3 samples, which also replicates their well documented relationship. When considered together, this study's data and results appear consistent with existing empirical evidence.
The major assumption for the Cox model is the constant proportionality of hazard functions. This assumption was checked with Morita etal.'s (1993: 1446-1449) recommended procedure. For each predictor variable, the Kaplan-Meier product limit estimate of the survivor function and the natural logarithm estimate of the cumulative hazard function were graphed. For dummy variables, these functions were graphed by the groups coded 0 versus 1; for continuous variables, these functions were graphed for the groups split at the median. Visual inspection of the resulting graphs indicated no major violation. Because the Cox model is quite robust, only major violation of the proportionality assumption is considered sufficient to reject its use (Morita et al., 1993). Thus, the assumption of constant proportionality was judged appropriate.
The effect of intention to leave on actual leaving was hypothesized to strengthen when one has a spouse (Hla), an employed spouse (H1b) and an increasing number of children at home (H1c). The interaction between intention to leave and having a spouse was statistically significant among the SWO and GUO, but the effects were in opposite directions (Table 4). For the SWO, the effect of intention to leave on leaving was stronger among married (than single) people. More specifically, the effect of intention to leave on leaving was exp (b) = 2.27 (p[less than].001) for married people and exp (b) = 1.67 (p[less than].001) for single people. In contrast, the effect was weaker among married people (exp (b) 1.53, p[less than].001) than single persons (exp (b) = 1.66, p[less than].001) for the GUO. The interaction between intention to leave and having an employed spouse was statistically significant only among the GUO, with the effect of intention to leave on leaving stronger among people with an employed spouse (exp ( b) = l.69, p[less than].001) than persons with a nonemployed spouse (exp (b) = 1.54, p[less than]001.). The interaction between intention to leave and an increasing number of children at home was statistically significant among the SWO and AWO, with the effect of intention to leave on leaving strengthening as the number of children at home increased. For the SWO and AWO with no children at home, the effects of intention to leave on leaving were, respectively, exp (b) = 1.76 and 1.91 (both p[less than].001). In contrast, the effects of intention to leave on leaving for the SWO and AWO with children at home were, respectively, exp (b) = 2.46 and 2.26 (both p[less than].001). In sum, hypotheses la was partially supported in one sample, hypothesis 1b was supported in one sample, and hypothesis 1c was supported in two samples.
The negative effect of organizational commitment on intention to leave was hypothesized to weaken as a function of having a spouse (H2a), having an employed spouse (H2b) and children at home (H2c). The interaction between organizational commitment and having a spouse was statistically significant only among the SWO (Table 5). More specifically, the negative effect of organizational commitment on intention to leave was weaker among married people (b = -.73, r = -.45, p[less than].001) than for single persons (b = -.1.12, r = - .58, p[less than].001). The interaction between organizational commitment and an increasing number of children at home was statistically significant among the SWO and AWO. Specifically, the negative effect of organizational commitment on intention to leave was weaker as the number of children at home increased. For the SWO and AWO with no children at home, the effects of organizational commitment on intention to leave were, respectively, b = -1.18 (r = -.61) and b = -1.03 (r -.54; all p [less than].001). In contrast, the effects of organizational commitment on intention to leave for the SWO and AWO with children at home were, respectively, b = -.57 (r = -.36) and b = -.61 = - .38; all p[less than].001). In sum, hypotheses 2a was supported in one sample, hypothesis 2b was not, and hypothesis 2c was supported in two samples.
Post Hoc Analyses. The previously specified survival equations (from hypothesis 1 and shown in Table 4) were recalculated, but with the perceived ease of movement and career satisfaction added as control variables. Consistent with prevailing turnover theories, the perceived ease of movement was positively predictive of turnover among the AWO and GUO, and career satisfaction was negatively predictive of turnover among the AWO and GUO. For the substantive analysis involving the predicted interactions from hypothesis 1, the same moderators, with comparably sized regression coefficients, were statistically significant as previously found. Thus, the earlier survival model does not appear under-specified.
This Research Report responded to appeals by applied psychologists to study work-family interfaces and to understand better the effects of families on voluntary turnover. Although virtually all psychologically-based turnover models include some family variables, researchers have not agreed upon which family characteristics are most relevant to quitting and the processes by which they operate on quitting. As a result, accumulation of knowledge has been problematic. Drawing from sociological theory and research on the effects of family structure, having a spouse, having an employed spouse and an increasing number of children at home were identified as potentially meaningful antecedents in the turnover process. In addition, the sociological evidence suggests that these family characteristics can control its members' behaviors by exerting social pressures and prompting allocation decisions on the time and energy available between work and family. This study's basic argument is that family structure increases the systematic social pressures on its members' allocation decisions of the time and energy devoted to the job or family and that these systematic social pressures produce testable effects on the linkages among organizational commitment, intention to leave and turnover.
Hypotheses were deduced and tested on archival data obtained for two samples of USN line officers (SWO and AWO) and one sample of USN staff officers (GUO). It was found that having a spouse and an increasing number of children at home weakened the negative relationship between organizational commitment and intention to leave in the SWO and AWO samples. Furthermore, an increasing number of children at home strengthened the predictive effect of intention to leave on subsequent leaving among the SWO and AWO. Having an employed spouse strengthened the predictive effect of intention to leave on leaving among the GUO. Whereas having a spouse strengthened the predictive effect of intention to leave on leaving among the SWO, it, however, weakened the predictive effect among the GUO. To explain this contradiction, we consulted representatives of the USN. Albeit speculative, they suggested that SWO were most often the families' primary income earner, while the GUO were most often the families' secondary income earner. As a result, leaving among the GUO was seen as more difficult than for the SWO. When taken together, family structure appears to be a meaningful contextual factor that systematically affects the well documented linkages among organizational commitment, intention to leave and voluntary turnover. It now seems timely for future research to aim at better understanding the processes by which family structure operates on voluntary turnover.
Suggested Theoretical and Empirical Implications
Over the years, turnover theorists have relegated family variables to tangential roles, with a possible trend toward relegating nonwork influences even further from the act of leaving. For example, Mobley's (1977) early intermediate linkages model had family variables operating directly on the intention to search, which can be seen as a secondary role. In the expanded version, Mobley et al. (1979) relegated family variables to affect the leaving process primarily through individual values, which would appear to move family into a tertiary role. In a recent turnover model, Hom and Griffeth (1995) theorized that family variables were subsumed by two larger summary concepts, which were themselves aggregated into a larger summary construct, called antecedents to commitment. These antecedents to commitment were asserted to affect withdrawal cognitions, which were, in turn, theorized to influence turnover. Thus, the Ham and Griffeth model appears to relegate the effects of family characteristics beyond a tertiary role. The results of this Research Report should call this trend into question and suggest that the effects of family occur closer to the act of quitting than traditionally thought. Moreover, turnover theorists have not agreed upon which specific family characteristics are most important to the leaving process. Our results suggest that having a spouse, having an employed spouse and an increasing number of children living at home might merit such consideration.
Limitations, Cautions and Possible Future Research Directions
In this Research Report, an initial "coarse-grain, first step" was taken towards delineation of some of the contextual factors that operate on turnover. In particular, family structure was conceptualized as an external, contextual factor that creates moderating conditions for the well documented linkages among organizational commitment, intention to leave and leaving. Social control and allocation decisions are established processes in the sociological research on families, and led to hypothesized predictions that connect family structure to the psychological withdrawal process. Nonetheless, it is critical to note that these theoretically established processes were unmeasured in this study. Given the empirical confirmation for many of this study's predictions, it seems timely and prudent for theorists and researchers to move toward "finer-grained" studies. Future empirical studies should include measures of the psychological linkages, family structure and the theorized process variables as well.
Although many of the hypothesized effects were empirically supported, this study presented a conservative view of the family that may have weakened the obtained results. Future researchers might take a less conservative view and, in doing so, may strengthen their empirical results. In particular, this study's conceptual arguments focused on being single versus married. Turnover researchers may need to expand their range of marital states and, thereby, capture more variance from this distinction. For example, cohabitating "unmarried people of the opposite sex," "unmarried people of the same sex" and "married people of the same sex" may need to be included. Furthermore, this study held a static view of the family. Although the hypothesized interactions predicted turnover up to six years later, improved prediction of leaving should be expected if future researchers considered the effects of changes to family structure on quitting. Simply put, ongoing family structure may be less salient to an individual than a major disruption to family structure. Whereas having children may gradually engage the psychological leaving process, the recent birth of a child seems likely to engage the withdrawal process more forcefully and quickly (Lee and Mitchell, 1994). Hence, stronger effects might be found. Finally, this study took a limited view on the nature of the spouse's employment; it was taken to be employed or not employed. Future researchers might take a broader view and access more variance in this distinction. They might consider whether the spouse's job is professional, career oriented, in the secondary labor market or full time. As speculated above, it would be informative to document which spouse is the primary income earner as well.
Intention to leave, having a spouse, having an employed spouse and children at home were measured with single items with unknown reliabilities. However, there may be empirical reason to believe that these single items may be sufficiently valid for research purposes. First, the single item for the intention to leave was found to predict voluntary turnover longitudinally and, second, to correlate with scores from the Organizational Commitment Questionnaire in three samples. Third, the single items for having a spouse and children at home were found to predict turnover longitudinally and interactively in two samples, and having an employed spouse was found to predict turnover longitudinally and interactively in one sample. Fourth, the single items for children at home and having a spouse correlated directly or interactively with intention to leave in two samples. In addition, there may be conceptual reason to believe that the single items for having a spouse, having an employed spouse and children may be suffici ently reliable. These single items assess what have been called "well rehearsed" concepts. In particular, Schmitt and Hunter (1996) suggest that such "well rehearsed" single item measures are highly predictable and likely have so little measurement error that any such error can be reasonably ignored. When taken together, the single-items measures adopted here appear sufficiently reliable for research purposes.
Because this study's data were collected during the early to mid 1980s, it is reasonable to consider whether our basic argument -- that family structure induces systematic social pressures on allocation of time and energy between the job and family -- holds true today. We suggest that these social pressures and allocation decisions, which occurred in the 1980s, occur in the 1990s and will likely occur in the next decade as well. More specifically, these arguments, and their corresponding empirical tests, are not time-based. Consider the following data from the US Bureau of the Census (Farley, 1996): (1) the number of dual-earner families increased from approximately 25 million in 1980 to 31 million in 1995, (2) from 1979 to 1989, the percentage increase in work force participation for men was .5%, but it was 25.6% for women, and (3) from 1980 to 1990, the percentage change for families in poverty was zero. These (and similar) statistics give reason to expect that family and work pressures are at least as grea t today as they were when this Research Report's data were collected. However, the non-time-dependence of this study's arguments is an issue of empirical replication. It may be that certain subprocesses (e.g., gender- or racebased roles), which determine a moderator's directionality, are time-dependent. The potential time-dependence of specific subprocesses remains an open theoretical (and empirical) issue, which should be examined through further empirical research.
Although the three samples were reasonably large and perform distinct organizational functions, all persons were members of the US Navy. As such, the generalizability from military to civilian populations (and the reverse as well) may be of some concern. Over the decades, turnover researchers have often studied military samples (e.g., Hom et at., 1979; Steel, 1996). While not eliminated, concern may be somewhat lessened by recent meta analytical findings that indicate no significant differences in the correlation between intention to leave and turnover between military and civilian samples (Hom et at., 1992: 898; t = .51, ns, n = 2,058 military and n = 2,955 civilian personnel). Nonetheless, it remains an empirical issue whether this study's basic theoretical argument--that family structure induces systematic social pressures on its members' allocation of time and energy devoted to the family, which in turn affects organizational commitment, intention to leave and leaving--holds across different populations.
Finally, researchers might study additional dimensions of organizational commitment. In the present Research Report, the equivalent of "affective commitment" was studied. Recent conceptual developments and corresponding empirical research have identified the additional dimensions of "continuance" and "normative" commitment (Meyer and Allen, 1997). A theoretical and empirical issue is whether continance and/or normative commitment interacts with family structure to affect voluntary turnover.
For managers, these results suggest the advisability of policies and practices that support the family structure as a means of reducing the effects of low organizational commitment while simultaneously enhancing employees' ability to cope with family issues that could otherwise exacerbate the likelihood of turnover. For instance, employers might influence the leaving process by implementing flextime work schedules, employee assistance plans, insurance coverage for family counseling, and/or active support for family and medical leaves. Such programs could facilitate an employee's involvement in the spouse's and/or children's activities (e.g., school, sports, music) and thereby reduce the desire to leave the firm in order to lessen work-family conflicts. In essence, these practices could increase the "family" as a reason to stay while simultaneously decreasing perceived work/family conflicts as reasons to leave.
A specific example of a managerial intervention that could increase family-related reasons for staying while simultaneously decreasing family-related reasons for leaving is the firm's childcare policy. Recent survey data indicate, for instance, that approximately 12% of US companies, which employ more than 100 persons, offer childcare assistance (Bureau of Labor Statistics, 1998; Kossek and Nichol, 1992). In addition, the US government has generally adopted a "hands-off" policy on day-care (Gomez-Mejia, et al., 1998). In our view, considerable potential exists for firms to implement "family friendly" child-care policies that may decrease child-care conflicts as a reason to leave, while simultaneously serving as a reason to stay. Based on over 1,500 interviews, for example, Grover and Crooker (1995) reported that the availability of child-care was inversely related to the turnover intentions among parents with young children. More specifically, they found that employees often interpreted the firm's child-care policy as a reason to stay because such policies reduced felt pressures to leave.
Recently, Lee and Mitchell (1994) described the research on employee turnover as cumulative normal science. Management scholars know a great deal about how individuals decide to leave. Nonetheless, they suggested that turnover researchers may have become overly focused on the internal cognitions which immediately precede quitting and that much more could be learned if some scholarly attention was redirected toward variables more external to the employee's cognition. This study may be taken as support for the recommendation to reorient at least some of the turnover research. As Zedeck (1992) observed, families matter and families affect work behavior. Yet not a great deal is known about the work-family interface. Furthermore, he suggested that much could be learned if some scholarly attention was redirected toward the relationships between family and organizational behaviors. It is hoped that this Research Report represents the onset for substantially more scholarly inquiry into the family-work-turnover inter faces and for the merging of two important research traditions.
(*.) We thank Drs. Rodger Griffeth (Georgia State University) and Terry Mitchell (University of Washington) for their generous help on prior versions.
Becker, G.S. 1991. A Treatise on the Family. Cambridge, MA: Harvard University Press.
___, 1985. "Human capital, effort and the sexual division of labor." Journal of Labor Economics 3: 33-58.
Biblarz, T.J. and A.E. Raftery. 1993. "The effects of family disruption on social mobility." American Sociological Review 58: 97-109.
Bureau of Labor Statistics. 1998. Employee Benefits Survey. Washington, DC: US Department of Labor.
Downey, D.B. 1995. "When bigger is not better: family size, parental resources and children's educational performance." American Sociological Review 60: 746-761.
Farley, R. 1996. The New American Reality. New York, NY: Russell Sage Foundation.
Gerhart, B. 1990. "Voluntary turnover and alternative job opportunities." Journal of Applied Psychology 75: 467-476.
Fishbein, M. and I. Ajzen. 1975. Belief, Attitude, Intention and Behavior. Reading, MA: Addison & Wesley.
Gomez-Mejia, L.R., D. Balkin and R.L. Cardy. 1998. Managing Human Resources (second edition). Upper Saddle River, NJ: Prentice Hall.
Grover, S.L. and K.J. Crooker. 1995. "Who appreciates family-responsive human resource policies: the impact of family-friendly policies on organizational attachment of parents and non-parents." Personnel Psychology 48: 271-288.
Hom, P.W., F. Caranikas-Walker G.E. Prussia, and R.W. Griffeth. 1992. "A meta-analytical structural equations analysis of a model of turnover." Journal of Applied Psychology 77: 890-909.
___ and R.W. Griffeth. 1995. Employee Turnover. Cincinnati, OH: Southwestern.
___, R. Katerberg, and C.L. Hulin. 1979. "Comparative examination of three approaches to the prediction of turnover." Journal of Applied Psychology 64: 280-290.
Kossek, E. and V. Nichol. 1992. "The effects of on-site child care on employee attitudes and performance." Personnel Psychology 45: 485-510.
Lee, T.W. and T.R. Mitchell. 1994. "An alternative approach: the unfolding model of voluntary employee turnover." Academy of Management Review 19: 51-89.
March, J.G. and H.A. Simon. 1958. Organizations. New York, NY: Wiley.
Mathieu, J.E. and D.M. Zajac. 1990. "A review and meta-analysis of the antecedents, correlates, and consequences of organizational commitment." Psychological Bulletin 108: 171-194.
Meyer, J.P. and N.J. Allen. 1997. Commitment in the Work Place. Thousand Oaks, CA: Sage.
Mobley, W.H. 1982. Employee Turnover: Causes, Consequences, and Control. Menlo Park, CA: Addison-Wesley.
___ 1977. "The intermediate linkages in the relationship between job satisfaction and employee turnover." Journal of Applied Psychology 62: 237-240.
___ R.W. Griffeth, H.H. Hand, and B.M. Meglino. 1979. "Review and conceptual analysis of the employee turnover process." Psychological Bulletin 86: 493-522.
Morita, J.G., T.W. Lee and R.T. Mowday. 1993. "The regression-analog to survival analysis: a selected application to turnover research." Academy of Management Journal 36: 1430- 1464.
Mowday, R.T., L.W. Porter and R.M. Steers. 1982. Employee-Organization Linkages. New York, NY: Academic Press.
___, R.M. Steers and L.W. Porter. 1979. "The measurement of organizational commitment." Journal of Vocational Behavior 14: 224-247.
Price, J.L. and C.W. Mueller. 1986. Absenteeism and Turnover of Hospital Employees. Greenwich, CT: JAI Press.
Schmitt, F.L. and J.E. Hunter. 1996. "Measurement error in psychological research: lessons from 26 research scenarios." Psychological Methods 1:199-223.
Steel, R.P. 1996. "Labor market dimensions as predictors of the reenlistment decisions of military personnel." Journal of Applied Psychology 81: 421-428.
Steers, R.M. and R.T. Mowday. 1981. "Employee turnover and post-decision accommodation processes." Research in Organizational Behavior 3: 235-281.
Stroh, L.K., J.M. Brett and A.H. Reilly. 1996. "Family structure, glass ceiling, and traditional explanations for the differential rate of turnover of female and male managers." Journal of Vocational Behavior 49: 99-118.
Thornton, A. 1991. "Influence of the marital history of parents on the marital and cohabitational experiences of children." American Journal of Sociology 96: 868-894.
Trevor, C.O., B. Gerhart and J.W. Boudreau. 1997. "Voluntary turnover and job performance: curvilinearity and the moderating influences of salary growth and promotions." Journal of Applied Psychology 82: 44-61.
Wu, L.L. 1996. "Effects of family instability, income, and income instability on the risk of premarital birth." American sociological Review 61: 386-406.
___ and B.C. Martinson. 1993. "Family structure and risk of premarital birth." American Sociological Review 58: 210-232.
Zedeck, S. 1992. Work, Families and Organizations. San Francisco, CA: Jossey Bass.
Means, Standard Deviations and Correlations Surface Warfare Officers Standard Correlations [b] Varlables [a] Mean Deviation n 1 2 1. Turnover .17 .37 2,422 .50 [***] 2. Intention to Leave 2.60 2.05 2,793 - 3. Organizational 5.09 .91 2,833 Commitment 4. Employed Spouse .37 .48 2,859 5. Number of Children 1.21 1.23 2,857 at Home 6. Spouse .79 .41 2,859 7. Race .03 .17 2,853 8. Sex .01 .08 2,859 9. Tenure 9.73 5.82 2,773 10. Perceived Ease 5.29 1.60 2,847 of Movement 11. Career Satisfaction 5.10 1.35 2,831 Varlables [a] 3 4 5 6 7 1. Turnover -.30 [***] .00 -.20 [***] -.14 [***] -.02 2. Intention to Leave -.51 [***] -.02 -.41 [***] -.34 [***] .06 [**] 3. Organizational - -.02 .16 [***] .16 [***] .00 Commitment 4. Employed Spouse -.05 [**] .35 [***] .02 5. Number of Children - .50 [***] -.04 [*] at Home 6. Spouse - -.06 [**] 7. Race - 8. Sex 9. Tenure 10. Perceived Ease of Movement 11. Career Satisfaction Varlables [a] 8 9 10 11 1. Turnover -.01 -.33 [***] .09 [***] -.37 [***] 2. Intention to Leave .05 [*] -.63 [***] .12 [***] -.59 [***] 3. Organizational .00 .20 [***] -.05 [**] .78 [***] Commitment 4. Employed Spouse -.01 .07 [***] .01 .00 5. Number of Children -.07 [***] .51 [***] -.04 [*] .18 [***] at Home 6. Spouse -.10 [***] .41 [***] -.04 [*] .18 [***] 7. Race -.01 -.11 [***] -.01 -.03 8. Sex - -.06 [***] .03 .00 9. Tenure - -.08 [***] .28 [***] 10. Perceived Ease - -.08 [***] of Movement 11. Career Satisfaction -
(*.)p [less than] .05, (**.)p [less than] .01, (***.)p [less than] .001; two-tailed tests
(a.)Dummy codes were as follows: Turnover: 0-stayer, 1-leaver; Employed spouse: 0-no, 1-yes; Spouse: 0-no,1-yes; Race: 0-white, 1-nonwhite; Sex: 0-males. 1-females.
(b.)Product moment correlations are reported. However, chi-squared tests for statistical significance of association are reported when 1 or 2.dummy variables are involved.
Means, Standard Deviations and Correlations Aviation Warefare Officers Standard [Correlations.sup.b] Variables [a] Mean Deviation n 1 2 1. Turnover .13 .34 4,490 - .46 [***] 2. Intention to Leave 2.45 1.75 4,984 - 3. Organizational 5.08 .82 5,029 Commitment 4. Employed Spouse .38 .49 5,051 5. Numberof Children 1.41 1.18 4,365 at Home 6. Spouse .73 .44 5.051 7. Race .01 .11 5,041 8. Sex .00 .06 5,051 9. Tenure .25 5.17 5,550 10. Perceived Ease 5.21 1.57 5.026 of Movement 11. Career Satisfaction 5.39 1.06 5,033 Variables [a] 3 4 5 6 7 1. Turnover -.26 [***] .01 -.16 [***] -.10 [***] -.03 [*] 2. Intention to Leave -.46 [***] -.01 -.35 [***] -.26 [***] .03 [*] 3. Organizational -.03 [*] .09 [***] .08 [***] -.02 [*] Commitment 4. Employed Spouse - -.15 [***] .22 [***] .01 5. Numberof Children - .31 [***] -.04 [*] at Home 6. Spouse - -.05 [***] 7. Race - 8. Sex 9. Tenure 10. Perceived Ease of Movement 11. Career Satisfaction Variables [a] 8 9 10 11 1. Turnover .05 [**] -.26 [***] .08 [***] -.25 [***] 2. Intention to Leave .04 [**] -.60 [***] .09 [***] -.41 [***] 3. Organizational .00 .13 [***] .01 .68 [***] Commitment 4. Employed Spouse .02 .04 [**] .01 -.02 5. Numberof Children -.06 [***] .46 [***] -.02 .04 [**] at Home 6. Spouse -.03 .33 [***] -.03 [*] .05 [***] 7. Race -.01 -.04 [**] .03 [*] -.02 8. Sex - -.07 [***] .01 .00 9. Tenure - -.06 [***] .07 [***] 10. Perceived Ease - .00 of Movement 11. Career Satisfaction -
(*.)p[less than or equal to].05,
(**.)p[less than or equal to].01,
(***.)p[less than or equal to].001; two-tailed tests
(a.)Dummy codes were as follows: Turnover: 0-stayer. 1-leaver; Employed spouse: 0-no, 1-yes; Spouse: 0-no, 1-yes; Race: 0-white, 1-nonwhite; Sex: 0-males, 1-females.
(b.)Product moment correlations are reported. However. chi-squared tests for statistical significance of association are reported when 1 or 2 dummy variables are involved.
Means, Standard Deviations and Correlations General Unrestricted Officers Standard Correlation [b] Variables [a] Mean Deviation n 1 2 1. Turnover .27 .44 1,160 - .39 [***] 2. Intention to Leave 3.35 1.97 1,183 - 3. Organization 5.33 .94 1,195 Commitment 4. Employed Spouse .42 .49 1,201 5. Number of Children .35 .80 950 at Home 6. Spouse .46 .50 1,201 7. Race .07 .26 1,181 8. Sex .85 .35 1,201 9. Tenure 4.16 4.73 1,114 10. Perceived Ease 5.30 1.55 1,193 of Movement 11. Career Satisfaction 5.16 1.21 1,190 Variables [a] 3 4 5 6 7 1. Turnover -.29 [***] .04 .06 .07 [*] .02 2. Intention to Leave -.57 [***] -.01 -.05 -.02 -.03 3. Organization - .00 .06 .02 .01 Commitment 4. Employed Spouse - -.14 [***] .79 [***] -.02 5. Number of Children - .33 [***] -.10 [**] at Home 6. Spouse - .01 7. Race - 8. Sex 9. Tenure 10. Perceived Ease of Movement 11. Career Satisfaction Variables [a] 8 9 10 11 1. Turnover -.32 [***] -.12 [***] .10 [***] .31 [***] 2. Intention to Leave -.06 [*] -.36 [***] .06 [*] -.55 [***] 3. Organization .08 [**] .07 [*] .01 .69 [***] Commitment 4. Employed Spouse .00 .13 [**] .08 [**] .05 5. Number of Children -.31 [***] .14 [**] .08 [**] .00 at Home 6. Spouse -.16 [***] .18 [***] .10 [***] .05 7. Race -.05 -.06 [*] .00 -.04 8. Sex - -.08 [**] -.10 [***] .16 [***] 9. Tenure - .06 [*] .21 [***] 10. Perceived Ease - .02 of Movement 11. Career Satisfaction -
(*.)p[less than or equal to].05.
(**.)p[less than or equal to].01.
(***.)p[less than or equal to].001; two-tailed tests
(a.)Dummy codes were as follows: Turnover: 0-stayer, 1-leaver; Employed spouse; 0-no, 1-yes; Spouse: 0-no 1-yes; Race: 0-white 1-nonwhite; Sex: 0-males, 1-females.
(b.)Product moment correlations are reported. However, chi-squared tests for statistical significance of association are reported when 1 or 2 dummy variables are involved.
Survival Analyses with Turnover as the Dependent Variable Across the SWO, AWO and GUO Samples [a] Samples [b] Predictors SWO AWO GUO 1. Race .52 .14 [**] 1.63 2. Sex .70 3.28 [**] .34 [***] 3. Intention to Leave 1.63 [***] 1.75 [***] 1.68 [***] 4. Organizational Commitment .89 .88 .96 5. Employed Spouse .72 .64 .47 6. Children at Home .36 .34 [*] .90 7. Spouse .26 .43 4.64 8. (3) x (5) .96 1.06 1.07 [***] 9. (3) x (6) 1.11 [*] 1.08 [*] 1.40 10. (3) x (7) 1.26 [*] 1.10 .67 [*] 11. (4) x (5) 1.11 1.03 .97 12. (4) x (6) 1.09 1.11 .98 13. (4) x (7) 1.05 1.09 .94 Global [X.sup.2] 805.51 [***] 1135.92 [***] 187.23 [**] Sample size 2,571 4,223 808
(a.)SWO: Surface Warfare Officers, AWO: Aviation Warfare Officers, GUO: General Unrestricted Officers.
(b.)Exponentiated regression coefficients are reported. Values above I indicate a positive effect, values below 1 indicate a negative effect, and a value of 1 indicates no effect.
(***.)p[less than].001; one-tailed tests.
OLS Regressions with Intention to Leave as the Dependent Variable Across SWO, AWO and GUO Samples [a] Samples [b] Predictors SWO AWO GUO 1. Tenure -.16 [***] -.17 [***] -.15 [***] 2. Race .05 .06 -.51 [**] 3. Sex .11 .18 -.19 ([delta][R.sup.2] .39 [***] .35 [***] .16 [***]) 4. Organizational Commitment -1.32 [***] -1.12 [***] -1.17 [***] 5. Employed Spouse .56 -.19 -.47 6. Children at Home -1.36 [***] -1.27 [***] .26 7. Spouse -1.74 [***] -.44 .34 ([delta][R.sup.2] .16 [***] .14 [***] .29 [***]) 8. (4) x (5) -.09 .04 .10 9. (4) x (6) .24 [***] .22 [***] -.05 10. (4) x (7) .30 [***] .06 -.05 ([delta][R.sup.2] .03 [***] .02 [***] .00) Intercept 11.10 [***] 9.88 [***] 10.46 [***] F 366.70 [***] 446.84 [***] 70.25 [***] Adjusted [R.sup.2] .58 .51 .45 Sample size 2,691 4,291 855
(a.)SWO: Surface Warfare Officers, AWO: Aviation Warfare Officers, GUO: General Unrestricted Officers.
(b.)Unstandardized regression coefficients are reported.
(*.)p[less than or equal to].05,
(**.)p[less than or equal to].01,
(***.)p[less than or equal to].001; one-tailed tests.
|Printer friendly Cite/link Email Feedback|
|Author:||Lee, Thomas W.; Maurer, Steven D.|
|Publication:||Journal of Managerial Issues|
|Article Type:||Statistical Data Included|
|Date:||Dec 22, 1999|
|Previous Article:||The Influence Of Corporate Governance On Investor Reactions To Layoff Announcements.|
|Next Article:||Journal of Managerial Issues.|