nonwork influences on withdrawal cognitions: an empirical examination of an overlooked issue
TRANSCRIPT
Nonwork Influences on Withdrawal Cogn itions:
An Empirical Examination of an Overlooked
Issue
Aaron Cohen1,2
The aim of this research is to examine several dimensions of nonwork domains
such as nonwork-to-work spillover, work/nonwork conflict, coping strategies, and
organizational support for nonwork, and the ir re lationship to withdrawal
cognitions. Questionnaires were mailed to all employees of a school district in
western Canada. A total of 300 usable questionnaires were returned: a response
rate of 42% . The findings show that nonwork domain variables are significantly
related to withdrawal cognitions. This was demonstrated mainly in two strong
interaction effects between the importance of nonwork and personal coping
strategie s on the one hand, and between the importance of nonwork and
perceived organizational support for nonwork on the other. The two interactions
remain significant when two work variables, i.e., organizational commitme nt and
job satisfaction, were entered into the regression as control variables. The paper
concludes with proposed directions for future research based on the findings of
this present work.
KEY WORDS: turnover; nonwork; spillover; work/nonwork conflict.
INTRODUCTION
Increasing change s in the workforce call for more atte ntion to the
effect of nonwork domains on attitude s and behaviors at work. Over half
of all married women are employe d, many in careers. Fathe rs are partici-
pating more in childrearing and child support activitie s. Time spent with
familie s in recreational activitie s has increased. Higher divorce rates have
made many workers single pare nts. These and a host of othe r social trends
have increased the pote ntial for competition between the organization and
the family. This compe tition has increased family resistance to moving,
Hum an Relations, Vol. 50, No. 12, 1997
1511
0018-7267/97/1200-1511 $12.50/1 Ó 1997 The Tavistock Institute
1Department of Political Science, University of Haifa, Mount Carmel, Haifa 31905, Israel.2Requests for reprints should be addressed to Aaron Cohen, Department of Political Science,
Unive rsity of Haifa, Mount Carme l, Haifa 31905, Israel.
and has transfe rred family stresses to the workplace (Orthne r & Pittman,
1986) . Yet, Wile y (1987) argued that studie s measure d work/nonwork role
conflict only in terms of work interference in the family role , neglecting
the individual ’s perceptions of the inte rference of family role demands in
work role obligations. Ste ffy and Jones (1988) justified the need for evalu-
ating the effects of family and house hold variable s on various commitment
domains base d upon the premise that just as workplace factors affect ex-
tra-work life , so too do family and house hold factors influe nce organiza-
tional and career attitude s and behaviors. Lambert (1990) argue d that to
docume nt the costs and benefits of maintaining a particular balance be-
tween work and home, research must specify the relationship between dif-
ferent patte rns of work/family intersection and measure s of organizationa l
well-be ing. Lambert recommende d the use of diverse outcome measure s
to capture the full range of effects of diffe rent patte rns of work/home
inte rsection on employing organizatio ns. Turnove r, which is considered
one of the most important indicators of organizational we ll-be ing
(Mowday, Porter, & Steers, 1982) , is the outcome measure examined in
this study in its relationship to nonwork domains.
Employe e turnove r is one of the aspects most studied in organiza-
tional research (Cotton & Tuttle , 1986; Mitra, Jenkins, Douglas, & Gupta,
1992) . Yet there are no firm conclusions as to the turnove r process (Cotton
& Tuttle , 1986) , and there is some disappointm ent in the results dealing
with the relationship between variable s representing individualistic theo-
ries (e .g., age , sex, and job satisfaction) and turnove r (Dalton & Todor,
1993) . One factor that has received little attention in turnove r research
is nonwork influe nces on staying or leaving ( Mowday et al. 1982) . The
relationships between individual nonwork relate d variable s and turnove r
are often neglected (Moble y, 1982) . Moreove r, studie s have found that
aspects of work/nonwork inte rface are relate d to work attitude s such as
job satisfaction and organizational commitment (Kirchmeyer, 1992; Wile y,
1987; Rice , Frone , & McFarlin, 1992; Rudd & McKenry, 1986; Ste ffy &
Jones, 1988), all conside red important ante cedents of turnove r (Tett &
Meyer, 1993; Cohe n, 1993) . This indicate s that nonwork aspe cts might be
related to turnove r. In light of the above , this research examine s several
dimensions of nonwork domains such as nonwork-to-work spillove r,
work/nonwork conflict, coping strategies and organizational response to
nonwork in the ir relationship to turnove r. It will also examine the effect
of nonwork domains relative to the effect of work domain variable s. The
possibility of a more comple x effect of nonwork domains on turnove r will
be examine d too by testing inte ractions among nonwork domain variable s
in the ir relationship to turnove r.
1512 Cohen
CONCEPTUAL FRAMEWORK AND HYPOTHESES
Nonwork Domain s and Turnover: Early Conceptualization s
Conceptually, very little work has been done on the effect of nonwork
domains on turnove r. An early and general perspective on this relationship
was offered by Sussman and Cogswe ll (1971) . Their main argument was
that there is a direct relationship between supply and demand of workers
in any occupational system and the conside rations of non-economic factors
in job movements. The greater the demand for workers in any occupational
system the greater the consideration give n to familial concerns such as work
aspirations of spouses, special needs of children, community activitie s, links
with kin, friends, and voluntary associations, physical and social environ-
ment, and conditions in the work situation. That is, in a market of few
options or of practically no jobs, and where survival is paramount, most
individuals will go where the work is for the available pay. On the other
hand, a worker enjoying great demand for her/his services, one in which
s/he has many job options, will input into a decision those non-mone tary
conside rations relevant to her/his social situation and personality. Another
way to conceptualize the effect of nonwork domains on turnove r is by the
side-bet theory (Becker, 1960) . Becker argues that over time, certain ac-
cruing costs make it more difficult to disengage from a consistent line of
activity, namely, maintaining membership in the organization. The threat
of losing these inve stments, along with a perceived lack of compensating
alte rnative s, commits the person to the organization. Becker phrase d his
argume nt as follows “...The man who hesitates to take a new job may be
deterred by a comple x of side-bets: the financial costs connected with a
pension fund he would lose if he moved; ...the loss of ease in domestic
living conseque nt on having to move his household, and so on ...” (Becker,
1960, pp. 38-39) . Nonwork domains, then, constitute Becker’s examples of
side-bet that might affect turnove r decision.
The Nature of the Relation sh ip Between Nonwork Domains and
Turnover
There is a consensus in the lite rature regarding the way nonwork con-
siderations affect turnove r. Steers and Mowday (1981) , Price and Mueller
(1981) , and Mobley (1982) argue d that nonwork considerations do not af-
fect turnove r directly. They affect the intention to leave or to stay in the
organization, and this intention is the variable that has a direct effect on
turnove r (Steel & Ovalle , 1984). There is, however, a major diffe rence be-
tween Price and Mueller (1981) on the one hand and Moble y (1982) and
Steers and Mowday (1981) on the other. Price and Mueller expected and
Influences on Withdrawal Cogn itions 1513
found a direct relationship of kinship responsibilitie s, which was the ir in-
dicator of nonwork domains, on the intention to stay in the organization.
Mobley and Steers and Mowday, however, hypothe sized that nonwork con-
siderations moderate the relationship between affective responses to the
job and inte ntion to stay or leave the organization. However, Lee and
Mowday (1987) , who examine d Steers and Mowday’s (1981) model, found
no support for an interaction between affective response s and nonwork in-
fluences. Their results suggest a direct relationship between affective re-
sponses and intention to stay or leave with no effect of nonwork influences.
Nevertheless, they argued that it would be inappropriate to recommend
deleting nonwork influences from the model on the basis of a single study.
A relate d issue regarding the relationship between nonwork domains
and turnove r is the nature and the scope of variable s defined as nonwork
domains. Most work has focused on what is generally labe led “family re-
sponsibility ” (Blegen, Mueller, & Price, 1988; Mobley, 1982; Sussman &
Cogswell, 1971; Muchinsky & Tuttle , 1979; Porter & Steers, 1973; Good,
Sisle r, & Gentry, 1988; Price & Mue lle r, 1981) . What is missing in the
lite rature is an empirical examination of othe r aspects of nonwork domains
regarding the ir relationship to turnove r. Mowday, Porte r, and Steers (1982)
argue d that the study of turnove r was just beginning to recognize a number
of nonwork factors that influe nce turnove r decisions, and that a constella-
tion of nonwork influe nces is often ignored in determining desire or intent
to leave . Many circumstances can be identified in which one may not like
a particular job but still does not seek termination. These include situations
where a spouse is limited geographically to a certain region and alte rnative
employment is scarce, or situations where the employe e’s central life in-
terest lie s outside of work. Mobley (1982) suggested that nonwork value s,
such as le isure or location preferences, can contribute to turnove r. Family
responsibilitie s, dual career familie s, and conflict between work and non-
work role s can also have an impact. The organization ’s analysis of turnove r
should include a diagnosis of nonwork value s and roles and the ir relation
to job behavior.
Recent research on nonwork domains has emphasized the perceptions
and reactions of individuals to work/nonwork interface and the ir effects on
the quality of life of the se individuals. Important conce pts such as
work/nonwork conflict (Frone, Russe ll, & Coope r, 1992; Williams & Alliger,
1994) , positive and negative nonwork to work spillove r (Croute r, 1984;
Kirchmeyer, 1992) , coping strategie s (Beute ll & Greenhaus, 1983; Paras-
uraman & Hansen, 1987) , and organizational support to employe es’ non-
work needs (Orthne r & Pittman, 1986) were not examined as possible
determinants of turnove r although they were found to be related to atti-
1514 Cohen
tudes which are turnove r antecedents, such as commitment and satisfaction
(Kirchmeyer, 1992; Steffy & Jones, 1988) .
In short, very little research has been performed on the relationship be-
tween nonwork domains and turnover. While the influence of nonwork fac-
tors on employe e turnove r remains perhaps one of the richest areas for
future work, fe w studie s have syste matically examined this relationship
(Mowday et al., 1982; Porte r & Steers, 1973) . As a result, many aspects of
this relationship have not been explored. For example , most of the above
research applie d a limited definition of nonwork, mainly its family responsi-
bility aspects. Few studie s have developed conceptual arguments regarding
the process by which different aspects of nonwork domains might be related
to turnove r. Also, little research has compared the effect of nonwork domains
with work variable s in their relationship to turnove r. This research attempts
to explore some of these issues. The following section suggests several hy-
potheses regarding the possible effect of nonwork domains on turnove r and
the process whereby they are expected to be relate d to turnover.
Research Hypotheses
Work/Nonwork Spillover
Croute r (1984) defined psychologica l spillove r as a transitory phe-
nomenon which include s the ways in which family life affects an individual ’senergy leve l, attention span, and mood, which in turn are brought into the
work setting by the worker. Kirchmeyer (1992) described how family and
other nonwork domains can affect attitude s and behaviors at work. She
argue d that by active participation in nonwork domains, such as family (e.g.,
pare nting), community (e .g., political partie s, charitie s) , and recreation
(e.g., social clubs, hobby associations) , the employe d individual can increase
the numbe r of privile ges which he or she can enjoy beyond work-re lated
one s, buffe r the failure s and strains of work, gain contacts and information
valuable for work, and develop skills and perspectives useful there. Such
resource enrichme nts involve not only the individual ’s capacity to meet
work demands and his or her value to the employing organization, but also
the sense of personal competence . Through such enhance ments, nonwork
participation could favorably influence attitude s and behaviors toward the
organization and the job.
This research anticipate s that a positive nonwork-to-work spillove r will
be related to withdrawal cognitions. Of all the propose d ante cedents of turn-
over, those concerning the work experience itself have demonstrated the
strongest relationships with turnover (Mowday et al., 1982; Lee & Mowday,
1987). The more positive the expe rience at work, the stronge r the intention
Influences on Withdrawal Cogn itions 1515
to remain in the organization tends to be. Similarly, the more nonwork par-
ticipation enhance s the work e xperience , the stronger the intention to remain
in the organization should be . Through its effect on the work expe rience,
nonwork participation may be related to withdrawal cognitions.
Hypothesis 1a. Low withdrawal cognitions will be associate d with high
positive nonwork to work spillove r.
Work/nonwork conflict reflects the goodne ss of fit between work life
and nonwork life (Rice et al., 1992) . The demands associated with one role
constrains the time and psychological resources that individuals can devote
to the other. Individuals experiencing high job demands may have limited
time and energy for family tasks. As a result, perceptions that work inter-
fe res with family would be high (Williams & Allige r, 1994) . Increased
work/nonwork conflict might lead employe es to look for an alte rnative work
setting which will offer them a better work schedule (e.g., no shiftwork, a
shorte r workday, no weekends and holiday work), or day-care arrange ments
that would reduce the nonwork demands and thereby decrease the pote n-
tial for work/nonwork conflict. Moreover, Frone et al. (1992) argue d that
individuals are less like ly to accept direct responsibility for managing the ir
work roles in a way that does not inte rfere with the ir family life . Rathe r,
individual s are more like ly to hold the ir organizat ions responsible for
work/nonwork conflict. This probably causes negative attitude s toward the
organization, increasing withdrawal cognitions.
Research has supporte d the notion that work¯family conflict and pres-
sures can cause employe es to quit their jobs. Sussman and Cogswe ll (1971)
offered specific arguments indicating how a certain type of family and life
cycle stage can cause differential propensities to move. The relationship
between family responsibility and turnove r intentions was found to be gen-
erally positive (Muchinsky & Tuttle , 1979) , indicating that work¯family con-
flict and pressures can cause employe es to quit the ir jobs. Porter and Steers
(1973) stated that increased family responsibility produces more turnove r
for women whereas the results for men are mixed. Good, Sisle r, and Gentry
(1988) found that retail managers with higher leve ls of work¯family conflict
were more like ly to leave the company.
Hypothesis 1b. High withdrawal cognitions will be associated with high
work/nonwork conflict.
Response to Nonwork
Importan ce of Nonwork. The two explanations offered for the relation-
ship between importance of nonwork and turnove r suggest that individuals
who assign high importance to nonwork domains will have weaker with-
drawal cognition s. The first one is base d on the expansion mode l of
1516 Cohen
work¯nonwork ties. Accordingly, individuals who value the ir nonwork do-
mains are more willing to cope with increasing extra-organizational role
demands by responding to them positive ly (Kabanoff, 1980; Marks, 1977) .
Adding new roles may liberate source s of energy for the individual and,
rathe r than having to pay for extensive social involve ment, individuals may
come away from new social involve ments more enriche d and vitalize d.
Many ties may be supportive of the individual and create energy for use
in other role performance (Randall, 1988) . Such individuals will not per-
ceive the organization as inte rfering in their extra-organizational role de-
mands and will not deve lop negative attitude s that might cause them to
leave the ir work setting.
Another explanation is based on the side-bet theory mentioned earlier.
It states that nonwork domains can be perceived as a side-bet, following
Becker’s (1960) theory. That is, one might hesitate to leave the workplace so
as not to lose any of the quality of one’s nonwork life in one’s current com-
munity. In many cases, leaving a workplace means having to relocate . Indi-
viduals who have deve loped strong ties to their community, friends, family,
social clubs will be less incline d to leave their organization if it means leaving
their location and their community. Thus, involve ment in nonwork domains
can be a side -bet that will prevent individuals from leaving in order not to
lose or reduce the quality of their extra-organizational life .
Hypothesis 2. Low withdrawal cognitions will be associated with high
importance of nonwork domains.
Personal Coping Strategies. Williams and Alliger (1994) argued that the
negative effects of role juggling on mood and work/family conflict sugge st
that individuals should structure their work and family roles to reduce the
potentials for role intrusions. The type of personal strategy used to cope
with the demands and responsibilitie s of multiple domains appears to affect
the expe rience of interdomain conflict (Hall, 1972; Kirchmeyer, 1993) . Gil-
bert and Holahan (1982) found that subje cts rating their characteristic ways
of coping as highly effective reporte d less conflict than subjects rating the ir
characte ristic strategies as less effective. Hall’s (1972) work on the strate-
gies of college-educate d women in coping with role conflict provide s the
framework for the concept of coping strategie s in this research. The women
in the sample were active in multiple domains and reported using 16 coping
strategie s classified into three types: structural role redefinition (e.g., a
proactive attempt to deal with the objective reality of one’s role s by reduc-
ing the role demands and changing othe r’s expectations) , personal role re-
definition (e.g., a more defensive approach where conflicts are reduced
through changing one’s personal attitude s and behaviors as oppose d to al-
tering role demands) , and reactive role behavior (e.g., no attempt to ad-
dress conflict, but rathe r the individual strives to improve his or her ability
Influences on Withdrawal Cogn itions 1517
to satisfy all demands) . Hall’s findings showed the reactive type to be the
least effective . Kirchmeyer (1993) replicated Hall’s research and found that
successfully coping with multiple domains involve s applying good personal
organization and deve loping an appropriate attitude . Strategies which aim
at altering one ’s own attitude s as oppose d to altering those of others, and
increasing one’s personal efficiency as oppose d to decreasing one’s activity
leve l or relying on others, appe ared to be most effective in he lping man-
agers cope with a multitude of life domains. These findings are in accord-
ance with those of Beutell and Greenhaus (1983) who found in a sample
of women that active coping strategies were perceived to be successful in
dealing with home and nonhome role conflicts.
Effectiveness in dealing with multiple domains should result in fewer
work¯nonwork conflicts and avoid the resulting negative attitude s toward
the work setting. For example , effective coping strategies were found to
affect job satisfaction and organizational commitment positive ly (Paras-
uraman & Hansen, 1987) . Employees who deal more effectively with mul-
tiple domains will experience less interdomain conflicts that might result
from ineffective coping. They will feel more comfortable with the ir current
setting and will be less incline d to view their work setting as the reason
for the ir conflicts and pressures. These employe es are expe cted to have
stronger intentions to remain in the organization than individuals who do
not cope effectively with multiple domains.
Hypothesis 3. Lower withdrawal cognitions will be associated with
greater personal coping strategies.
Organization al Support for Nonwork
Moble y (1982) argued that the understanding, prediction, and man-
agement of turnove r require the assessment of the extent to which policie s,
practice s, and conditions are perceived by employees to facilitate or inter-
fere with the attainme nt of nonwork value s. The logic underlying many
corporations’ decision to offer employe r-base d family supports such as
childcare and flexible work schedule s may be that such benefits will facili-
tate employe es’ abilitie s to handle family matters (Croute r, 1984) . This will
reduce the potential for work¯nonwork conflict for many employees and
will enhance the ir work performance. Williams and Allige r (1994) similarly
argue d that organizational inte rventions may reduce work¯family juggling
and conflict. Flexible work schedules and on-site day care , for example ,
provide workers with greater control over the ir work and nonwork sched-
ules and may make it easier for them to manage the daily demands of
work and family roles. This process will generate positive attitude s toward
the organization and will increase the intention to remain in it. Another
1518 Cohen
reason to expect a relationship between nonwork domains and withdrawal
cognitions is base d on the side -bet theory approach (Becker, 1960) . Here,
organizational support can be perceived as a type of benefit difficult to
obtain elsewhere.
Rudd and McKenry (1986) , for example , found that satisfactory child-
care incre ases job satisfaction of employe d mothers, and O rthne r and
Pittman (1986) found in a sample of military personnel that perceived or-
ganizational support led to stronger job morale , a stronger intent to pursue
an air force career and a stronger perceived quality of job performance.
Orthne r and Pittman (1986) conclude d that improve d family support poli-
cies are like ly to yield benefits for organizations in such areas as employe e
performance, morale , and retention. In light of the above it is anticipate d
that the greater the support provide d by the organization with regard to
its employe es’ nonwork domains, the lower will be the withdrawal cogni-
tions of these employees.
Hypothesis 4. Supportive organizational responses to nonwork domains
of employe es will be relate d negative ly to withdrawal cognitions.
Interaction Effects
This pape r argue s that the effect of the variable s “personal coping
strategie s” and “organizational support” on withdrawal cognitions might
not be direct, but a more complex inte raction effect. Moble y (1982) argued
that whether or not individuals translate the ir present and future evaluation
of the job into turnove r intentions may be relate d to the degree to which
the job is perceived to facilitate or inte rfere with important nonwork values
or with nonwork roles. As with work values, individual differences must be
recognize d. For individuals whose central life values are nonwork-re lated
the ir job choice and turnove r decision would be expected to be strongly
relate d to nonwork domains. Conversely, for individuals whose central life
value s are predominantly work-re lated, the relationships of nonwork do-
mains with turnove r intentions should be less strong. This difference is an-
ticipated to have an impact on the way individual s and organizat ions
respond to work/nonwork interface . Individuals are expose d to the effect
of nonwork domains regardle ss of their importance to them. But the im-
portance they assign to nonwork determines the way they respond to the
pressures from multiple domains.
For employees who value the ir nonwork domains, coping effectively
with the pressures from multiple domains will cause them to adjust better
to their work setting and, as a result, they will be more capable of fulfilling
the ir nonwork plans. The work setting will not be perceived as something
that inte rferes with fulfilling their nonwork plans, which they consider an
Influences on Withdrawal Cogn itions 1519
important aspe ct of the ir lives, and therefore they will be less incline d to
leave the organization. On the othe r hand, employe es who assign low im-
portance to their nonwork domains are less exposed to pressures from mul-
tiple domains. Whate ver happe ns in their nonwork domains is not expe cted
to be significantly relate d to their behavior and attitude s at work because
nonwork domains are not an important aspect of their live s. Therefore,
the way they cope with pressures from the ir work and multiple domains
will not be relate d to their attitude s towards the organization, or their with-
drawal cognitions.
Another inte raction expected is between the variable s “importance of
nonwork domains” and “organizational support”. For employe es who value
their nonwork domains, more organizational support will increase the ir
ability to cope with demands from multiple domains. That will prevent
negative attitude s toward the organization and will result in higher levels
of intention to remain in the organization. The opposite is expected when
the organization is not supportive of its employe es. In such cases, employ-
ees who value their nonwork domain might feel frustrated by the ir inability
to fulfil the ir nonwork responsibilitie s and needs. Some of this frustration
will be attribute d to the work setting, the organization in particular, and
will result in stronge r intentions to leave.
Finally, the use of three moderators in the relationship between non-
work domains and withdrawal cognitions raises the possibility of a three-
way inte raction effe ct. It is re asonable to e xpe ct that the inte raction
between the variable s “importance of nonwork domains” and “personal
coping strategies” will diffe r in a way depending on the support of the
organization for nonwork. That is, the interaction of importance of non-
work domains and personal coping strategie s will ope rate differently when
the organization supports the nonwork domains of its employees from when
it does not. When the organization supports the nonwork domains, the ef-
fect of the variable importance of nonwork domains on withdrawal cogni-
tions will be positive for employees who cope effectively with their nonwork
domains. Those who value their nonwork domains and use effective coping
strategie s to prevent work¯nonwork conflicts will respond with lower levels
of withdrawal cognitions when the organization supports the ir nonwork do-
mains. On the othe r hand, those who value the ir nonwork domains and
use effective coping strategie s to prevent work¯nonwork conflicts, but re-
alize that the organization doe s not support their efforts to cope, will react
negative ly and thus show higher levels of withdrawal cognitions.
Hypothesis 5(a). The effect of personal coping on withdrawal cognitions
will be stronger for employe es who assign high importance to nonwork do-
mains than for employees who assign them lower importance .
1520 Cohen
Hypothesis 5(b). The effect of organization support for nonwork will
be stronger for employe es who assign high importance to nonwork domains
than for employees who assign them lower importance .
Hypothesis 5(c). The effect of nonwork domain on withdrawal cogni-
tions will vary depending on personal coping strategies as well as organ-
izational support for nonwork domains.
Nonwork Domains and Work Attitudes. The final hypothe sis tests the
incremental effect of nonwork domains on withdrawal cognitions beyond
the effect of work-related variable s. Near, Rice , and Hunt (1980) argued
that a research que stion which emerged as worthy of immediate study was
the effect of nonwork aspe cts on work behaviors and attitude s relative to
work setting variable s. Such research has important implications for policy
issues because it may be necessary to conside r change s in off-the-job con-
ditions to improve the experienced quality of working life . For example , a
strong effect of the nonwork domain on turnove r would mean greater em-
phasis on improving the quality of life outside of work. It is thus important
to examine the effect of nonwork domains on turnove r in relationship to
the effect of affective responses to job. Job satisfaction and organizational
commitment have been extensive ly researched in their relationship to turn-
ove r (Tett & Meyer, 1993) , and are examined here. The two attitude s will
be examined according to Tett and Meyer’s (1993) meta-analysis findings
that job satisfaction and organizational commitment are distinguishable
constructs, each contributing unique ly to withdrawal cognitions. If nonwork
domain conside rations are important determinants of turnove r intentions,
as argue d in the above -mentioned literature , then they should have an in-
cremental effect beyond the effect of commitment and satisfaction. Other-
wise , it can be conclude d that nonwork domains are related to withdrawal
cognitions only through their effect on commitment and satisfaction.
Hypothesis 6. Nonwork domain variable s will have a significant effect
on withdrawal cognition beyond the effect of organizational commitment
and job satisfaction.
METHOD
Participan ts
All 720 employees of a school district in western Canada were sur-
veyed. Questionnaire s were mailed and a total of 300 usable que stionnaire s
were returne d; response rate was 42% . Thirty-thre e percent of the sample
were male s. The mean age of the respondents was 42.1 years and the mean
tenure in the local school, school district and occupation was 7.6, 11.2 and
14.2 years, respective ly. Sixty-se ven percent of the sample had completed
Influences on Withdrawal Cogn itions 1521
the ir university education. Of the university graduate s, 16.4% possessed an
MA or PhD degree. Seventy percent of the sample were teachers, 10%
were in administration, and 20% were in clerical, maintenance , or caretak-
ing positions.
Measures
Control Variables
The following control variable s were used: gender (0 = female ; 1 =
male), education (a scale of 1-elementary school, to 7-highe r university de-
gree), years in occupation, and annual income; these were measured as
interval variable s.
Nonwork Domain Variables
Positive Non work-to-Work Spillover. E ight ite ms from Kirchme yer’s(1992) measure were employed here. Each item began with the phrase “Be-
ing involve d in nonwork activitie s,” and example s of the continuation in-
cluded “earns me certain rights and privile ges that otherwise I could not
enjoy,” “give s me support so I can face the difficultie s of work,” and “im-
proves my image at work.” The reliability of this scale was .74, similar to
that (alpha = .76) reported by Cohen and Kirchmeyer (1995).
Work-Nonwork Conflict. Shamir’s (1983) six-ite m scale was employed.
Items referred to the disruption of nonwork by work and include “One of
the difficultie s of my job is that I am not at home enough,” and “Work
often prevents me from participating in leisure activitie s that take place at
the same time.” A 7-point scale of agre ement was used for responding.
The reliability of the instrument here (alpha = .80) is slightly higher than
that reported by Shamir (alpha = .78) .
Importan ce of Nonwork Domains. This variable was measure d based
on the basis of Randall’s measure (1988) . Responde nts were aske d: “How
important is this activity to you? ” Seven potential outside claims were ex-
plore d: jobs outside the work organization, hobbie s and recreational activi-
ties, religious organizations, political partie s, family and relative s, friends,
and a miscellaneous category of other organizations. A 5-point scale format
was used for all response categorie s, where 1 indicate d “not at all impor-
tant” and 5 indicate d “very important.” The total importance of external
claims was calculate d by summing the perceived importance of each claim.
This procedure was applie d by Randall (1988) .
Personal Coping with Multiple Domains. This variable was measured us-
ing eight items from a scale develope d by Kirchmeyer (1993) base d on
1522 Cohen
Hall (1972). These items represent effective strategie s identified in he r
study and suggest a coping theme of good personal organization and an
appropriate attitude . Examples include “Increase my efficiency by sched-
uling and organizing role activitie s carefully,” and “Develop attitude s which
put role demands in a positive light.” A 5-point scale ranging from “not
typical of me” to “very typical” was applie d. The reliability here (alpha =
.72) was slightly lower than the .76 reported by Kirchmeyer (1993) .
Organizational Support. Five items from Kirchmeyer’s (1995) measure
of organizational support for nonwork were selected here. These particular
items dealt with individual perceptions of the respect that organizations
show for nonwork, and include “Considers employe es’ personal live s when
making important decisions about careers,” “Accommodate s employe es’special nonwork needs,” and “Is flexible about employees’ work schedules.”A 5-point scale ranging from “not typical of my organization ” to “very typi-
cal” was applie d. The reliablity found here (alpha = .74) is similar to that
(alpha = .75) reporte d by Kirchmeyer (1995) .
Withdrawal Cognitions . This variable was measure d by three items
base d on Mobley, Griffeth, Hand, and Meglino’s (1979) definition and fol-
lowing similar measure s applie d in the lite rature (Mille r, Katerberg, &
Hulin, 1979; Michae ls & Spector, 1982; Mowday, Koberg, & McArthur,
1984) . Accordingly, responde nts were aske d to indicate the ir agre ement
with the following three items on a 5-point scale: (1) “I think a lot about
leaving the organization ”; (2) “I am active ly searching for an alte rnative
to the organization ”; (3) “As soon as it is possible , I will leave the organi-
zation.” There is a tende ncy in recent literature to view turnove r as a mul-
tidimensional construct (Cohen, 1993) . Therefore data on four dimensions
of withdrawal cognitions were colle cted in this study: from the organization,
the head office, the job, and the occupation. This was done by using the
same items, but replacing the term “organization ” with “job” or “occupa-
tion,” or “school district.” The scale range d from 1 (strongly agree) to 5
(strongly disagre e), which indicate s that a higher score means weaker with-
drawal cognition s.
Organizational Commitment. This variable was measure d by a nine-item
scale developed by Cohe n (1993) , base d on the approach sugge sted by
O’Reilly and Chatman (1986) in which organizational commitment is de-
fined as a psychological attachment to the organization. This affective at-
tachment can take one or more of the following three dimensions: (a)
Identification —adoption of the goals and value s of the organization as one’sown; (b) Affiliation—feelings of be longing to the organization, of be ing
“part of it”; and (c) Moral involvement—internalizing the roles of the com-
mitment obje cts demonstrated in feelings of care and concern for the or-
ganization. The focus of organizational commitment in this research was
Influences on Withdrawal Cogn itions 1523
the local unit, the school in most cases. The reliability here (.76) was lower
than that reported by Cohen (.86) .
Job Satisfaction . The variable was measure d by a 5-ite m scale based
on that developed by Schrie she im and Tsui (1980) . Responde nts were asked
to indicate on a 7-point scale how satisfie d they were with their work: pay,
promotion, supervision, co-workers, job. This measure was deve lope d by
Schriesheim and Tsui (1980) . Note that the reliability in this sample (.78)
is higher than that reported by Tsui, Egan, and O’Reilly (1992) in an Ameri-
can sample (.73) .
Data Analysis
Research hypothe ses were tested by hierarchical regression analysis,
conducte d in the following steps: first the control variable s (gende r, edu-
cation, tenure in occupation, and income) were entered into the equation.
Second, the two spillove r variable s (positive nonwork to work spillove r, and
work/nonwork conflict) were entered. Third, the three response variable s
(importance of nonwork, coping strategie s, and organization support) were
entered. The reason for this orde r is that the nonwork aspects should be
relate d to withdrawal cognitions beyond the effect of the control variable s.
Moreove r, of the two groups of nonwork variable s, the response variable s
were entered after the spillove r variable s because it was assumed that one
first expe riences work/nonwork interface as demonstrated by positive spil-
lover and work/nonwork conflict, and this experience might bear on the
way one reacts to this interface: coping strategie s and perceptions of the
support provide d by the organization. In the final step, the four inte ractions
were entered into the equation, as were three two-way interaction effects
(importance of nonwork ´ coping strategies; importance of nonwork ´ or-
ganization response ; coping strategie s ´ organization response ) and the
three-way inte raction effect (importance of nonwork ´ coping strategie s ´organization response ). Each interaction is presented to allow an inte rpre-
tation of the two-way inte ractions if the three-way interaction is not sig-
nificant, and an interpretation of the main effects when both the two-way
and the three-way inte ractions are not significant. Following Aiken and
West (1991) , the three variable s include d in the inte raction (importance of
nonwork, coping strategies, and organization response) were centered to
minimize multicolline arity. To test Hypothesis 6, regarding the incremental
effect of nonwork domains an additional analysis was performed. Organ-
izational commitment and job satisfaction were entered into the equation
after the demographic control variable s. Only then were the nonwork do-
main variable s and the interactions entered in identical orde r to the one
described above .
1524 Cohen
FINDINGS
Table I shows descriptive statistics, reliabilitie s, and the intercorrela-
tions among the research variable s, demonstrating acceptable psychometric
properties of the variable s. The correlations among the four depende nt
variable s were very high and range d from .73 to .86. This finding demon-
strates the difficultie s of the respondents in differentiating among the with-
drawal cognition forms. Therefore it was decided to combine the 12 items
of the four variable s to form a single measure of withdrawal cognitions.
The corre lations among the inde pendent variable s do not indicate any
proble m of multicolline arity.
Tables II and III present the results of the regression analyse s. Table
II presents the analysis for withdrawal cognitions as a dependent variable ,
and nonwork domains as inde pendent variable s. Table III shows the same
analysis but with organizational commitment and job satisfaction entered
in the second step to test Hypothe sis 6. One should note the consiste nt
effect of the control variable s. The variable s of gender, education, years in
occupation, and income were all significantly relate d to withdrawal cogni-
tions (see Table II). But the effect of the control variable s became much
weake r when organizational commitment and job satisfaction were entered
in the equation (see Table III).
Spillover Variables. Table II shows no relationship between positive non-
work-to-work spillove r and withdrawal cognitions. Table III, which include s
commitment and satisfaction, shows a consistently negative effect of posi-
tive spillove r on withdrawal cognitions. This relationship is contrary to that
expected in Hypothesis 1(a), which anticipate d a positive relationship. Thus
no support for Hypothesis 1(a) is provide d by the data. Hypothe sis 1(b),
which expected a relationship of withdrawal cognitions to work/nonwork
conflict, received some support from the data. High work/nonwork conflict
was associate d with high withdrawal cognitions in steps 2 and 3 of the re-
gressions (see Table II). However, the effect of positive spillove r disap-
peared when the interactions were entered into the equations in step 4.
No effect was found between the two variable s when commitment and job
satisfaction were include d in the equations (see Table III).
Response Variables. Hypotheses 2, 3, and 4 predicted an important re-
lationship among nonwork, coping strategie s, organization support, and
withdrawal cognitions. However, Table s II and III show no relationship be-
tween any of the three response variable s and withdrawal cognitions. There-
fore no empirical support is provide d for these hypothe ses.
Interaction Effects. Hypothesis 5(a), which predicted an interaction ef-
fect of importance of nonwork and coping strategie s on withdrawal cogni-
tions, was strongly supporte d by the data. As can be seen in Table II, this
interaction has a strong effect on withdrawal cognitions. The inte raction
Influences on Withdrawal Cogn itions 1525
Table
I.D
esc
rip
tive
Sta
tist
ics,
Re
lia
bil
itie
sa
nd
Inte
rco
rre
lati
on
sA
mo
ng
Re
sea
rch
Va
ria
ble
s(R
eli
ab
ilit
ies
inP
are
nth
ese
s)a
Me
an
SD
12
34
56
78
91
01
11
21
31
41
51
6
Dep
end
en
tvaria
ble
1.
Wit
hd
raw
al
cog
nit
ion
s4
9.4
21
1.8
5(.
95
)
Con
trol
va
riab
les
2.
Ge
nd
erb
0.3
20
.47
¯.21
3.
Ed
uca
tio
n6
.93
2.0
8¯.1
8.2
04
.Y
ea
rsin
occ
up
ati
on
14
.18
8.5
8¯.1
7.3
0.2
7
5.
Inco
me
39
58
71
76
76
¯.12
.44
.69
.52
Ind
ep
en
den
tvari
able
s
6.
Imp
ort
an
ceo
fn
on
wo
rkd
om
ain
28
.92
5.0
6¯.0
4¯.0
4.0
5¯.0
0.0
27
.P
osi
tive
no
nw
ork
-to
-wo
rksp
illo
ver
30
.09
5.2
9¯.0
8¯.0
5.1
4¯.1
3.0
3.2
3(.
74
)
8.
Wo
rk/n
on
wo
rkco
nfl
ict
19
.57
7.8
0¯.1
9.1
1.2
9.0
0.2
7.1
7.1
6(.
80
)9
.P
ers
on
al
cop
ing
28
.92
5.0
6.1
1¯.1
4.0
7.0
1.0
6.2
3.2
3¯.0
5(.
72
)
10
.O
rga
niz
ati
on
al
sup
po
rtfo
rn
on
wo
rk9
.00
3.7
0.1
3.0
1¯.2
2¯.1
4¯.1
8¯.1
1.0
5¯.2
5.1
8(.
74
)1
1.
Org
an
iza
tio
na
lco
mm
itm
en
t4
7.9
67
.40
.37
¯.14
.11
¯.05
.04
.10
.17
.00
.20
.13
(.7
6)
12
.Jo
bsa
tisf
act
ion
27
.15
5.4
6.5
0¯.1
1¯.0
3¯.0
7.0
5¯.0
2.0
9¯.1
6.1
2.2
4.4
6(.
78
)In
tera
ctio
nte
rms
13
.Im
po
rta
nce
of
no
nw
ork
do
ma
in´
pe
rso
na
lco
pin
g4
.19
19
.69
¯.24
.09
.05
.03
.11
¯.06
¯.03
¯.01
¯.01
.03
¯.13
¯.15
14
.Im
po
rta
nce
of
no
nw
ork
do
ma
in´
org
an
iza
tio
na
lsu
pp
ort
for
no
nw
ork
¯1.4
41
4.0
8.1
9.0
4¯.0
3.0
5.0
1.1
0.0
7¯.0
4.0
3¯.0
1.1
6.1
5¯.0
7
15
.P
ero
nsa
lco
pin
g´
org
an
iza
tio
na
l
sup
po
rtfo
rn
on
wo
rk
3.4
21
8.7
5.0
5¯.0
9.0
0¯.0
5¯.0
6.0
4.0
8.0
5¯.0
2.2
8.1
5.2
4¯.2
2.2
4
16
.Im
po
rta
nce
of
no
nw
ork
do
ma
in´
pe
rso
na
lco
pin
g´
org
an
iza
tio
na
lsu
pp
ort
for
no
nw
ork
2.1
58
5.6
9.1
8¯.0
3¯.0
6¯.0
5¯.0
7¯.0
8¯.0
8¯.0
7¯.1
6.1
9¯.0
2¯.0
1.1
8¯.0
1¯.0
0
aC
orr
ela
tio
ns
gre
ate
rth
an
.10
are
sig
nif
ica
nt
at
p<
.05
,th
ose
gre
ate
rth
an
.15
are
sig
nif
ica
nt
at
p<
.01
,a
nd
tho
seg
rea
ter
tha
n.1
8a
resi
gn
ific
an
t
at
p<
.00
1.
b0
=fe
ma
le1
=m
ale
.
1526 Cohen
was plotte d and is presented in Fig. 1(a). The pattern of the inte raction
shown in Fig. 1(a) was une xpected. The plot shows that among employe es
who attached low importance to nonwork domains there was a positive
relationship between personal coping strategie s and withdrawal cognitions;
among employe es who attached high importance to nonwork domains there
was a negative relationship between personal coping and withdrawal cog-
nitions. Hypothe sis 5(a) anticipate d the opposite patte rn.
Hypothesis 5(b), which predicted an inte raction effect between im-
portance of nonwork and organizational support, was strongly supporte d
by the data. As Table II shows, this inte raction had a strong effect on
withdrawal cognitions. The interaction was plotte d and is presented in Fig.
1(b). The patte rn of the interaction as shown in Fig. 1(b) was as expe cted
according to Hypothe sis 5(b). The plot shows that employees who assigne d
high importance to nonwork domains and perceived more support by the
Table II. Regression Results (Standardized Coefficients) for Withdrawal Cognitions as De-
pendent Variables
Withdrawal cognitions
Variables Step 1 Step 2 Step 3 Step 4
Control VariablesGende ra
¯.21** ¯.21** ¯.20** ¯.20**
Education ¯.26** ¯.22* ¯.20* ¯.19*Years in occupation ¯.17** ¯.21** ¯.20* ¯.22**
Income .24* .28* .26* .29**Spillover var iables
Positive nonwork-to-work spillover ¯.06 ¯.09 ¯.10Work/nonwork conflict ¯.17* ¯.15* ¯.13
Respon se variablesImportance of nonwork ¯.01 ¯.04
Personal coping .09 .09Organization support for nonwork .05 .08
Interaction sImportance of nonwork ´ personal coping ¯.25***
Importance of nonwork ´ organizationalsupport
.20**
Personal coping ´ organizational support ¯.06Importance of nonwork ´ personal coping ´organizational support
.05
R2 .09 .12 .13 .23
Adjusted R2 .07 .10 .10 .18F 5.32*** 4.94*** 3.58*** 4.72***
m R2.03 .01 .10
F for m R2 3.91* 0.87 6.42***
a0 = female 1 = male.
*p £ .05.
**p £ .01.
***p £ .001.
Influences on Withdrawal Cogn itions 1527
organizatio n had lower withdrawal cognitions; employe es who assigne d
low importance to nonwork domains and perceived more organizationa l
support had highe r withdrawal cognitions. Hypothe sis 5(c) , which pre-
dicted a three-way inte raction effect, was not supporte d by the data. Fi-
nally, Table III shows an inte resting inte raction of coping strategie s with
organizational support. The plot of this interaction (not presented in this
paper) shows that for employees who perceived the organization as sup-
portive , the more coping strategie s they applie d the weaker were their
withdrawal cognitions. For those who perceived the organization as not
Table III. Regression Results (Standardized Coefficients) for Withdrawal Cognitions as
Dependent Variables with Organizational Commitment and Job Satisfaction as ControlVariables
Withdrawal cognitions
Variables Step 1 Step 2 Step 3 Step 4 Step 5
Control variables (demograp hics )
Gende ra¯.21** ¯.11 ¯.11 ¯.10 ¯.11
Education ¯.26** ¯.20* ¯.16* ¯.17* ¯.15
Years in occupation ¯.17* ¯.10 ¯.14* ¯.14* ¯.15*Income .24* .09 .11 .09 .12
Control variables (attitudes)Organizational commitme nt .18** .20** .20** .17**
Job satisfaction .39*** .38*** .39*** .37***Spillover var iables
Positive nonwork-to-work spillover ¯.14* ¯.14* ¯.14*Work/nonwork conflict ¯.08 ¯.08 ¯.07
Respon se variablesImportance of nonwork ¯.02 ¯.04
Personal coping .06 .06Organization support ¯.06 ¯.01
Interaction sImportance of nonwork ´ personal
coping¯.18**
Importance of nonwork ´ organiza-
tion support
.14*
Personal coping ´ organization sup-
port¯.13*
Importance of nonwork ´ personal
coping ´ organization support
.05
R2 .09 .33 .36 .36 .41
Adjusted R2 .07 .31 .33 .33 .37F 5.32*** 17.23*** 14.36*** 10.52*** 9.44***
m R2.24 .03 .00 .05
F for m R2 37.39*** 4.20* 0.53 4.19**
a0 = female 1 = male.
*p £ .05.
**p £ .01.
***p £ .001.
1528 Cohen
Fig. 1. Interaction effects: (a) Interaction effect on withdrawal cognitions (importance of non-
work ´ personal coping strategies) . (b) Interaction effect on withdrawal cognitions (importance
of nonwork ´ organizational support).
Influences on Withdrawal Cogn itions 1529
supportive , the more coping strategies they applie d the stronge r were their
withdrawal cognitions.
Nonwork Domains and Work Attitudes. Hypothesis 6 examine d whether
nonwork variable s had an effect on withdrawal cognitions beyond the effect
of the two work-re lated variable s examined here, namely organizational
commitment and job satisfaction. The findings in Table III support Hy-
pothe sis 6. Despite the strong relationship of the two work attitude s, non-
work domain variable s entered in steps three to five , and in particular the
inte ractions, contributed significantly to the variance explaine d by the
demographic control variable s and the two work attitude s.
DISCUSSION
The aim of this paper was to augme nt the limited research into the
influences of nonwork domains on turnove r decision. Mobley (1982) argued
that as dual career familie s become more prevale nt, as nonwork values be-
come more central, and as more young people attach less importance to
a stable and secure career, prediction and unde rstanding of turnove r will
require inclusion of nonwork variable s. This study tested nonwork variable s
other than those reflecting family responsibilitie s, the nonwork variable s
commonly tested so far. Anothe r contribution was testing the relative effect
of nonwork and work-related variable s. The findings show that nonwork
domain variable s were meaningfully and significantly relate d to withdrawal
cognitions beyond the effect of work-related variable s. But this was a more
complex effect, as demonstrated in the two strong interaction effects: they
were stronge r than the main effects. The spillove r and response variable s
had a weak main effect and little can be conclude d about the effect of
the ir ordering in the regression equations.
The variance explaine d by the nonwork variable s examined here is not
very high. But nonwork domains are not expected to be a major determi-
nant of turnove r. Lee and Mowday (1987) who found no effect of nonwork
domains on withdrawal cognitions argued that more research is needed be-
fore any conclusions may be drawn about discontinuing the use of nonwork
domains in turnove r models. What can be conclude d from this study is
that nonwork conside rations should be incorporate d in future mode ls of
turnove r. Clearly, part of the reason for Lee and Mowday’s (1987) findings
is their use of only one indicator on nonwork domains. Lambert (1990)
argue d that in considering the full work/family nexus, research identifying
factors that lead workers to balance work and home will contribute greatly
to our unde rstanding of the relationship between work and family. While
the little research that has e xamine d nonwork domain relationship with
turnove r has focused on family responsibilitie s, this paper propose s several
1530 Cohen
diffe rent aspects of nonwork domains, such as personal and organizational
responses to work/nonwork inte rface and positive nonwork-to-work spil-
lover. The relationships of the variable s tested here show the need for di-
versity in the nonwork domains examined in the ir relation to turnove r.
The strong interaction effects found here show that the relationship of
nonwork domains with turnove r is complex, rather than simple and direct.
An earlier indication of the existence of moderators in the relationship of
nonwork domains to turnover was revealed by Muchinsky and Tuttle (1979) .
They found that the relationship between family re sponsibilitie s and turnove r
was moderated by whether the employe e was the primary or secondary wage
earner. The pattern of the inte ractions found in this study shows that non-
work variable s had diffe rential relationships depending on the importance of
nonwork domains to employe es. The two groups of employees, name ly those
who valued nonwork domains and those who did not, seemed to respond
diffe rently to variable s representing work/nonwork interface.
The inte raction of the variable s “importance of nonwork domains” ´“organizational support for nonwork” shows that organizations can reduce
withdrawal cognitions of employe es by being more supportive of the ir non-
work needs. This is important in light of Lambert’s argument (1990) that
we must conside r how workplace policie s are he lping workers balance work
and family responsibiliti es in orde r to evaluate the ir merit adequate ly.
Crouter (1984) argue d that only persuasive data would convince employe rs
that it may be in their best inte rest to support familie s in ways that mini-
mized the like lihood that employees would bring family proble ms with them
to the work. This inte raction also shows that organizations should realize
that being more supportive will have a positive impact only for employe es
who assign high importance to the ir nonwork domains. More research in
this area is imperative , since most employe rs are like ly to pursue policie s
and practice s that support familie s only if there is persuasive documenta-
tion that such support will enhance the opportunity for the positive aspects
of family life to make the ir mark on the work organization (Crouter, 1984) .
Kanter (1977) identifie d two opposing types of responses to work¯non-
work issues adopte d by employers. One response is based on the myth of
separate work and nonwork worlds. Such employers expect employees to
leave their nonwork live s at the office or factory door, and are concerned
mainly with e mploye e s’ work behaviors. A second type of response is
termed “integration.” Here employers take over responsibilitie s not only
for employe es’ work lives but also for aspects of their nonwork lives. This
response aims to close the work¯nonwork gap, can reduce the conflict be-
tween work and nonwork domains, and should result in more favorable
attitude s toward the organization. The findings here sugge st that organiza-
tions who wish to decrease withdrawal cognitions should consider imple-
Influences on Withdrawal Cogn itions 1531
menting respect practices especially as the numbers of employed women
and dual earner familie s continue to grow.
The inte raction of the variable s “importance of nonwork domains” ´“personal coping” shows that employees who gave low importance to non-
work domains but were more effective in the ir strategies for coping with
multiple domains were more like ly to remain with the organization. This
patte rn of interaction was not anticipate d. One possible explanation is that
for people who regard nonwork domains as unimportant, applying effective
coping strategie s to deal with pressures from multiple domains is enough
to reduce work/nonwork interface tensions, which are minimal for them.
For this type of employe e, more effective coping strategie s should decrease
the ir withdrawal cognitions. On the othe r hand, for those who regard non-
work domains as important, additional coping strategies may be insufficie nt
to bridge the conflicts that arise from work/nonwork interface . Because of
the high importance they assign to nonwork domains, there is greater po-
tential for conflict. For this type of employe e more coping strategie s lead
to more frustrations resulting from their failure to cope with claims from
multiple domains, and can result in a greater desire to leave the situation
in orde r to find a setting in which they will be able to fulfill their nonwork
needs better. This explanation, as well as this particular finding, need rep-
lications in future research. Such research should further examine the pos-
sibility that the relationship between nonwork domains and withdrawal
cognitions is not a simple line ar one, and search for othe r inte raction effects
among nonwork domain variable s. For example , the interaction between
coping and organizational support was not anticipate d in this study and
may sugge st some directions for conceptual developme nts and future re-
search. More research into the interrelationships among nonwork variable s
is needed to increase our understanding of the processes through which
nonwork variable s are related to turnove r.
This research responds to the call of Near, Rice, and Hunt (1980) to
test the relationship between nonwork aspects to work behaviors and atti-
tudes relative to the work-se tting variable s because such research has im-
portant implications for policy issue s. The fact that organizational
commitment and job satisfaction explaine d much of the variance in with-
drawal cognitions suggests that the conditions in the work setting are the
main factor in shaping an employe e’s decision to stay or to leave . However,
policym ake rs cannot ignore the nonwork domain aspe cts as they were
found to have a significant relationship beyond the effect of the two work
variable s. The fact that the effect of nonwork domains, and the inte ractions
in particular, remained strong and consistent in the equations with and
without commitment and job satisfaction, emphasize s the importance of
the nonwork variable s examined here as determinants of turnove r.
1532 Cohen
The relationship of the control variable s, gende r in particular, is in-
teresting. The latte r is considered important because of the need to study
me n and wome n simultane ously in orde r to assess systematically how
work/family linkage s vary for men and women workers (Lambert, 1990) .
The relationship of all four control variable s was significant in all steps of
the regression equations that analyze d the relationship of research variable s
without commitment and satisfaction (see Table II). However, when com-
mitment and job satisfaction were entered (see Table III, step 2) the effect
of the control variable s became much weake r and disappe ared in some
cases. This finding shows that the control variable s were relate d to com-
mitment and satisfaction, and not to the nonwork domain variable s exam-
ined here. The disappe arance of the effect of gende r in the equations with
commitment and satisfaction supports Crouter’s (1984) argument that the
effect of nonwork doe s not depend so much on gende r per se, as on gen-
der-relate d roles. A logical extension of this argument is that single -parent
fathe rs are like ly to resemble mothers in the ir depiction of familial influ-
ence on work, as are fathers who are nontraditional and take on respon-
sibility for childcare and household tasks.
Finally, several limitations of this research should be mentioned. First,
the use of cross-sectional, corre lational data does not allow us to draw
causal inferences concerning the various hypothe sized relationships. Sec-
ond, because the study is based on self-report data, incurring the possibility
of source bias or general method variance , data must be inte rpreted cau-
tiously. The design of this research in terms of the depende nt variable is
appropriate however, because conceptually nonwork domains are expe cted
to be relate d to withdrawal cognitions and not actual turnove r (Steers &
Mowday, 1981; Price & Mueller, 1981) . Third, one can argue that turnove r
intentions are predicted to a greater extent by the presence of negative job
attitude s such as stress or burnout. Therefore, in orde r to support the case
that nonwork variable s explain work withdrawal beyond the contribution
of work-specific variable s future research should test negative work atti-
tude s inste ad of commitment and satisfaction, which represent positive
work attitude s and were tested here. Despite the ir limitations, the findings
of this study have shown that trying to explain work attitude s and behaviors
strictly in terms of work conditions and experiences is limiting. To try to
unde rstand the individual at work, not only his or her work life , but also
his or her life away from work must be considered (Kirchmeyer, 1992) . As
suggested by Croute r (1984) , future research should continue in this direc-
tion and attempt to measure specific outcomes of spillove r from the family
to the workplace . Some such outcome s may be sensitive to family influe nces
while others may not. As for turnove r, the influence of nonwork factors
on employe e turnove r is perhaps one of the richest areas for future work
Influences on Withdrawal Cogn itions 1533
(Mowday, Porter, & Steers, 1982) . With the increasing proportions of dual
career familie s and women in the workforce , nonwork domain considera-
tions may have an even large r role in turnove r decisions. Hence, the need
arise s for much more work on this issue. Nonwork influe nces on turnove r
could generate a stream of research that will lead us to a greater under-
standing of turnove r. The present paper advocate s further research into
this area.
ACKNOWLEDGMENTS
The author thanks two anonymous reviewers for the ir helpful com-
ments and sugge stions.
REFERENCES
AIKEN, L. S., & WEST, S. G. Multiple regression: Testing and interpreting interactions, Ne wburyPark: Sage Publications, 1991.
BECKER, H. S. Notes on the concept of commitment. American Journal of Sociology, 1960,66, 32-40.
BEUTELL, N. J., & GREENHAUS, J. H. Integration of home and nonhome roles: Women’sconflict and coping behavior. Journal of Applied Psychology, 1983, 68, 43-48.
BLEGE N, M. A., MUE LLE R, C. W., & PRICE, J. L. Measureme nt of kinship responsibilityfor organizational research. Journal of Applied Psychology, 1988, 73, 402-409.
COHEN, A. Work commitment in relation to withdrawal intentions and union effectiveness.Journal of Business Research , 1993, 26, 75-90.
COHEN, A., & KIRCHMEYER, C. A multidimensional approach to the relationship betweenorganizational commitment and nonwork domains. Journal of Vocational Behavior, 1955,
46, 189-202.COTTON, J. L., & TUTTLE , J. M. Employe e turnover: A me ta-analysis and review with
implications for research. Academ y of Managem ent Review, 1986, 11, 55-70.CROUTE R, A. C. Spillover from family to work: The neglecte d side of the work¯family in-
terface. Hum an Relations, 1984, 37, 425-442.DALTON, D. R., & TODOR, W. D. Turnover, transfer, absentee ism: An interdependent per-
spective. Journal of Managem ent, 1993, 19, 193-219.FRONE, M. R., RUSSELL, M., & COOPER, L. M. Antecedents and outcomes of work¯fam-
ily conflict: Testing a mode l of the work¯family interface. Journal of Applied Psychology,1992, 77, 65-78.
GILBERT, L. A., & HOLAHAN, C. K. Conflicts between student/professional, pare ntal, and
self-development roles: A comparison of high and low effective copers. Hum an Relations,1982, 35, 635-648.
GOOD, L. K., SISLER, G. F., & GENTRY, J. W. Antece dents of turnover intentions amongretail manage ment personnel. Journal of Retailing, 1988, 64, 295-314.
HALL, D. T. A model of coping with role conflict: The role behavior of college educatedwomen. Administrative Science Quarterly, 1972, 17, 471-486.
HALL, D. T., & RICHTER, J. Balancing work life and home life: What can organizationsdo to help? Academ y of Man agem ent Executive, 1988, 2, 213-223.
KABANO FF, B. Work and nonwork: A review of models, me thods, and findings. Psychological
Bulletin, 1980, 88, 60-77.KANTER, R. M. Work and family in the United States: A critical review and agenda for research
and policy. New York: Russell Sage Foundation, 1977.KIRCHMEYE R, C. Nonwork participation and work attitudes: A test of scarcity vs. expansion
models of personal resources. Hum an Relations, 1992, 45, 775-795.
1534 Cohen
KIRCHMEYE R, C. Nonwork to work spillover: A more balanced view of the expe riences
and coping of professional women and men. Sex Roles, 1993, 28, 1-22.
KIRCHMEYE R, C. Managing the work-nonwork boundary: An assessment of organizational
responses. Hum an Relations, 1995, 48, 515-536.
LAMBERT S. J. Processe s linking work and family: A critical review and rese arch age nda.
Hum an Relations, 1990, 43, 239-257.
LEE, T. W., & MOWDAY, R. T. Voluntarily leaving an organization: An empirical investigation
of Steers and Mowday’s mode l of turnover. Academ y of Managem ent Journal, 1987, 30,
721-743.
MARKS, S. R. Multiple roles and role strain: Some notes on human energy, time and com-
mitment. American Sociological Review, 1977, 42, 921-936.
MICHAELS, C. E., & SPECTOR, P. E. Causes of employee turnover: A test of the Mobley,
Griffet, Hand & Meglino model. Journal of Applied Psychology, 1982, 67, 53-59.
MILLER, H. E., KATERBERG, R., & HULIN, C. L. Evaluation of the Mobley, Horner,
and Hollingsworth mode l of employee turnover. Journal of Applied Psychology, 1979, 64,
509-517.
MITRA, A., JENKINS, G., DOUGLAS, & GUPTA, N. A me ta-analytic review of the rela-
tionship between absence and turnover. Journal of Applied Psychology, 1992, 77, 879-889.
MOBLE Y, W. H. Em ployee turnover, causes, consequences , and control. Reading, Addison-
Wesley, 1982.
MOBLEY, W. H., GRIFFETH, R. H., HAND, H. H., & MEGLINO , B. M. Review and
conceptual analysis of the employee turnover process. Psychological Bulletin, 1979, 86,
493-522.
MOWDAY, R. T., KOBERG, C. S., & McARTHUR, A. W. The psychology of the withdrawal
process: A cross validational test of Mobley’s interme diate linkages model of turnover in
two samples. Academ y of Managem ent Journ al, 1984, 27, 79-94.
MOWDAY, R. T., PORTER, L. M., & STEERS, R. M. Em ployee-organizational linkage. Ne w
York: Acade mic Press, 1982.
MUCHINSKY, P. M., & TUTTLE , M. L. Employe e turnover: An empirical and methodo-
logical asse ssment. Journal of Vocational Behavior, 1979, 14, 43-77.
NEAR, J. P., RICE, R. W., & HUNT, R. G. The re lationship between work and nonwork
domains: A review of empirical research. Academ y of Managem ent Review, 1980, 5, 415-
429.
O’REILLY, C. A., & CHATMAN, J. Organizational commitment and psychological attach-
me nt: The effects of compliance, identification and internalization on prosocial behavior.
Journal of Applied Psychology, 1986, 71, 492-499.
ORTHNER, D. K., & PITTMAN, J. F. Family contributions to work commitme nt. Journ al of
Marriage and Fam ily, 1986, 48, 573-581.
PARASURAMAN, S., & HANSEN, D. Coping with work stressors in nursing: Effects of adap-
tive ve rsus maladaptive strategies. Work and Occupations , 1987, 14, 88-105.
PORTER, L. W., & STEERS, R. M. Organizational, work, and personal factors in employee
turnover and abse nteeism. Psychological Bulletin, 1973, 80, 151-176.
PRICE, J. L., & MUE LLE R, C. W. Professional turnover: The case of nurses, SP Medical and
Scientific Books, New York, 1981.
RANDALL, D. M. Multiple roles and organizational commitment. Journal of Organizational
Behavior, 1988, 9, 309-317.
RICE, R. W., FRONE, M. R., & McFARLIN, D. B. Work-nonwork conflict and the perceived
quality of life. Journal of Organizational Behavior, 1992, 13, 155-168.
RUDD N. M., & McKENRY. P. C. Family influences on the job satisfaction of employed
mothers. 1986, 10, 363-372.
SCHRIESHEIM, C., & TSUI, A. S. De velopment and validation of a short satisfaction in-
strument for use in survey feedback interventions. Paper presented at the Western Academ y
of Managem ent Meeting., 1980.
SHAMIR, B. Soe antecede nts of work-nonwork conflict. Journal of Vocational Behavior, 1993,
23, 98-111.
Influences on Withdrawal Cogn itions 1535
STEERS, R. M., & MOWDAY, R. T. Employee turnover and post-decision accommodation
processes. In L. L. Cummings and B. M. Staw (Eds.), Research in organizational behavior(Vol. 3) , Greenwich, CT: JAI Press, 1981, pp. 235-281.
STEFFY, B. D., & JONES, J. W. The impact of family and caree r planning variables onorganizational, care er, and community commitment of professional women. Journal of
Vocational Behavior, 1988, 32, 196-212.STEEL, R. P., & OVALLE, N. K. A review and me ta-analysis of research on the relationship
between behavioral intentions and employee turnover. Journal of Applied Psychology, 1984,
69, 673-686.SUSSMAN, M. B., & COGSWELL, B. E. Family influences on job move ment. Hum an Re-
lations, 1971, 24, 477-487.TETT, R. P., & MEYER, J. P. Job satisfaction, organizational commitment, turnover intention,
and turnover: Path analyses based on meta-analytic findings. Personnel Psychology, 1993,46, 259-293.
TSUI, A. S., EGAN, T. D., & O’REILLY, C. H. Being different: Relational demography andorganizational attachment. Administrative Science Quarterly, 1992, 37, 549-579.
WILEY, D. L. The relationship between work/nonwork role conflict and job-related outcomes:Some unanticipated findings. Journal of Managem ent, 1987, 13, 467-472.
WILLIAMS, K. J., & ALLIGER, J. M. Role stressors, mood spillover, and perceptions ofwork-family conflict in employed parents. Academ y of Managem ent Review, 1994, 37, 837-
868.
BIOGRAPHICAL NOTE
AARON COHEN is a senior Lecturer in the De partment of Political Science at the Unive rsityof Haifa, Israel. He rece ived his PhD in Manageme nt at the Technion-Israe l Institute of Tech-
nology and taught 3 years at the University of Lethbridge in Alberta, Canada. His currentresearch interests include work commitment, and in particular organizational commitme nt and
union commitment, turnover, and union participation.
1536 Cohen