Professional Documents
Culture Documents
Edwards 2000
Edwards 2000
Constructs
Author(s): Jeffrey R. Edwards and Nancy P. Rothbard
Source: The Academy of Management Review, Vol. 25, No. 1 (Jan., 2000), pp. 178-199
Published by: Academy of Management
Stable URL: http://www.jstor.org/stable/259269 .
Accessed: 09/05/2014 18:46
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to The Academy
of Management Review.
http://www.jstor.org
NANCY P. ROTHBARD
NorthwesternUniversity
In recent years the amount of research into the to implement family-friendlypolicies (McShul-
linkages between work and family has grown skis, 1997). These policies help ease family de-
dramatically (Burke & Greenglass, 1987; Ecken- mands and, by doing so, reduce employee ab-
rode & Gore, 1990; Zedeck, 1992). This research senteeism and turnover(Landauer, 1997). Thus,
has been stimulated by fundamental changes in linkages between work and familyaffectorgan-
the substance and structureof work and family izational performance and family functioning,
roles, such as the increasing prevalence of dual- both of which are importantmarkers of societal
earner couples, the influx of women into the well-being (Diener & Suh, 1997).
workforce,and family arrangements that devi- Work-familyresearchers have identified nu-
ate fromtraditional gender-based roles. This re- merous mechanisms linking work and family
search has debunked the myth that work and (Burke & Greenglass, 1987; Evans & Bartolome,
family are independent (Blood & Wolfe, 1960; 1986; Lambert, 1990; Payton-Miyazaki & Bray-
Dubin, 1973), demonstrating instead that work field, 1976; Zedeck, 1992). Although their re-
and family are closely interconnected domains search has provided rich, detailed descriptions
of human life (Burke & Greenglass, 1987;Kanter, of the work-familyinterface, it has two major
1977; Voydanoff,1987). shortcomings. One is the sheer number of link-
Linkages between work and familyare impor- ing mechanisms described. Differenttermshave
tant to organizations, families, and society. been used for fundamentally similar mecha-
Many organizations are expanding operations nisms (Staines, 1980;Zedeck, 1992),and common
globally (van Bergeijk & Mensink, 1997) and, themes across mechanisms have been over-
therefore,require key employees to travel or looked. This profusion of terminology implies
work abroad. These assignments can strain distinctionsamong linking mechanisms that are
family relationships and compel employees to not conceptually meaningful and inhibits the
withdraw or resign (Shaffer & Harrison, 1998), accumulation of knowledge regarding a core set
which, in turn,hinders global operations. Anal- of linking mechanisms.
ogously, many families require income from Second, linking mechanisms are typically de-
both parents to cover expenses, and these dual- scribed with metaphoric language that provides
earner families place pressure on organizations no clear translation into propositions regarding
the relationship between specific workand fam-
We thank Jane E. Dutton, Kamela R. Edwards, and three ily constructs (Lambert, 1990;Rice, Near, & Hunt,
anonymous reviewers fortheir helpful comments during the 1980). This translation is necessary to move be-
development of this article. yond descriptive accounts of the work-familyin-
178
terface to rigorous research into causal relation- emergentproducts [that] summarize progress,
ships that link the work and family domains give direction,and serve as placemarkers.They
have vestiges of theorybut are not themselves
(Frone, Russell, & Cooper, 1994; Kingston, 1989; theory.Then again, few thingsare full-fledged
Lambert, 1990; Near, Rice, & Hunt, 1980; Rice et theories.The keylies in thecontext-whatcame
al., 1980). before,what comes next?(1995:389).
Here, we review and critique linking mecha-
nisms examined in work-familyresearch, and Withregard to work-familylinking mechanisms,
we translate these mechanisms into causal re- we review and critique what has come before,
lationships between specific work and family and we provide a foundation for developing
constructs. The article contains three major sec- what we believe should come next.
tions. In the firstwe review linking mechanisms
discussed in the work-familyliterature, noting
conceptual redundancies, clarifying important
MECHANISMSLINKINGWORKAND FAMILY
distinctions,and identifyingcommon themes. In
the second we explain how currentconceptual- As stated previously, numerous mechanisms
izations of linking mechanisms fail to specify linking work and family have been identified
basic properties of relationships between work (Burke & Greenglass, 1987; Evans & Bartolome,
and family constructs, such as the sign and 1984; Lambert, 1990; Payton-Miyazaki & Bray-
causal structureof these relationships. And in field, 1976; Zedeck, 1992). Here, we organize
the third we translate linking mechanisms into these mechanisms into six general categories:
theoretical statements framed as causal rela- spillover, compensation, segmentation, resource
tionships between work and family constructs. drain, congruence, and work-familyconflict.In
This translation resolves ambiguities regarding this section we review and integrate these
the causal processes underlying work-family mechanisms by consolidating terms that de-
linking mechanisms and yields conceptual scribe fundamentally similar processes, draw-
building blocks researchers may use to develop ing distinctions within and between each mech-
comprehensive theories of the work-familyin- anism, and identifying points of similarity
terface. across mechanisms. This review provides the
Before proceeding, we should clarify the the- necessary foundation for our critique and re-
oretical contribution we intend to provide. We specification of linking mechanisms as causal
do not presume to develop a comprehensive the- relationships between work and family con-
ory of the work-familyinterface.Such an under- structs.
taking would exceed the scope of a journal arti- To establish boundary conditions for this re-
cle, given what is required to transformwork- view, we firstdefine the termswork,family,and
family linking mechanisms from ambiguous linking mechanism. We define work as instru-
metaphor to formal theory. Rather, our goal is mental activity intended to provide goods and
more modest: we intend to respecifywork-family services to support life (Piotrkowski,Rapoport, &
linking mechanisms as causal relationships be- Rapoport, 1987).Work typically entails member-
tween work and familyconstructs.This goal be- ship in a market or employing organization that
fits the current status of theory regarding the compensates the worker forhis or her contribu-
work-familyinterface, which has yet to resolve tions (Burke & Greenglass, 1987;Kabanoff, 1980).
basic conceptual issues concerning the mean- Work may provide intrinsic rewards (Deci &
ing and causal structureof work-familylinking Ryan, 1985), but its primary goal is to obtain
mechanisms. extrinsic rewards (Locke & Latham, 1990).
Our respecification of work-family linking Family we define as persons related by bio-
mechanisms yields conceptual building blocks logical ties, marriage, social custom, or adop-
researchers may use to develop comprehensive tion (Burke & Greenglass, 1987; Piotrkowski et
theories of the work-familyinterface,as we later al., 1987). Like work, family signifies member-
illustrate. Our objectives echo Whetten's (1989) ship in a social organization to which the person
observation that most theorists do not generate contributes (Zedeck, 1992). However, these con-
new theoryfromscratch but, rather,improve on tributions are intended not to earn goods and
what currentlyexists. In this spirit we provide a services but, rather,to maintain the family and
version of what Weick describes as enhance its well-being.
These definitions of work and family are in- terms of work and family affect (i.e., mood and
tentionally broad, encompassing not only nu- satisfaction), values (i.e., the importance as-
clear families in which one or both parents work cribed to work and family pursuits), skills, and
but also working teenagers; single working overt behaviors. Other terms that capture the
adults with siblings, parents, or other relations; essence of spillover include generalization, iso-
and other persons who work and have immedi- morphism, continuation, extension, familiarity,
ate or extended families. and similarity (Staines, 1980; Zedeck, 1992).
We define a linking mechanism as a relation- Two versions of spillover have been discussed
ship between a work construct and a family in the work-familyliterature.One characterizes
construct. Linking mechanisms can exist only spillover as similarity between a construct in
when work and familyare conceptually distinct. the work domain and a distinct but related con-
Thus, we exclude situations in which work and structin the family domain (Judge& Watanabe,
family are too closely intertwinedto be consid- 1994; Zedeck, 1992). This version of spillover is
ered separately, a condition labeled identityor exemplified by the positive association between
integration in the work-familyliterature (Burke job and family satisfaction (Gutek, Repetti,
& Greenglass, 1987; Morf,1989; Payton-Miyazaki & Silver, 1988; Near et al., 1980; Piotrkowski et
& Brayfield,1976;Zedeck, 1992).This condition is al., 1987) and between work and family values
exemplified by family-runbusinesses, in which (Payton-Miyazaki & Brayfield,1976; Piotrkowski,
family members are also supervisors, cowork- 1979).
ers, or subordinates of one another (Payton- A second version describes spillover as expe-
Miyazaki & Brayfield, 1976). In addition, linking riences transferred intact between domains
mechanisms entail relationships that span the (Near, 1984; Near et al., 1980; Payton-Miyazaki &
work and family domains, as opposed to rela- Brayfield, 1976; Repetti, 1987),as when work fa-
tionships among constructs within either do- tigue is displayed at home (Eckenrode & Gore,
main (Lambert, 1990; Near et al., 1980). Finally, 1990). This version of spillover does not repre-
linking mechanisms may represent either sent a linking mechanism, because, by itself,it
causal or noncausal relationships (Morf,1989). does not entail a relationship between a work
We later elaborate alternative causal structures construct and a family construct. For instance,
underlying work-familylinkages. the display of work fatigue at home indicates
A final boundary condition concerns our focus that an experience generated in one domain is
on the individual level of analysis. Many re- exhibited in another domain, but it does not
searchers have examined linkages between in- indicate that a constructin the latter domain is
dividuals' work and family experiences and the affected. If such an effectoccurs, as when work
psychological dynamics that influence these fatigue inhibits the fulfillmentof familyrole de-
linkages (Burke & Greenglass, 1987; Eckenrode mands (Eckenrode & Gore, 1990; Greenhaus &
& Gore, 1990; Lambert, 1990; Zedeck, 1992). The Beutell, 1985),then a link between a work con-
volume and importance of this research justify struct and a family construct has been estab-
our examination of work-familylinkages at the lished.
individual level. We acknowledge the value of
studying work-familylinkages at other levels of
Compensation
analysis, such as dual-earner couples (Bielby &
Bielby, 1989),familysystems (Cox & Paley, 1997), Compensation represents effortsto offsetdis-
work and family institutions (Dubin, 1973), and satisfaction in one domain by seeking satisfac-
national cultures (Aryee, 1992; Keller, 1987). tion in another domain (Burke & Greenglass,
1987; Champoux, 1978; Lambert, 1990; Zedeck,
1992).Other termsthat capture the core meaning
Spillover
of compensation include contrast,complementa-
Spillover refers to effects of work and family rity,competition, regeneration, and heteromor-
on one another that generate similarities be- phism (Staines, 1980: Zedeck, 1992).
tween the two domains (Burke & Greenglass, Two formsof compensation have been distin-
1987; Evans & Bartolome, 1986; Lambert, 1990; guished in the work-familyliterature. First, a
Near et al., 1980; Staines, 1980; Zedeck, 1992). person may decrease involvement in the dissat-
These similarities usually are described in isfying domain and increase involvement in a
potentially satisfying domain (Champoux, 1978; domain, and (2) shifting involvement fromone
Evans & Bartolome, 1984;Lambert, 1990;Staines, domain to another will enhance overall satisfac-
1980; Zedeck, 1992). Involvement itself has been tion only if the latter domain provides valued
defined as the perceived importance of a do- rewards (Edwards, 1992; Rice, McFarlin, Hunt, &
main (Champoux, 1978; Lambert, 1990; Lobel, Near, 1985).
1991),time spent in a domain (Lobel, 1991; Small
& Riley, 1990),and attentiondevoted to a domain
Segmentation
(Kanter, 1977; Small & Riley, 1990; Voydanoff,
1987). Hence, this formof compensation may be Segmentation is the separation of work and
conceived as the reallocation of importance, family,such that the two domains do not affect
time,or attention froma dissatisfying domain to one another (Burke & Greenglass, 1987;Lambert,
a potentially satisfying domain. 1990;Zedeck, 1992).Segmentation originally was
Second, the person may respond to dissatis- viewed as a natural division of workand family,
faction in one domain by pursuing rewards in owing to the separation of the two domains in
another (Champoux, 1978; Kando & Summers, time and space and to the inherently different
1971; Zedeck, 1992). By "rewards," we mean ex- functionsthey serve (Blood & Wolfe, 1960;Dubin,
periences that may fulfillthe person's desires 1973). However, this view of segmentation has
and, by doing so, enhance his or her satisfaction been challenged by researchers who have dem-
(Porter & Lawler, 1968). This formof compensa- onstrated that work and family are closely re-
tion has been differentiatedfurtherinto supple- lated domains of human life (Burke & Green-
mental and reactive compensation (Kando & glass, 1987; Kanter, 1977; Voydanoff, 1987).
Summer, 1971; Zedeck, 1992). Consequently, segmentation is now viewed as
Supplemental compensation occurs when re- an active process whereby people maintain a
wards that are insufficientin one domain are boundary between work and family (Eckenrode
sought in the other domain (Evans & Bartolome, & Gore, 1990; Lambert, 1990; Morf, 1989; Near,
1986; Kando & Summers, 1971; Zedeck, 1992). 1984).
Here, the person seeks rewards in the latter do- This process was described by Piotrkowski
main that add to those in the formerdomain so (1979),who found that people may actively sup-
that the cumulative rewards across domains are press work-related thoughts, feelings, and be-
fulfilling (Kabanoff & O'Brien, 1980; Staines, haviors while in the family domain, and vice
1980). For example, a person with little auton- versa. The active separation of work and family
omy at work may seek autonomy outside ofwork may be viewed as a method of coping with
(Evans & Bartolome, 1984; Kabanoff, 1980).Reac- stress from either domain (Lambert, 1990;
tive compensation occurs when undesirable ex- Piotrkowski,1979; Rice et al., 1980) or as way of
periences in one domain are redressed by an maintaining a preferred degree of connection
individual's seeking contrasting experiences in between work and family (Kanter, 1977).
the other domain (Kando & Summers, 1971; Ze- Other termsthat capture the substance of seg-
deck, 1992),such as resting at home aftera tiring mentation include compartmentalization, inde-
day at work or throwing oneself into work to pendence, separateness, disengagement, neu-
avoid thinkingabout family problems (Evans & trality,and detachment (Lambert, 1990; Pryor,
Bartolome, 1986; Kando & Summers, 1971; Ze- 1987; Zedeck, 1992).
deck, 1992).
Although both supplemental and reactive
Resource Drain
compensation involve the pursuit of rewards in
an alternative domain, supplemental compen- Resource drain refers to the transferof finite
sation is prompted by insufficientpositive expe- personal resources, such as time, attention,and
riences, whereas reactive compensation arises energy, fromone domain to another (Eckenrode
fromexcess negative experiences. Both supple- & Gore, 1990; Piotrkowski, 1979; Small & Riley,
mental and reactive compensation are related 1990; Staines, 1980; Tenbrunsel, Brett, Maoz,
to compensation that entails shifting involve- Stroh, & Reilly, 1995). Resource drain is analo-
ment between domains, in that (1) in seeking gous to those formsof compensation that entail
rewards in another domain, the person may a shift of time or attention between domains
need to reallocate time and attention to that (Lobel, 1991; Small & Riley, 1990). However, as
PROBLEMSWITHCURRENT 1990; Near, 1984; Near et al., 1980). With few ex-
CONCEPTUALIZATIONSOF WORK-FAMILY ceptions, work-familylinkages are specified as
LINKINGMECHANISMS simple bivariate associations. This specification
overlooks numerous alternative causal struc-
The linking mechanisms reviewed above
tures that may generate relationships between
have undeniably enhanced our understanding
work and family constructs. For example, emo-
of the work-familyinterface. However, attempts tional spillover is often characterized as a pos-
to translate these mechanisms into relation- itive correlation between job and family satis-
ships between work and family constructs re- faction (Gutek et al., 1988; Judge & Watanabe,
veal several importantambiguities. Perhaps the 1994; Rice et al., 1980; Staines, 1980). However,
most fundamental ambiguity concerns the sign this correlation may be spurious, attributable to
of the relationship, meaning whether an in- a common cause, such as dispositional affect,
crease in one construct is associated with an rather than a causal relationship between emo-
increase or decrease in the other construct. tion in the two domains (Frone et al., 1994;Morf,
In some cases the sign of the relationship im- 1989; Staines, 1980; Zedeck, 1992). Likewise, a
plied by a linking mechanism is explicitly negative correlation between work and family
stated. For example, several investigators have involvement interpreted as compensation
asserted that spillover, compensation, and seg- (Champoux, 1978)may be spurious, arising from
mentation represent positive, negative, and null subscription to traditional gender roles that en-
relationships, respectively (Judge & Watanabe, courage greater work involvement formen and
1994;Near, 1984; Staines, 1980;Tenbrunsel et al., greater family involvement forwomen (Bielby &
1995; Zedeck, 1992). However, in many cases the Bielby, 1989; Gutek, Searle, & Klepa, 1991; Voy-
sign of the relationship is unstated or con- danoff,1988).
founded with the benefit or harm resulting from A null relationship between work and family
the relationship (Tenbrunsel et al., 1995).For ex- constructs is typically equated with segmenta-
ample, researchers often describe "positive" tion (Judge & Watanabe, 1994) but may result
spillover as work satisfaction that enhances from countervailing positive and negative ef-
family functioningand "negative" spillover as fects representing simultaneous spillover and
work dissatisfaction that hinders family func- compensation, as when job dissatisfaction in-
tioning (Eckenrode & Gore, 1990; Evans & Bar- duces family dissatisfaction but also prompts
tolome, 1984; Payton-Miyazaki & Brayfield,1976; the person to seek greater satisfaction with fam-
Piotrkowski,1979;Voydanoff,1989).Although the ily (Champoux, 1978;Kabanoff, 1980;Tenbrunsel
terms positive and negative suggest a differ- et al., 1995).
ence in sign, both of these forms of spillover Conversely, work and family constructs may
represent a single statistically positive relation- be correlated even when segmentation pre-
ship between work satisfaction and familyfunc- cludes a causal relationship between them, as
tioning. when dispositional variables create spurious re-
A more concrete example is provided by Lam- lationships between work and familyconstructs
bert, who presents path diagrams of relation- (Frone et al., 1994; Morf,1989; Zedeck, 1992). As
ships between work and family constructs and these examples illustrate, relationships be-
labels each path + or -, not to specify the sign tween work and family constructs cannot be
of the relationship but,rather,to convey whether meaningfully conceptualized or interpretedun-
the relationship yields "positive or negative re- less theircausal structuresare clearly specified
sults" (1990: 248). This use of + and - labels (Frone et al., 1994; Near, 1984; Near et al., 1980).
likely will hinder the translation of Lambert's A thirdambiguity entails the forces that give
(1990) diagrams into causal models, which in- rise to relationships between work and family
variably use + and - to represent the statistical constructs.These forces include the intentof the
sign of a relationship, as opposed to the benefit person; the behavior of others in the person's
or harm resulting from a relationship (James, work and familyenvironments;and policies and
Mulaik, & Brett,1982). practices attributable to organizations, govern-
A second ambiguity concerns the causal ments, and society. Given our emphasis on the
structureof the relationship between work and psychological dynamics of work-family link-
family constructs (Frone et al., 1994; Lambert, ages, we focus on intent, meaning whether a
family domains (e.g., renegotiate role demands Similarly, Nolen-Hoeksema, Parker, and Larson
and acquire resources needed to meet de- (1994) found that negative moods were associ-
mands), adapt to conditions in the domains (e.g., ated with pessimism and rumination, which
devalue or ignore disappointing aspects of work may cause people to neglect demands in the
or family), or avoid a domain in part or whole other domain.
(e.g., spend less time at workor with family).The The effects of mood in one domain on role
goal of coping is to enhance well-being associ- performance, rewards, and mood in the other
ated with work and family and, by doing so, domain are depicted in Figure lA, paths a, d,
increase overall well-being. Although these and e.' Figure 1A also shows that domain-
principles are anchored in role theory,they are specific mood (i.e., mood originating in a partic-
common to theories of how people interact with ular domain) influences general mood (path b;
situations and how these interactions influence Rice et al., 1985),which may also affect perfor-
affectand behavior (e.g., Edwards, 1992; French mance, rewards, and mood in the other domain
et al., 1982; Locke, 1976; Rice et al., 1985). (paths c, d, and e).
Mood spillover is largely unintentional, be-
cause it operates throughcognitive and motiva-
Spillover
tional processes that do not require intent. For
Recall that spillover refers to the effects of instance, negative moods inhibit problem solv-
work and family on one another that make the ing and reduce self-efficacy(Staw et al., 1994),
two domains similar. Similarity connotes a pos- and people need not tryto make these effects
itive relationship between a work constructand occur. However, mood spillover is more likely
a family construct (Judge & Watanabe, 1994; intentional when it refers to expressed (i.e.,
Tenbrunsel et al., 1995;Zedeck, 1992),regardless overtlydisplayed) rather than felt (i.e., internal)
of whether this relationship benefits or harms moods. Work and family roles carry expecta-
the person (cf. Lambert, 1990). The causal struc- tions regarding mood expression (Ashforth&
ture and intent of this relationship vary accord- Humphrey, 1993; Rafaeli & Sutton, 1987), and
ing to the work and family constructs involved. these expectations promptpeople to selectively
Here we examine mood, values, skills, and be- display appropriate moods. For instance, anger
havior, given the prevalence of these constructs and irritationfromfamily problems may be in-
in spillover research (Burke & Greenglass, 1987; tentionally concealed by service workers (Ash-
Eckenrode & Gore, 1990; Lambert, 1990; Near, forth& Humphrey, 1993) but deliberately dis-
1984; Near et al., 1980; Staines, 1980; Voydanoff, played by bill collectors (Sutton, 1991). Thus,
1989; Zedeck, 1992). intentregulates the degree to which feltmood is
Mood. Mood spillover occurs when mood in manifested as expressed mood, and people reg-
one domain affects mood in the other domain. ulate expressed moods to fulfill role expecta-
This effect may be explained in terms of a tions, enhance role performance, and receive
causal sequence that links mood in one domain role rewards.
to performance and rewards in the other.Specif- Values. The spillover of values between work
ically, positive moods enhance cognitive func- and familysuggests two causal structures.First,
tioning, increase task activity and persistence, work and family are socializing forces that af-
and promote positive interactions with others, fectvalues regarding life as a whole, and these
each of which facilitates role performance life values influence values specific to a domain
(Staw, Sutton, & Pelled, 1994). Role performance (Kanter, 1977;Payton-Miyazaki & Brayfield,1976;
brings intrinsicand extrinsic rewards, which, in Piotrkowski, 1979; Repetti, 1987). This process
turn,enhance mood. signifies an indirect relationship between work
By the same argument, negative moods from and familyvalues, mediated by general life val-
one domain inhibit role performance and re- ues (Figure lB, paths b and c). Second, values in
wards in the other domain, producing negative one domain may directly affect values in the
moods in that domain. For example, Barling and
Macewen (1992) found that negative moods orig-
inating from work interferedwith family rela- ' Note that family mood refers to how the person feels
tionships and familyrole performance,which, in about his or her family-not the mood of members of the
turn,were linked to dissatisfaction with family. person's family.
FIGURE 1
Causal StructuresforSpilloverBetweenWorkand Family
b b
+ a General + a General
Wo Work
skills > behav or
,~~~ b
+ a knowledge + a scripts,and
'-
-Fam ily AFamily
behavior
other domain (Figure lB, path a). For example, & Kernan, 1987; Lord & Maher, 1991) that apply
people in jobs where obedience is valued over across life domains (Figure lC, paths b and c).
self-directiontend to emphasize obedience over For example, problem-solving skills developed
self-directionin their children (Kohn, 1963; Pay- at work may enhance one's analytical reasoning
ton-Miyazaki & Brayfield, 1976; Pearlin & Kohn, capabilities, which may, in turn,facilitate solv-
1966). For either causal structure,the relation- ing familyproblems, even when specific aspects
ship between work and familyvalues may arise of problems in the two domains differ.
from the unintentional transmission of in- Second, skills obtained in one domain may be
grained value structuresbetween domains (Lord directlyexported to the otherdomain (Figure 1C,
& Maher, 1991) or fromintentional strivings for path a; Repetti, 1987). For instance, Crouter
value consistency between work and family, (1984a) found that employees applied participa-
particularly forpeople who desire a consistent tive management skills acquired at workto fam-
self-concept (Cialdini, Trost,& Newsom, 1995). ily situations. This transferof specific skills be-
Skills. The spillover of work and family skills tween domains does not entail the development
implies two causal pathways. First, skills ob- of general schema but, instead, implies a direct
tained in one domain may be abstracted into transportof skills fromone domain to another.
general knowledge structures,or schema (Lord Research on learning transfer (Guberman &
Greenfield, 1991) suggests that the skill transfer person to increase involvement or seek rewards
between work and family is more likely when in the other. Compensation refersto a decrease
skills are abstracted into general knowledge in involvement or rewards in the dissatisfying
structures, because this abstraction enables domain, coupled with an increase in these con-
skill transfer, despite differences in specific structs in the other domain. Therefore,the rela-
problems in the two domains. Consequently, the tionship between these constructs across do-
indirect effect in Figure 1C (paths b and c) is mains is negative (Staines, 1980; Tenbrunsel et
more likely than the direct effect(path a). Both al., 1995; Zedeck, 1992).Moreover, compensation
indirect and direct skill transfer are typically is intentional,forit represents active attempts to
intentional, assuming people deliberately mus- reallocate involvement or seek alternative re-
ter skills to meet role expectations and enhance wards (Burke & Greenglass, 1987; Champoux,
role performance.However, unintentional trans- 1978; Lambert, 1990; Zedeck, 1992). However,
fermay occur when skills become embedded in compensation implies different causal struc-
schema that are firmlyingrained and require tures, depending on whether it refers to the re-
little conscious informationprocessing (Lord & lationship between work and family involve-
Maher, 1991). ment or rewards.
Behavior. Behavioral spillover may follow a Recall that involvement may be viewed as
causal structure similar to that for values and time, attention,or importance associated with a
skills. For instance, behaviors developed in one domain. The relationship linking these con-
domain may become ingrained as habits or structsacross domains is direct. For instance, in
scripts that influence behaviors across domains response to declining work satisfaction, a per-
(Figure 1D, paths b and c; Champoux, 1978; son may devote less time to work and more to
Greenhaus & Beutell, 1985; Lord & Kernan, 1987; family (Figure 2A, path b; Staines, 1980;Zedeck,
Staines, 1980; Zedeck, 1992). Alternately,behav- 1992).2 This relationship is imperfect,because
iors in one domain may directly influence be- time may be allocated to and from domains
haviors in the other domain without becoming other than work or family, such as personal or
generalized as habits or scripts (Figure 1D, path communityactivities. The relationship between
a). For example, studies show that teachers de- work and family attention is also direct, since
velop interaction patterns with students that di- the person may shiftattentionfroma dissatisfy-
rectly shape their behavior as parents (Ispa, ing domain directly to a potentially satisfying
Gray, & Thornburg,1984). domain (Figure 2B, path b; Edwards, 1992;Evans
The direct transferof behavior between work & Bartolome, 1986; Lambert, 1990). Finally, the
and family is likely when situational cues (e.g., relationship between work and family impor-
work and family role requirements) in the two tance is direct (Figure 2C, path b), because a
domains are similar (Mischel, 1977). These two person who assigns less importance to a dissat-
effectscombine additively, such that the trans- isfyingdomain may seek satisfaction by ascrib-
ferof behaviors between domains is most likely ing greater importance to other potentially sat-
when behaviors have been internalized as hab- isfyingdomains (Lobel, 1991).
its or scripts and situational cues in the domains Like time,the relationships between workand
are similar (Pervin, 1989; Schneider, 1983). We family attention and importance are imperfect,
posit that behavioral spillover is unintentional given that attentionand importance may be dis-
forbehaviors that are habitual or scripted (Ash- tributed among domains other than work and
forth& Fried, 1988;Lord & Kernan, 1987)or when
situational cues in work and family are similar
and strong (Mischel, 1977). In contrast, behav- 2Some researchers may preferto conceptualize compen-
sation as a response to increased dissatisfaction, as op-
ioral spillover is intentional for behaviors the
posed to decreased satisfaction. To accommodate this alter-
person deems applicable to role expectations in native conceptualization, one can modify the causal
both domains (Crouter, 1984a,b). structuresin Figure 2 by (1) replacing satisfaction with dis-
satisfaction and (2) reversing the signs leading to and from
satisfaction so that the arrows fromsatisfaction in Figures
Compensation 2A, 2B, and 2C are negative and the arrows to and from
satisfaction in Figure 2D are negative and positive, respec-
As discussed previously, compensation occurs tively.These modified causal structureswill be conceptually
when dissatisfaction in one domain prompts a equivalent to those shown in Figure 2.
FIGURE 2
Causal StructuresforCompensation Between Work and Family
_b _b
am~r
ily F
l a Wor 0W a Work
satisfacto + importance rewards + satisfaction
b ~~~~b
family. Shiftingtime, attention,and importance tle social contact, a person may performfamily
to another domain may increase satisfaction for roles to obtain contact with family members) or
two reasons. First, time and attention are re- extrinsic (e.g., insufficientrecognition at work
sources formeeting role expectations and there- may prompt the person to performfamily roles
fore may enhance role performance and bring to earn recognition from family members). Al-
rewards, contributing to satisfaction. Second, though the relationship between work and fam-
time, attention,and importance intensifythe ef- ily rewards is mediated by work satisfaction
fects of rewards on satisfaction, because focus- and family role performance, the link between
ing time and attention on a reward heightens domains (i.e., path b) is not mediated by a gen-
satisfaction with that reward (Edwards, 1992; eral life construct and, therefore,is a direct ef-
Greenhaus & Beutell, 1985), and important re- fect.
wards elicit greater satisfaction than unimpor- This causal structureapplies to both supple-
tant rewards (Rice et al., 1985). mental and reactive compensation. As stated
Figure 2D shows a hypothesized causal struc- earlier, supplemental compensation occurs
ture for compensation as seeking rewards in when insufficientrewards in one domain de-
another domain. This structure indicates that crease satisfaction in that domain and invoke
decreased work rewards lead to decreased work the pursuit of rewards in the other domain-a
satisfaction, which then leads to familyrole per- process that corresponds directly to Figure 2D.
formance intended to yield desired family re- In contrast,reactive compensation occurs when
wards. These rewards may be intrinsic to role excess undesirable experiences in a domain de-
performance itself (e.g., when work provides lit- crease satisfaction in that domain and lead to
FIGURE 3
Causal Structures forSegmentation Between Work and Family
p or mood
upe on General
V ofwork )_> V mo
h obehavior
Note:Formsimpli ,thee
fuuppression t effects
the
ofabitse a
FoSipiiy
Note a egmentation that
hsefgrsdpc
suppessenso wr theeairo
effectsof aiybhvo
work on family.Each figure may be reversed to depict segmentation that dampens the
effects of family to work. Also, Figure 3A refers to positive mood (e.g., satisfaction,
pleasure, joy), rather than negative mood (e.g., dissatisfaction, distress, anger).
the pursuit of contrasting pleasurable experi- family constructs (Staines, 1980; Zedeck, 1992).
ences in the other domain. Although these two However, a null relationship alone does not nec-
formsof compensation differon the surface, un- essarily signifysegmentation, because positive
desirable experiences that promptreactive com- and negative effectsbetween work and family
pensation typically can be reframedas the lack constructs may combine to yield a null relation-
of pleasurable or rewarding experiences (e.g., ship (Champoux, 1978; Kabanoff, 1980; Tenbrun-
excess noise implies insufficientquiet and ex- sel et al., 1995). Rather, segmentation requires
cess fatigue implies insufficientrest). Thus, the that all direct and indirecteffectsbetween work
causal structurein Figure 2D applies to reactive and family constructs are null, thereby indicat-
compensation, if undesirable experiences are ing a true lack of causal linkages between do-
reconstrued as insufficientrewards. mains. Segmentation is also intentional, for it
represents deliberate efforts to inhibit un-
wanted interferencebetween domains (Lambert,
Segmentation
1990; Piotrkowski, 1979; Rice et al., 1980) or to
Segmentation refers to the active separation maintain a preferred degree of connection be-
of work and family so that the two domains do tween domains (Kanter, 1977).
not influence one another. Segmentation pro- Regarding causal structure, segmentation
duces a null relationship between work and represents attempts to reduce the relationship
FIGURE 5
Causal StructuresforCongruence Between Work and Family
mood alue
g fective % enerl
disposition life values
? ?
b b
Family Family
a ?
General K eeaviloral
aptitudeEs X stylesv
evening shiftwork makes it physically impossi- Watson & Clark, 1984). Likewise, overarching
ble to spend time with children during theirnor- life values may create similarity between val-
mal waking hours (Burke & McKeen, 1993; ues pertaining to work and family (Figure 5B;
Shamir, 1983), or when the intrusion of family Schwartz & Bilsky, 1987). Analogously, general
problems prevents an individual fromfocusing aptitudes and intelligence may contribute to
attention on work responsibilities (Cooke & skills specific to work and family (Figure 5C;
Rousseau, 1984;Crouter, 1984b;Voydanoff,1988). Lubinski & Dawis, 1990; Staines, 1980). Finally,
similarity of behaviors exhibited in work and
family domains (Greenhaus & Beutell, 1985;
Congruence
Staines, 1980) may arise from general behav-
Congruence indicates that work and family ioral styles (Figure 5D; Ispa et al., 1984;Murphy,
constructsare similar, owing to a common cause 1996).
(Frone et al., 1994; Morf, 1989; Staines, 1980; We suggest that these relationships are pri-
Zedeck, 1992),thereby signifyinga positive spu- marily unintentional, because the general con-
rious relationship. This relationship may apply structs shown in Figure 5 induce similarity be-
to a variety of constructs, such as those linked tween work and family constructs without
through spillover. For example, a relationship conscious effortby the person. However, inten-
between work and family mood may arise from tional relationships may occur, as when people
dispositional affect(Figure 5A; Frone et al., 1994; strive for consistency in values or behaviors
FIGURE 6
Causal StructuresforWork-FamilyConflict
across life domains (Cialdini et al., 1995;Mischel compete with familydemands forfamilytime or
& Peake, 1983),or they purposely apply general attention,as indicated by the negative and pos-
skills to specific work and family problems itive signs on paths a and b, respectively. As
(Pearlin & Schooler, 1978). with resource drain, the shiftof time or attention
between domains results fromintentional allo-
cation decisions.
Work-FamilyConflict
Strain-based conflict suggests that increased
The three formsof work-familyconflictidenti- strain in one domain reduces personal capabil-
fied by Greenhaus and Beutell (1985) imply dif- ities needed to meet demands in the other do-
ferent relationships between work and family main, thereby inhibiting role performance (Fig-
constructs. As argued previously, time-based ure 6B). The impact of strain in one domain on
conflictis a formofresource drain, in which time capabilities in the other domain may be direct
or attention transferredfroma domain hinders (Figure 6B, path a) or indirect,mediated by over-
role performance in that domain but facilitates all well-being (e.g., life satisfaction and mental
role performance in the receiving domain. This and physical health; Figure 6B, paths b and c).
process is illustrated in Figure 6A, which shows Available evidence suggests that this relation-
that a decrease in family time or attention ship is unintentional. For example, work may
would reduce family role performance (path e) produce fatigue, tension, and frustrationthat in-
and increase work time or attention (path c), terferewith familyrole performance(Greenhaus
thereby increasing work role performance (path & Beutell, 1985), without intentional effortsto
d). Figure 6A also shows that work demands create this interference.
TABLE 1
Sign, Causal Structure,and Intent of Work-FamilyLinking Mechanisms
Sign
Causal structure 0 +
Intentional relationships
Unintentional relationships
Direct effect Resource drain Spillover
Time-based conflict Behavior-based conflict
Strain-based conflict
Finally, behavior-based conflict represents a 1987; Rice et cl., 1980; Voydanoff,1989; Zedeck,
formof behavioral spillover (Figure 6C, path a or 1992).
paths b and c) in which behavior transferred This specification also has several important
from one domain inhibits role performance in implications forwork-familyresearch. First,the
another domain (Figure 6C, path d). This rela- location of each linking mechanism within Ta-
tionship is presumably unintentional, because ble 1 constitutes a set of theoretical conditions
the transferredbehavior interfereswith role per- that may be examined empirically-for exam-
formance and, therefore,reduces rewards re- ple, compensation is specified as a negative,
ceived by the person (Greenhaus & Beutell, direct, intentional relationship between work
1985). and family constructs. All three of these condi-
tions should be empirically verified before con-
cluding that compensation has occurred. In con-
IMPLICATIONSFOR RESEARCHON WORK-
trast,in previous studies scholars have inferred
FAMILYLINKINGMECHANISMS
compensation solely froma negative correlation
Our specification of sign, causal structure, between work and familyconstructs(Bolger, De-
and intentregarding work-familylinking mech- Longis, Kessler, & Wethington,1989;Champoux,
anisms is summarized in Table 1. This specifi- 1978; Kabanoff & O'Brien, 1980), thereby over-
cation provides theoretical precision lacking in looking the crucial issues of causal structure
previous discussions of work-family linking and intent(Judge & Watanabe, 1994;Near et al.,
mechanisms-a deficiency that scholars have 1987).
lamented repeatedly in the work-familylitera- Second, linking mechanisms within a single
ture (Burke & Greenglass, 1987; Kabanoff, 1980; cell constitute alternative explanations for ob-
Kingston, 1989; Lambert, 1990; Near et al., 1980, served relationships between work and family
FIGURE 7
Continued
+ s-r~~~~~
performance rewards + MOod timre
+ im
Suppressio -enera
of mood mood A~ ~~~~
allocatim
llcation
e
(D) Adding intentional time allocation decisions and segmentation as the suppression
of mood effectson work and familyrole performance
Table 1 captures mostofthe work-family linking causes. Third,forces other than intentthat influ-
mechanisms studied to date and suggests new ence work-familylinkages may be considered,
categories of relationships between work and such as familystructure,workplace policies, and
family constructs. However, one may elaborate societal and cultural norms. Thus, Table 1 pro-
upon Table 1 to generate furtherpossibilities. vides a useful point of departure fordeveloping
First,the classification of sign as positive, nega- more elaborate views of relationships between
tive,and null may be extended to include nonlin- work and familyconstructs.
ear relationships. For example, mood spillover
may be morepronounced fornegative moods than
for positive moods, implying a curvilinear rela- AREASFOR FURTHERTHEORETICAL
tionship between work and familymood. Second, DEVELOPMENT
causal structuremay be expanded to include in- Our central goal in this article has been to
directeffectswithmultiplestages ofmediation,or translate work-familylinking mechanisms from
spurious associations owing to several common their currentmetaphoric state into causal rela-
tionships between work and family constructs. family mood (Figure 3A). This model (Figure 7D)
This translation provides theoretical building emphasizes that personal intent moderates re-
blocks that may be used to develop comprehen- lationships that link workand familyconstructs,
sive theories of the work-familyinterface. To therebycasting the person as an active agent in
illustrate how these building blocks may be managing the work-familyinterface.
used, we draw fromour specification of mood We do not claim that the model in Figure 7D is
spillover in Figure IA. We first extend this comprehensive or superior to other possible
model to capture mood spillover not only from models. Rather, we use it to illustrate how our
work to family but also from family to work, respecified work-family linking mechanisms
which yields Figure 7A. serve as building blocks to develop theories of
This extended model provides two insights not the work-familyinterface.These building blocks
evident fromFigure lA. First,mood spillover may may also be woven into existing theories in
operate throughreciprocal relationships,in which work-familyresearch. For example, Frone, Yard-
work mood influences family role performance, ley, and Markel (1997) present a model of the
which,in turn,affectsfamilymood, workrole per- work-familyinterface that includes work and
formance,and work mood. These reciprocal ef- familytime, behavior, and satisfaction, but they
fectsmay lead to upward or downward spirals in do not link these constructs across domains.
workand familymood over time. Second, general These linkages may be developed by drawing
mood may influence role performance for both fromour respecified work-familylinking mech-
work and family. These effectsfollow a causal anisms. Other models of the work-familyinter-
structuresimilar to mood congruence (Figure 5A) face may be enhanced similarly.
but show that general mood may influence do- In addition to developing more complete mod-
main-specificmood-not directly,but by affecting els of the work-familyinterface,we believe an-
role performanceand rewards. other agenda for furthertheoretical develop-
Next, we elaborate the model in Figure 7A by ment is to specify conditions under which
adding the effects of domain-specific mood on differentlinking mechanisms will occur. In this
domain role performance,based on the reason- article we explained why each individual link-
ing that domain mood can influence role perfor- ing mechanism occurs, and we noted that link-
mance not only in other domains but also in the ing mechanisms may operate simultaneously.
same domain (Staw et al., 1994). The resulting The next logical step is to explain when and
model (Figure 7B) captures relationships be- why certain linking mechanisms occur and oth-
tween mood and role performance that not only ers do not. For example, under what conditions
span domains but also occur within a domain. will the effectsofcompensation shown in Figure
To this model we add compensation by add- 7D swamp the effects of spillover so that the
ing the reallocation of time between domains overall relationship between work and family
(Figure 2A), depicting time reallocations in both mood is negative, rather than positive? Under
directions, and specifying mood as the impetus what conditions will effortsto suppress mood
for these reallocations. We furtherincorporate fail, so that mood spillover overwhelms segmen-
the notion that time devoted to a domain may tation? Such questions can be investigated by
enhance role performance in that domain (Fig- identifying person and situation factors that
ure 6A). The resulting model (Figure 7C) shows promotecertain linking mechanisms and inhibit
that mood affects role performance not only di- others-a task that awaits futureresearch.
rectly but also indirectly,through time alloca- A final area forfurthertheoreticaldevelopment
tion. For instance, a person experiencing nega- is to relax the boundary conditions we have spec-
tive work mood may performfamily roles less ified in this article. For instance, by emphasizing
effectivelybut, at the same time, may reallocate the psychological dynamics of work and family,
time fromwork to family,which would facilitate we have devoted littleattentionto social, organi-
family role performance. zational, and cultural factorsthat may influence
Finally, we add two moderator variables- work-family linkages. For example, for dual-
one that depicts intentional time allocation de- earner couples, personal decisions to reallocate
cisions that dampen the tradeoffbetween work investmentbetween work and family must take
and family time and another that incorporates into account these same decisions made by one's
segmentation as the suppression of work and spouse (Bielby & Bielby, 1989).Similarly,linkages
between workand familymay be affectednotonly Bielby, W. T., & Bielby, D. D. 1989. Family ties: Balancing
by variables in these two domains but also by commitmentsto work and family in dual earner house-
holds. American Sociological Review, 54: 776-789.
variables in the broader cultural contextthat sur-
rounds workand family(Aryee,1992;.Keller,1987). Blood, R. O., & Wolfe, D. M. 1960. Husbands and wives. New
York: Macmillan.
Moreover, work-familylinkages may operate
differentlyfor differenttypes of work and for Bolger, N., DeLongis, A., Kessler, R. C., & Wethington,E. 1989.
The contagion of stress across multiple roles. Journalof
differentfamily structures. For instance, work- Marriage and the Family, 51: 175-183.
family segmentation is perhaps easier when
Burke, R. J.,& Greenglass, E. 1987. Work and family.In C. L.
work involves extensive travel but is more diffi-
Cooper & I. T. Robertson (Eds.), International review of
cult for family-runbusinesses. Likewise, work- industrial and organizational psychology: 273-320. New
family conflict may be manageable for a mar- York: Wiley.
ried working parent with a supportive extended Burke, R. J.,& McKeen, C. A. 1993. Career prioritypatterns
family but overwhelming for a single working among managerial and professional women. Applied
parent with no extended family. Thus, differ- Psychology: An International Review, 42: 341-352.
ences in work and family arrangements should Champoux, J.E. 1978. Perceptions of work and nonwork: A
be considered when developing comprehensive reexamination of the compensatory and spillover mod-
models of the work-familyinterface. els. Sociology of Workand Occupations, 5: 402-422.
Cialdini, R. B., Trost,M. R., & Newsom, J.T. 1995. Preference
for consistency: The development of a valid measure
CONCLUSION and the discovery of surprising behavioral implications.
Journalof Personalityand Social Psychology,69: 318-328.
We have reviewed and critiqued linking Cooke, R. A., & Rousseau, D. M. 1984. Stress and strain from
mechanisms examined in work-familyresearch family roles and work role expectations. Journal of Ap-
and respecified these mechanisms as relation- plied Psychology, 69: 252-260.
ships between work and family constructs. For Cox, M. J.,& Paley, B. 1997. Families as systems. Annual
each relationship we have explained its sign Review of Psychology, 48: 243-267.
and causal structureand whether the relation- Crouter, A. C. 1984a. Participative work as an influence on
ship arises fromintentional actions by the per- human development. Journalof Applied Developmental
son. The properties of sign, causal structure,and Psychology, 5: 71-90.
intentprovide a theoretical frameworkforexam- Crouter, A. C. 1984b. Spillover from family to work: The
ining relationships between work and family neglected side of the work-family interface. Human
constructs-a frameworkthat incorporates link- Relations, 37: 425-441.
ing mechanisms examined in previous work- Deci, E. L., & Ryan, R. M. 1985. Intrinsicmotivation and self-
family research and reveals mechanisms yet to determination in human behavior. New York: Plenum.
be considered. Relationships between work and Diener, E., & Suh, E. 1997. Measuring quality of life: Eco-
family constructs embedded in this framework nomic, social, and subjective indicators. Social Indica-
tors Research, 40: 189-216.
constitute theoretical building blocks that may
be used to develop comprehensive theories of Dubin, R. 1973. Work and non-work: Institutional perspec-
tives. In M. D. Dunnette (Ed.), Workand non-workin the
the work-familyinterface and to embellish ex-
year 2001: 53-68. Monterey,CA: Brooks/Cole.
isting theories in the work-familyliterature.
Eckenrode, J., & Gore, S. 1990. Stress and coping at the
boundary of work and family.In J.Eckenrode & S. Gore
(Eds.), Stress between work and family: 1-16. New York:
REFERENCES Plenum.
Aryee, S. 1992. Antecedents and outcomes of work-family Edwards, J.R. 1992.A cybernetictheoryof stress, coping, and
conflict among married professional women: Evidence well-being in organizations. Academy of Management
fromSingapore. Human Relations, 45: 813-837. Review, 17: 238-274.
Ashforth,B. E., & Fried, Y. 1988. The mindlessness of organ- Evans, P., & Bartolome, F. 1984.The changing pictures of the
izational behaviors. Human Relations, 41: 305-329. relationship between career and family.Journal of Oc-
Ashforth,B. E., & Humphrey,R. H. 1993. Emotional labor in cupational Behavior, 5: 9-21.
service roles: The influence of identity. Academy of Evans, P., & Bartolome, F. 1986.The dynamics of work-family
Management Review, 18: 88-115. relationships in managerial lives. International Review
Barling, J.,& Macewen, K. E. 1992. Linking work experiences of Applied Psychology, 35(Special Issue): 371-395.
to facets of marital functioning.Journal of Organiza- French, J.R. P., Jr.,Caplan, R. D., & Harrison, R. V. 1982. The
tional Behavior, 13: 573-583. mechanisms of job stress and strain. London: Wiley.
Repetti, R. L. 1987. Linkages between work and family roles. Voydanoff,P. 1989.Workand family:A review and expanded
Applied Social Psychology Annual, 7: 98-127. conceptualization. In E. B. Goldsmith (Ed.), Work and
family: Theory,research, and applications: 1-22. New-
Rice, R. W., McFarlin, D. B., Hunt, R. G., & Near, J. P. 1985. bury Park, CA: Sage.
Organizational work and the perceived quality of life:
Toward a conceptual model. Academy of Management Watson, D., & Clark, L. A. 1984. Negative affectivity:The
Review, 10: 296-310. disposition to experience aversive emotional states.
Psychological Bulletin, 96: 465-490.
Rice, R. W., Near, J.P., & Hunt,R. G. 1980.The job-satisfaction/
life-satisfaction relationship: A review of empirical re- Weick, K. A. 1995. What theory is not, theorizing is. Admin-
search. Basic and Applied Social Psychology, 1: 37-64. istrative Science Quarterly, 40: 385-390.
Schneider, B. 1983. Interactional psychology and organiza- Whetten,D. A. 1989.What constitutes a theoretical contribu-
tional behavior. In B. M. Staw & L. L. Cummings (Eds.), tion? Academy of Management Review, 14: 490-495.
Research in organizational behavior, vol. 5:1-31. Green- Williams, K. J.,& Alliger, G. M. 1994. Role stressors, mood
wich, CT: JAIPress. spillover, and perceptions of work-familyconflictin em-
Schwartz, S. H., & Bilsky, W. 1987. Toward a universal psy- ployed parents. Academy of Management Journal, 37:
chological structureof human values. Journalof Person- 837-868.
ality and Social Psychology, 53: 550-562.
Zedeck, S. (Ed.). 1992. Work,families, and organizations. San
Shaffer,M. A., & Harrison,D. A. 1998.Expatriates'psychological Francisco: Jossey-Bass.