Construction and Initial Validation of a Multidimensional
Measure of WorkFamily Conflict
Dawn S. Carlson Department of Management, Hankamer School of Business, Baylor University K. Michele Kacmar Department of Management, College of Business, Florida State University and Larry J. Williams Department of Management, School of Business, Virginia Commonealth University This manuscript reports on three studies that utilized five different samples ! 5 !"!!# to construct and validate a multidimensional measure of wor$%family conflict. The si& dimensions of conflict measured include the com'ination of three forms of wor$%family conflict time( strain( and 'ehavior# and two directions of wor$%family conflict wor$ interference with family and family interference with wor$#. The three studies assessed the content ade)uacy( dimensionality( relia'ility( factor structure invariance( and construct validity of the scale. The desi*n of the final scale provides future researchers the fle&i'ility to measure any of the si& dimensions of wor$%family conflict individually. + ",,, -cademic .ress Wor$%family conflict is a source of stress that many individuals e&perience. Wor$%family conflict has 'een defined as /a form of interrole conflict in which the role pressures from the wor$ and family domains are mutually incompati'le is some respect0 1reenhaus 2 3eutell( !456( p. 77#. 8esearch on wor$%family conflict has found that this varia'le influences a num'er of outcomes includin* psycholo*ical distress( 9o' satisfaction( or*anization commitment( turnover( and life satisfaction :rone( 8ussell( 2 Cooper( !44"; <i**ins( Du&'ury( 2 =rvin*( !44"; >?Driscoll( =l*en( 2 <ildreth( !44"; .arasuraman( 1reenhaus( 8a'inowitz( The authors than$ Michael 8. :rone for his helpful comments on an earlier version of this manuscript. -ddress correspondence and reprint re)uests to Dawn S. Carlson( Department of Mana*ement( 3aylor @niversity( ..>. 3o& 45,,A( Waco( TB 7A745%5,,A. Journal of Cocational 3ehavior 56, "D4%"7A ",,,# doiE!,.!,,AF9v'e.!444.!7!G( availa'le online at httpEFFwww.ideali'rary.com on "D4 ,,,!H574!F,, IG6.,, Copyri*ht + ",,, 'y -cademic .ress -ll ri*hts of reproduction in any form reserved. 3edeian( 2 Mossholder( !454#. Thus( wor$%family conflict has 'ecome a much investi*ated topic in today?s or*anizational 'ehavior research. 8esearchers have measured wor$%family conflict in many ways. Traditionally( researchers have measured wor$%family conflict unidirectionally. That is( they studied the conflict that occurred when wor$ interfered with family 1reenhaus 2 3eutell( !456#. More recently researchers have 'e*un to reco*nize the duality of wor$%family conflict 'y considerin* 'oth directions" wor$ interference with family and family interference with wor$ e.*.( Du&'ury( <i**ins( 2 Mills( !44"; :rone et al.( !44"; 1ute$( Searle( 2 Klepa( !44!#. To fully understand the wor$%family interface( 'oth directions of wor$%family conflict W=: and :=W# must 'e considered :rone et al.( !44"; 1reenhaus 2 3eutell( !456#. 8esearchers also have 'e*un to consider the different forms of wor$%family conflict Jetemeyer( 3oles 2 McMurrian( !44A; Stephens 2 Sommer( !44G#. Consistent with 1reenhaus and 3eutell?s !456# definition( three forms of wor$% family conflict have 'een identified in the literatureE a# timeH'ased conflict( '# strainH'ased conflict( and c# #ehaviorH'ased conflict. TimeH'ased conflict may occur when time devoted to one role ma$es it difficult to participate in another role( strainH'ased conflict su**ests that strain e&perienced in one role intrudes into and interferes with participation in another role( and 'ehaviorH'ased conflict occurs when specific 'ehaviors re)uired in one role are incompati'le with 'ehavioral e&pectation in another role 1reenhaus 2 3eutell( !456#. =n !44!( 1ute$ et al. ar*ued that each of these three forms of wor$%family conflict has two directions" a# conflict due to wor$ interferin* with family W=:# and '# conflict due to family interferin* with wor$ :=W#. When these three forms and two directions are com'ined si& dimensions of wor$%family conflict resultE !# timeH'ased W=:( "# timeH'ased :=W( G# strainH'ased W=:( D# strainH'ased :=W( 6# 'ehaviorH'ased W=:( and A# 'ehaviorH'ased :=W. While there is some a*reement in terms of the forms and directions of wor$%family conflict( researchers use a wide variety of scales to measure it. 8ecently( Jetemeyer et al. !44A# constructed and validated a !,Hitem measure that included items for 'oth directions of wor$%family conflict W=: and :=W#. <owever( the authors did not consider all three of the forms of wor$%family conflict. 8e*ardin* their measure they stated it is /not as useful as scales that use a multidimensional approach to the measurement of W:C and :WC0 p. D,5#. -nother scale recently developed included items from each of the three forms of wor$%family conflict Stephens 2 Sommer( !44A#. <owever( it considers these forms from only one direction W=:#. -s a result( these authors ac$nowled*e that /further study is necessary to ade)uately measure family to wor$ conflict0 p. D56#. =n a recent metaHanalysis of wor$%family conflict the authors su**ested that differences in research results were often due to difference in measures Kosse$ 2 >ze$i( !445#. They ar*ue that researchers should strive for /*reater consistency and construct development of measures0 and that the measures needed to distin*uish more clearly 'etween nature and direction of conflict. <ence( there "6, C-8LS>J( K-CM-8( -JD W=LL=-MS still remains a need for a wor$%family conflict measure that incorporates all si& dimensions of wor$%family conflict. The purpose of this study was to develop and validate a scale that captures all si& uni)ue dimensions of wor$%family conflict. To do this( scale development procedures which are descri'ed in the psychometric literature were followed i.e.( 3a*ozzi 2 Ki( !455; Cortina( !44G; DeCillis( !44!; Schriesheim( .owers( Scandura( 1ardiner( 2 Lan$au( !44G#. =n all( three studies were conducted to develop and initially validate the final scale. >ur *oal was to produce a comprehensive( yet versatile( measure of wor$%family conflict that can 'e used to advance understandin* of this comple& phenomenon. SIX-DIMENSION! "E#S"EC$IVE OF WO#%FMI!& CONF!IC$ -s can 'e seen in :i*. !( the com'ination of the forms and directions of conflict result in si& uni)ue dimensions of wor$%family conflict. L&aminin* wor$%family conflict from this perspective raises )uestions a'out the de*ree to which the si& dimensions have 'een incorporated in prior measures and research. :or 'ac$*round purposes and as a startin* point( an investi*ation of which forms and directions of wor$%family conflict have 'een measured in past research was conducted. -n -3=F=J:>8M search of seven top 9ournals $nown to pu'lish wor$%family conflict articles $cademy of Management %ournal, Human &elations, %ournal of $pplied 'sychology, %ournal of Management, %ournal of (rgani)ational Behavior, %ournal of Vocational Behavior, and (rgani)ational Behavior and Human Decision 'rocesses# was conducted. The years included in this search were !45A%!44A. The studies and scales from articles prior to !45A are reviewed in 1reenhaus and 3eutell !456#. - total of "6 articles were located. Ta'le ! provides a listin* of the authors( the nature of wor$%family conflict studied( the source of the scales used to measure wor$%family conflict( the num'er of items in each scale( and the relia'ility coefficient for the scales as reported in the articles. Lach scale also was evaluated to determine if it distin*uished 'etween the direction of conflict( the form of conflict( and if it included FIG. 1. Dimensions of wor$%family conflict. W>8K%:-M=LK C>J:L=CT "6! T-3LL ! 8epresentation of the Si& Dimensions of Wor$%:amily Conflict in L&istin* Measures -uthor Conflict measuredM Source of scale Jum'er of items -lpha Distin*uish 'etween directionN W=:F:=W# Distin*uish 'etween formN time( strain( 'ehavior# -ll A dimensions representedN -dams( Kin*( and Kin* J-.( !44A# Time 2 strain W=: Kopelman( 1reenhaus( and Connoly !45G# D .7" Kes Jo Jo Time 2 strain :=W 3urley !454# D .AA -ryee <8( !44"# 1eneral W:C Small and 8iley( !44,# !6 .75( .77( .7G Jo Jo Jo -ryee and Lu$ JC3( !44A# 1eneral W:C Kopelman et al. !45G# D .5D Jo Jo Jo 3acharach( 3am'er*er( and Conley J>3( !44!# 1eneral W=: <olohan and 1il'ert !474# D .77( .57 Kes Jo Jo 3edeian( 3ur$e( and Moffett J>M( !455# 1eneral W:C 3ur$e( Weir( and Du Wors !45,# 5 .4" Jo Jo Jo Du&'ury and <i**ins J-.( !44!# Strain W:C 3ohen and CiverosHLon* !45!#; .lec$ !475# !A .55 M# .4, :# Jo Kes Jo :rone( 8ussell( and Cooper J-.( !44"a# Time 2 strain W=: Developed 'y authors " .7A Kes Jo Jo Time 2 strain :=W Developed 'y authors " .6A :rone( 8ussell( and Cooper J>3( !44"'# Time 2 strain W=: Developed 'y authors " .7A Kes Jo Jo Time 2 strain :=W Developed 'y authors " .6A :rone( 8ussell( and Cooper J>3( !44G# Time 2 strain W:C Developed 'y authors; see :rone et al. !44"a# D .A6 Jo Jo Jo "6" C-8LS>J( K-CM-8( -JD W=LL=-MS :rone( 8ussell( and Cooper J>M( !44D# Time 2 strain W=: Developed 'y authors " .7A Kes Jo Jo 1reenhaus( .arasuraman( 1ranrose( 8a'inowitz( and 3eutell JC3( !454# TimeH'ased W:C Kopelman et al. !45G# A .7"( .5" Jo Kes Jo StrainH'ased W:C Kopelman et al. !45G# A .7"( .7" 1ute$( Searle( and Klepa J-.( !44!# Time 2 strain W=: Kopelman et al. !45G# D .5!F.5G Kes Jo Jo Time 2 strain :=W 3urley !454# D .74F.5G <i**ins( Du&'ury( and =rvin* >3<D.( !44"# Strain W:C 3ohen and CiverosHLon* !45!#; .lec$ !475# !A .4! Jo Kes Jo Jud*e( 3oudreau( and 3retz J-.( !44D# Time 2 strain W=: 1ute$ et al. !44!#; :rone et al. !44"a# D .5" Kes Jo Jo Time 2 strain :=W 1ute$ et al. !44!# :rone et al. !44"a# D .7A Loerch( 8ussel( and 8ush JC3( !454# TimeH'ased W:C Thompson !456# 5 .5A Jo Kes Jo StrainH'ased W:C Wiley !45G# 6 .A5 3ehaviorH'ased W:C Developed 'y authors 6 .5" Matsui( >hsawa( and >n*latco JC3( !446# Time 2 strain W=: Developed 'y authors 6 .56 Kes Jo Jo Time 2 strain :=W Developed 'y authors 6 .5G Jetemeyer( 3oles( and McMurrian J-.( !44A# Time 2 strain W=: Developed 'y authors 6 .55F.54F .55 Kes Jo Jo Time 2 strain :=W Developed 'y authors 6 .5AF.5GF .54 Strain W:C Kopelman et al. !45G# D .7" M# .7" :# W>8K%:-M=LK C>J:L=CT "6G T-3LL !*Continued -uthor Conflict measuredM Source of scale Jum'er of items -lpha Distin*uish 'etween directionN W=:F:=W# Distin*uish 'etween formN time( strain( 'ehavior# -ll A dimensions representedN >?Driscoll( =l*en( and <ildreth J-.( !44"# Time W=JW Developed 'y authors 7 .57 Kes Kes Jo Time JW=W Developed 'y authors 7 .74 .arasuraman( 1reenhaus( and 1ranrose J>3( !44"# Time W:C Kopelman et al. !45G# A .7" M# +,- .F/ Jo Kes Jo .arasuraman( 1reenhaus( 8a'inowitz( 3edeian( and Mossholder -MJ( !454# 1eneral W:C 3ur$e( Weir( and Du Wors !474# 5 .4" Jo Jo Jo .arasuraman( .urhoit( 1odshal$( and 3eutell JC3( !44A# Time W=: Kopelman et al. !45G# A .57 Kes Jo Jo Strain :=W Kopelman et al. !45G# D .AD Kes Jo Jo 8ice( :rone( and Mc:arlin J>3( !44"# Strain W:C Developed 'y authors ! J- Jo Kes Jo Thomas and 1anster J-.( !446# Time 2 strain W:C Kopelman et al. !45G# 5 .57 Jo Jo Jo Wiley J>M( !457# 1eneral WJC 3ur$e( Weir( and Du Wors !474# "" .5A( .7G +01, +23 Jo Jo Jo Williams and -lli*er -MJ( !44D# 1eneral W:C Developed 'y authors Diaries J- Jo Jo Jo !ote+ -''reviationsE W=:( wor$ interference with family conflict; :=W( family interference with wor$ conflict; time( timeH'ased conflict; 'ehavior( 'ehavior'ased conflict; WJC( wor$%nonwor$ conflict 'oth directions#; W=JW( wor$ interference with nonwor$ conflict; JW=W( nonwor$ interference with wor$ conflict; strain( strainH'ased conflict; *eneral( form not specified. a W:C( wor$%family conflict 'oth directions( W=: 2 :=W#. "6D C-8LS>J( K-CM-8( -JD W=LL=-MS all si& dimensions of wor$%family conflict. The results of this investi*ation appear in Ta'le !. -s can 'e seen in Ta'le !( researchers distin*uish items 'y direction 'etween W=: and :=W in their scales in less than half of the "6 studies reviewed. Specifically( !" out of "6 researchers separated the direction of conflict in their scales. 8esearchers distin*uish 'etween the forms of conflict i.e.( time( strain( 'ehavior# in their scales even less often. =n only 7 of the "6 studies did researchers distin*uish 'etween the forms of conflict. :urther( only one of the scales e&amined included 'ehaviorH'ased conflict introduced 'y 1reenhaus and 3eutell !456#. >verall( !7 of the "6 measures do ma$e some $ind of distinction whether it is 'y form or direction in measurin* wor$%family conflict. :inally and perhaps most importantly( of all the scales e&amined( none included items that represent all si& of the dimensions of wor$%family conflict. S$'D& () EXIS$IN* WO#%FMI!& CONF!IC$ SC!ES =n Study !( e&istin* items from the literature were collected and used as the initial foundation of the scale. These items were included in a content ade)uacy analysis .art !# to determine which( if any( form or direction of wor$%family conflict they 'est represented. -dditional data collected on the retained items were analyzed via e&ploratory factor analysis .art "# to determine the underlyin* factor structure of the items. Methods*'art 4 5tem generation+ - total of G! nonredundant items were *enerated from e&istin* measures in the literature see -ppendi& -#. =tems were incorporated from 3ohen and CiverosHLon* !45!#; 3urley !454#; Du&'ury et al. !44"#; :rone et al. !44"#; 1ute$ et al. !44!#; Kopelman( 1reenhaus( and Connolly !45G#; .lec$ !475#; and Stephens and Sommer !44G#. The items developed and used 'y -ryee !44"#; 3edeian( 3ur$e( and Moffett !455#; >?Driscoll et al. !44"#; and Wiley !457# were not included 'ecause these various measures specifically considered 9o' demands or nonwor$ conflict and did not fit the scope of the present study. =n addition( Jetemeyer et al. !44A# had not 'een pu'lished when we collected the items used in Study ! so their items were not included. 'rocedure+ The G! items were included in a content ade)uacy test followin* the *uidelines provided 'y Schriesheim et al. !44G#. - respondent was as$ed to determine the de*ree to which each of the wor$%family conflict items represented a wor$%family conflict definition. The si& wor$%family conflict dimensions previously discussed :i*. !# were used. The definitions of each dimension were 'ased on the wor$ of 1reenhaus and 3eutell !456# and Du&'ury et al. !44"#. =n order to not fati*ue the raters and ris$ a reduction in the accuracy of their ratin*s( 9ud*es only rated two dimensions that were randomly assi*ned to them. This re)uired them to ma$e only A" 9ud*ements rather than !5A A 3 G!#. 'articipants+ The raters consisted of "GA under*raduates enrolled in an upper level 'usiness course at a southern university. >f the "GA( !"6 6GO# were male. W>8K%:-M=LK C>J:L=CT "66 The avera*e a*e of the sample was "!.5 years. @sin* colle*e students as content ade)uacy raters has 'een endorsed in the literature. Schriesheim et al. !44G# noted that the main re)uirement for a content ade)uacy 9ud*e is /that they possess sufficient intellectual a'ility to perform the item ratin* tas$ and that they 'e relatively free of serious potential 'ias0 p. D,7#. 1iven this re)uirement( colle*e students appear to 'e a hi*hly appropriate choice for content ade)uacy 9ud*es as they would have the capa'ility to read and understand the ratin* tas$ instructions( items( and theoretical definitions Schriesheim et al.( !44G#. $nalyses and results+ The mean score of the responses on each item provided was calculated for each dimension. =n order to 'e retained( an item?s mean had to pass two tests. :irst( an item?s hi*hest mean had to correspond to the intended wor$%family conflict dimension. =n addition( to eliminate items that did not discriminate 'etween dimensions( an item?s hi*hest mean had to 'e sufficiently different from the ratin*s o'tained for the other cate*ories. =f the difference 'etween the hi*hest and the ne&t hi*hest mean was not at least .",( the item was discarded. :our items i.e.( "( !!( !6( and G!# were dropped 'ecause they failed to score hi*hest on their intended dimension. -n additional seven items i.e.( !( 6( A( 7( 5( !G( and !4# were removed due to failure to discriminate 'etween dimensions. The ", retained items are mar$ed with an asteris$ in -ppendi& -. Methods*'art - 'rocedure+ - survey was administered to employees in a division of a state *overnment a*ency in the Southeast. The survey was comprised of the ", items retained from the content ade)uacy analyses. Lmployees rated the de*ree to which they felt that they e&perienced the conflict represented in each of the items. 8esponses were made on a Li$ertHdirection scale with the anchors 'ein* stron*ly a*ree 6# and stron*ly disa*ree !#. 'articipants+ The state *overnment a*ency sample provided G4, usa'le surveys. The sample included "GD males A,O# whose a*es avera*ed to D" years. With respect to marital status( "67 AAO# of the respondents indicated they were married and """ 67O# had children. $nalyses and results+ The responses to the items were factor analyzed with an e&ploratory factor analysis L:-# applyin* an o'li)ue rotation. Multiple criteria for determinin* the num'er of factors to retain were used :ord( MacCallum( 2 Tait( !45A; Kim 2 Mueller( !475; Stevens( !44"#. The specific criteria used wereE Kaiser?s criterion( where only factors with ei*envalues *reater than !., are retained; cumulative percenta*e of variance e&plained; and the scree plot of the factor ei*envalues. Three factors were identified. The ei*envalues for the three factors were( 6.5( ".5( and !.7 respectively. These three factors e&plained 6".GO of the variance. Lach item loaded on only one factor. -ll of the items had loadin*s *reater than .D6( e&cept for item !A( which loaded at .G,. =tem !A was the only strain 'ased "6A C-8LS>J( K-CM-8( -JD W=LL=-MS W=: item. Therefore( it would not 'e e&pected to load as stron*ly on a factor that did not distin*uish 'etween 'oth form and direction of conflict. The ", items used for this analysis did not e)ually represent each of the si& dimensions of conflict. :or e&le( the 'ehavior 'ased :=W dimension was not represented. :urther( only one item measured the strainH'ased W=: dimension( only two items measured the timeH'ased :=W dimensions( and only three items measured the strainH'ased :=W dimension. While these ", items provide a solid 'e*innin* of a comprehensive wor$%family conflict scale( additional items were needed to cover all si& dimensions. S$'D& +) '*MEN$IN* EXIS$IN* SC!ES Methods 5tem development+ The ne&t step was to develop new wor$%family conflict items to au*ment each of the si& dimensions and have them rated for content ade)uacy. The items developed were 'ased on a review of the literature as well as on personal and anecdotal e&perience. -n additional GD items( which can 'e found in -ppendi& 3( were developed so that each dimension contained a representative set of items. 'articipants+ The respondents who served as 9ud*es for the content ade)uacy analysis consisted of !G" M3- students enrolled in a 'usiness course at a western university. - total of 54 A5O# were male( the avera*e a*e was "A." years( and 7DO were employed at least partHtime. 'rocedure+ To test the content ade)uacy of the 6D items( ", retained from Study ! and GD *enerated for Study "( two different approaches were usedE cate*orization and ratin*. =n the first approach( !! randomly selected respondents used a stac$in* procedure. These individuals were *iven the items on separate sheets of paper and as$ed to stac$ the sheets on top of the definition they most closely fit. The remainin* !"! individuals also were as$ed to place each item in one dimension( 'ut the items were listed on one sheet of paper( not separate ones. =n this case( respondents placed a num'er from ! to A in front of each item to reflect which dimension definition most accurately represented each item. =n the second approach( all of the 9ud*es also were as$ed to follow the content ade)uacy *uidelines outlined 'y Schriesheim et al. !44G# i.e.( the procedure used in Study !#. Lach rater rated all 6D items on three of the si& dimension definitions selected at random. $nalyses and results+ :or the cate*orization portion of the data( the num'er of 9ud*es who placed an item in a dimension was counted. :or the ratin* portion of the data( the mean for each item on each dimension was calculated. =n order for an item to 'e retained( it had to pass 'oth a cate*orization and a ratin* content ade)uacy test. To pass the cate*orization test( an item had to 'e assi*ned to the correct definition at least 7,O of the time. This test was performed on data from the cate*orization techni)ues. :or the ratin*Hcontent ade)uacy testin*( a mean score of G.6 or hi*her 7,O# for an item on the correct definition was considered W>8K%:-M=LK C>J:L=CT "67 accepta'le. The 7,O cutHoff is consistent with the criterion used in previous content ade)uacy research i.e.( Schriesheim 2 <in$in( !44,#. -pplyin* these rules to the data indicated that "! items did not pass 'oth tests( leavin* GG items. -ll GG of these items could have 'een included in the final scale( 'ut for parsimony and e)ual representation across dimensions( three additional items were removed leavin* only the 6 'est items for each dimension. The G, items included in the final scale are shown with an asteris$ in -ppendi& 3. S$'D& ,) SC!E V!ID$ION Study G was desi*ned to validate the scale developed in Studies ! and ". .art ! of this validation effort included further measure purification analyses. .art " used a second sample to e&amine the dimensionality( relia'ility( and discriminant validity of the scale. -lso in .art "( the factor structure from .art ! was applied to various samples and tested on a sample split on *ender. :inally( differential relationships were e&amined. To *ather the data needed to perform these tests( a survey composed only of the G, wor$%family conflict scale items retained in Study " was administered. Met-ods."art ( 'articipants 4 The participants consisted of ""5 *raduates from an L&ecutive M3- pro*ram at a lar*e western university. -ppro&imately G5, surveys were distri'uted to individuals from a mailin* list of past *raduates response rate A,O#. - cover letter was included *uaranteein* confidentiality and e&plainin* the purpose of the survey. 8espondents were supplied reply envelopes and as$ed to return the survey to the researchers throu*h the mail. The participants included !6! AAO# males( were an avera*e a*e of D, years old( and had an avera*e or*anizational tenure of 7.4 years. With respect to marital status( !7, 76O# of the respondents indicated they were married and !G7 A,O# had children. #esults Measure 'urification Structural e)uation modelin* SLM# was applied to the G, item measure from Study " to isolate items that performed well across a num'er of different criteria. - si&Hfactor confirmatory model with five items reflectin* each of the si& factors esta'lished in Study " was specified usin* L=S8LL 5 JoPres$o* 2 SoPr'om( !44G#. To determine which items should 'e removed, we applied su**estions found in the scale development literature 3a*ozzi 2 Ki( !455; DeCillis( !44!#. :irst( we deleted any items that had completely standardized factor loadin*s of less than .6,. Je&t( we inspected the modification indices and e&pected chan*e values for all the factor loadin*s to ensure that an item was not more stron*ly associated with any factor other than the one for which it was intended. =f it was( it was eliminated. :inally( we removed items that consistently resulted in correH "65 C-8LS>J( K-CM-8( -JD W=LL=-MS lated measurement error either within factors( across factors( or 'oth. That is( items were dropped if consistently si*nificant standardized residuals were found. -pplyin* these criteria resulted in the removal of !! of the G, itemsE 6 items due to correlated measurement error( " items due to factor loadin* issues( " items due to values for modification and e&pected chan*e parameters( and " items which were pro'lematic on multiple criteria. >ne final item was removed from the scale due to the redundancy of its wordin*. The purification process produced an !5Hitem scale with G items measurin* each of the A dimensions. >f the remainin* !5 items( 6 were from e&istin* scales and !G items were new. The final items appear in Ta'le ". Met-ods."art + .art " of Study G was desi*ned to assess dimensionality( relia'ility( and discriminant validity of the scale and to determine if the factor structure of the scale held for a new sample and across *ender. :urthermore( several antecedents and conse)uences of wor$%family conflict were collected for construct validation of the new !5 item measure. <ence( in .art "( not only were responses collected for the wor$%family conflict items( 'ut several antecedents and conse)uences of wor$%family conflict also were included in the survey. The antecedents included were role conflict( role am'i*uity( and social support from 'oth the wor$ and family domain as well as wor$ involvement. The outcomes studied were 9o' satisfaction( family satisfaction( life satisfaction( and or*anizational commitment. -ll of these varia'les have 'een found to 'e si*nificantly related to wor$%family conflict. =t was e&pected that the antecedents of role overload and role am'i*uity( and involvement from each domain( would 'e positively related to the respective domains of wor$%family conflict e.*.( -dams( Kin*( 2 Kin*( !44A; :rone( Kardley( 2 Mar$el( !447#( while the antecedent of social support from each domain would 'e ne*atively related to domain specific wor$%family conflict 1reenhaus( 3edeian( 2 Mossholder( !457; Schau'roec$( Cotton( 2 Jennin*s( !454#. The three satisfaction outcomes i.e.( 9o'( family( life# and or*anizational commitment were e&pected to decrease as wor$%family conflict increases( so a ne*ative relationship is predicted <i**ins et al.( !44"( .arasuraman et al.( !454; >?Driscoll et al.( !44"; 8ice( :rone( 2 Mc:arlin( !44"#. Data collected from the second survey .art "# were used to perform differential prediction analyses. 'articipants - Data were collected from ""6 individuals who were employed fullHtime. The respondents were employed 'y numerous or*anizations in a midwestern city and secured throu*h a snow'all samplin* approach. The principal sample included individuals who were enrolled as fullHtime students in an evenin* pro*ram caterin* to wor$in* adults finishin* their under*raduate de*rees. 3esides completin* the survey themselves( these individuals were as$ed to distri'ute five surveys to collea*ues at their places of employment who would 'e willin* to complete a )uestionnaire e&aminin* wor$%family conflict. The only selection W>8K%:-M=LK C>J:L=CT "64 criterion applied was that respondents hold fullHtime 9o's. The sample consisted of 5G G7O# males who were an avera*e a*e of G6.6 years old. - total of !DD ADO# were married and !D" AGO# had children livin* at home. The results were T-3LL " :inal Cersion of Wor$%:amily Conflict Scale Wor$%family conflict items TimeH'ased wor$ interference with family !. My wor$ $eeps me from my family activities more than = would li$e.a ". The time = must devote to my 9o' $eeps me from participatin* e)ually in household responsi'ilities and activities.a G. = have to miss family activities due to the amount of time = must spend on wor$ responsi'ilities. TimeH'ased family interference with wor$ D. The time = spend on family responsi'ilities often interfere with my wor$ responsi'ilities. 6. The time = spend with my family often causes me not to spend time in activities at wor$ that could 'e helpful to my career. A. = have to miss wor$ activities due to the amount of time = must spend on family responsi'ilities. StrainH'ased wor$ interference with family 7. When = *et home from wor$ = am often too frazzled to participate in family activitiesF responsi'ilities. 5. = am often so emotionally drained when = *et home from wor$ that it prevents me from contri'utin* to my family. 4. Due to all the pressures at wor$( sometimes when = come home = am too stressed to do the thin*s = en9oy. StrainH'ased family interference with wor$ !,. Due to stress at home( = am often preoccupied with family matters at wor$. !!. 3ecause = am often stressed from family responsi'ilities( = have a hard time concentratin* on my wor$. !". Tension and an&iety from my family life often wea$ens my a'ility to do my 9o'. 3ehaviorH'ased wor$ interference with family !G. The pro'lemHsolvin* 'ehaviors = use in my 9o' are not effective in resolvin* pro'lems at home.a !D. 3ehavior that is effective and necessary for me at wor$ would 'e counterproductive at home.a !6. The 'ehaviors = perform that ma$e me effective at wor$ do not help me to 'e a 'etter parent and spouse.a 3ehaviorH'ased family interference with wor$ !A. The 'ehaviors that wor$ for me at home do not seem to 'e effective at wor$. !7. 3ehavior that is effective and necessary for me at home would 'e counterproductive at wor$. !5. The pro'lemHsolvin* 'ehavior that wor$ for me at home does not seem to 'e as useful at wor$. a =tems from Stephens and Sommer !44A#. "A, C-8LS>J( K-CM-8( -JD W=LL=-MS e&amined to determine if the sample was confounded 'y includin* a small *roup of individuals who were not married and had no children livin* at home. The comparisons for the model 'ased on the full and more constrained sample su**ested no differences 'etween samples. Thus( the results for the full sample are reported herein. Measures &ole conflict+ Wor$Hrelated role conflict was measured usin* 8izzo( <ouse( and Lirtzman?s !47,# ei*htHitem measure of role conflict. - sample item is /= must do thin*s that should 'e done differently.0 The internal relia'ility was .4, for the participants in this study. The same ei*ht items were used to measure familyHrelated role conflict. <owever( each item was modified to reflect the family domain. The Cron'ach alpha was .56 for the participants in this study. &ole am#iguity+ Wor$ role am'i*uity was measured with 8izzo et al.?s !47,# role am'i*uity scale. This scale consists of si& items and produced a Cron'ach alpha of .5". - sample item is /= $now e&actly what my responsi'ilities are.0 The same items( ad9usted for the family domain( were used to measure family role am'i*uity. The internal consistency relia'ility estimate for these si& items was .5G. Social support+ Social support from the wor$ domain was measured with a !A item measure of or*anizational support developed 'y Lisen'er*er( <untin*ton( <utchison( and Sowa !45A#. - representative item is /<elp is availa'le from the or*anization when = have a pro'lem.0 The alpha coefficient was .4D for the participants in this study. :or the family domain these items were adapted to tap the support received from family sources. The alpha coefficient for this scale was .4G for the participants in this study. 5nvolvement+ Two )uestions e.*.( /= would li$e more time to spend wor$in*0# ori*inally from Quinn and Staines !474# and used 'y <i**ins et al. !44"# were used to tap wor$ involvement. =n addition( two )uestions from 3uchanan !47D# e.*.( /= am very much personally involved in my wor$0# were included that were desi*ned to measure a'sorption in the activities of one?s role. The alpha coefficient for this scale was .5D. These items were modified to measure the familyrelated domain as well. The Cron'ach alpha coefficient produced 'y the family scale was .5". %o# satisfaction+ The 9o' satisfaction scale was an overall measure of the de*ree to which an individual is satisfied or happy with his or her 9o'. >ur threeHitem measure of 9o' satisfaction was desi*ned and used 'y Cammann( :ichman( Jen$ins( and Klesh !474# and Seashore( Lawler( Mirvis( and Cammann !45"#. >ne of the items from this scale is /-ll in all( = am satisfied with my 9o'.0 The Cron'ach alpha for this scale was .4!. (rgani)ational commitment+ The or*anizational commitment scale measures the de*ree to which individuals are committed to the or*anization. The nine items used were developed 'y 3alfour and Wechsler !44A#. - sample item is /= am W>8K%:-M=LK C>J:L=CT "A! )uite proud to 'e a'le to tell people who it is = wor$ for.0 The relia'ility for this scale was .4!. Family satisfaction+ The family satisfaction scale is an overall measure of the de*ree to which an individual is satisfied with his or her family life. The threeHitem scale was developed 'y Staines and .lec$ !45G#. - sample item is /= am happy with my family life.0 The internal relia'ility for this scale was .56. 6ife satisfaction+ The life satisfaction scale measures an individual?s perceptions re*ardin* the )uality of his or her life in *eneral. The fiveHitem scale developed 'y Diener( Lmmons( Larsen( and 1riffin !456# was used. -n item from this scale is /= am satisfied with my life.0 The Cron'ach alpha estimate for this scale was .57. $nalyses The dimensionality of the items was assessed with confirmatory factor analysis. Je&t( the relia'ility of the scales was esta'lished with coefficient alpha. Discriminant validity of the scales was e&amined with SLM. :urther( a multiple *roup SLM test was conducted to determine if the si&Hfactor structure held across samples. The .articipants ! and .articipants " data from Study G were used for this analysis. =n addition( a multiple *roup SLM test was conducted on .articipants " to determine if the si& factor structure held across *ender. Differential predictions were investi*ated throu*h e&aminin* path coefficients in structural e)uation models usin* the measures developed to tap the wor$%family conflict dimensions with antecedents and outcomes. This analysis was 'ased on .articipants ". #esults Dimensionality Confirmatory factor analysis was used to assess a si&Hfactor model where each of the si& cate*ories were represented separately -nderson 2 1er'in*( !455#. :or comparison purposes( three other possi'le models similar to models used in prior scales were e&amined. :irst( a threeHfactor model( which represented the three forms of wor$%family conflict( time( strain( and 'ehavior collapsin* across direction#( was tested. Je&t( a twoHfactor model representin* the two directions of wor$%family conflict( W=: and :=W collapsin* across form#( was estimated. :inally( a oneHfactor model representin* a *eneral wor$%family conflict perspective was e&amined. =n each model the items were forced to load on a specified factor and the factors were allowed to correlate. Ta'le G presents the 7"( comparative fit statistic C:=#( and rootHmeanHs)uare error of appro&imation for each of the four models. The indices show that the si&Hfactor model is the 'est fittin* model. :urther e&amination of the si&Hfactor model indicated that the factor loadin*s were all si*nificant. The completely standardized factor loadin*s for each of the !5 items appear in :i*. ". "A" C-8LS>J( K-CM-8( -JD W=LL=-MS 5nternal Consistency The internal consistency of each of the si& dimensions was estimated with coefficient alpha. The relia'ilities e&ceeded the conventional level of acceptance of .7, Junnally( !475#E timeH'ased W=: 5 .57; timeH'ased :=W 5 .74; strain'ased W=: 5 .56; strainH'ased :=W 5 .57; 'ehaviorH'ased W=: 5 .75; 'ehavior'ased :=W 5 .56. Discriminant Validity Discriminant validity was assessed 'y e&aminin* the factor correlations from the confirmatory factor analysis. The correlations of the si& factors( found in Ta'le D( ran*ed from ."D to .5G. >nly two of the correlations were a'ove .A,. Thus( discriminant validity was shown. Factor Structure 8ests To determine if the factor structure of the si&Hdimensional model was invariant across various samples( a L=S8LL twoH*roup measurement procedure was performed. This procedure was used 'ecause it allows the factor loadin*s( correlations( and error variances to 'e held invariant individually or in com'ination. Tests of this nature provide a ri*orous assessment of the measurement properties of the models 3a*ozzi 2 Ki( !455; 3ollen( !454; Marsh( !446#. :our twoH*roup models for the si&Hdimensional wor$%family conflict approach were estimated for comparison purposes. The first model re)uired the factor loadin*s( factor correlations( and the error variances for 'oth data sets to 'e e)uivalent. The second model still held the factor loadin*s and correlations invariant( 'ut allowed the error variances to 'e different for each dataset. The T-3LL G Lstimates of :it =ndicesRSample " Model 7" df p Comparative fit inde& 8oot mean s)uare error of appro&imation Si&Hdimensional modelE @ni)ue cate*ories of wor$%family conflict "G7.D, !", .,, .46 .,A ThreeHdimensional modelE :orms of wor$%family conflict !!AA.!D !G" .,, .AA .!4 TwoHdimensional modelE Directions of wor$% family conflict !G"A.44 !GD .,, .A! .!4 >neHdimensional modelE 1eneral wor$%family conflict !A77.G4 !G6 .,, .6, ."G !ote+ ! 5 ""6. W>8K%:-M=LK C>J:L=CT "AG ne&t model allowed the factor correlations and error variances to vary( 'ut the factor loadin*s remained invariant. The final model allowed the factor loadin*s( correlations( and error variances to vary across the samples. The fit for each of the four models as well as the 7" difference tests 'etween the 'aseline model and each of the other models can 'e found in Ta'le 6. These results su**est that the two different data sets map well to the model with respect to the factor loadin*s( factor correlations( and error variances( indicatin* that the model is *eneraliza'le across the data sets. The 'aseline model was not si*nificantly different from the model with the factor loadin*s held FIG. 2. Completely standardized path loadin*s for !5Hitem scale. "AD C-8LS>J( K-CM-8( -JD W=LL=-MS invariant or from the model with the factor loadin*s and factor correlations held invariant. The only instance where the factor structure did not hold across the samples was the most constrained model( where factor loadin*s( factor correlations( and error variances were all invariant. <owever( invariant error variances are considered the least important in testin* measurement property invariance across *roups 3ollen( !454; Jetemeyer et al.( !44A#. :urthermore( statistical tests of invariance have limitations so fit indices also should 'e used to assess invariance Marsh( !446; Williams( 3ozdo*an( 2 -imanHSmith( !44A#. -n e&amination of the fit statistics for the model in which factor loadin*s( correlations( and error variances were fi&ed indicated ade)uate fit on all indices. Thus( evidence of measurement invariance across samples was found( further confirmin* the structure of the si&Hfactor model. 9ender Differences To determine if the factor structure of the si&Hdimensional model was invariant across *ender( the same analytic procedure used to e&amine the factor structure in the T-3LL 6 Test of Measurement =nvariance Si&Hdimensional model 7" df 7diff " dfdiff Comparative fit inde& 8ootHmeanHs)uare estimate of appro&imation Jo constraints 'aseline model# D6!.",M "DA .4A .,G4 :actor loadin*s invariant D7".75M "AD "!.65 !5 .4A .,G4 :actor loadin*s 2 factor correlations invariant D57.6"M "7G GA.G" "7 .4A .,G4 :actor loadin*s( factor correlations( 2 error variances invariant 674.4,M "4! !!4.7,M D6 .4D .,DG M p , .,!. T-3LL D Discriminant Calidity of the Si& Dimensions of Wor$%:amily ConflictE .hi Matri& from C:- -nalysis Dimension of wor$%family conflict ! " G D 6 A !. TimeH'ased wor$ interference with family R ". TimeH'ased family interference with wor$ .G! R G. StrainH'ased wor$ interference with family .65 .D6 R D. StrainH'ased family interference with wor$ ."D .7A .D5 R 6. 3ehaviorH'ased wor$ interference with family .G! .D, .6D .D7 R A. 3ehaviorH'ased family interference with wor$ ."5 ."A .6! .D6 .5G R W>8K%:-M=LK C>J:L=CT "A6 previous section was applied. - L=S8LL twoH*roup measurement procedure was performed in which four twoH*roup models i.e.( male versus female# for the si& dimensional wor$%family conflict approach were estimated for comparison purposes. The fit for each of the four models as well as the 7" difference tests 'etween the 'aseline model and each of the other models appear in Ta'le A. These results su**est that the two different data sets map well to the model with respect to the factor loadin*s. The 'aseline model was not si*nificantly different from the model when the factor loadin*s were held invariant. <owever( there were differences across *ender when the factor loadin*s and factor correlations were held invariant and in the most constrained model in which the factor loadin*s( factor correlations( and error variances were all invariant. While it is not surprisin* to find differences in error variance the differences in factor correlations su**est that women and men may e&perience conflict differently. L&amination of the factor correlations su**est that men and women had the same pattern of si*nificance. The avera*e overall correlation for males was .D7 and for females was .D6. :urthermore( twoHthirds of the individual differences were less than .", and the lar*est difference 'etween correlations was .G7. T-3LL A Test of 1ender Differences 7" df 7diff " dfdiff Comparitive fit inde& 8ootHmeanHs)uare estimate of appro&imation Si&Hdimensional model Jo constraints 'aseline model# D"7.AGM "DA .4" .,7 :actor loadin*s invariant DA,.64M "AD G".4A !5 .4" .,5 :actor loadin*s 2 factor correlations invariant D57."7M "7G 64.5DM "7 .4! .,5 :actor loadin*s( factor correlations 2 error variances invariant 6!5.74M "4! 4!.!AM D6 .4, .,5 t test for 1ender Differences Dimension Mean for males Mean for females t p TimeH'ased wor$ interference with family ".4! ".5" .6" .A,! TimeH'ased family interference with wor$ !.77 ".,! 2".,6 .,D" StrainH'ased wor$ interference with family ".D6 ".5! 2".6" .,!G StrainH'ased family interference with wor$ !.7! !.4G 2".," .,D6 3ehaviorH'ased wor$ interference with family ".DG ".AG 2!.65 .!!A 3ehaviorH'ased family interference with wor$ ".GA ".A6 2".,4 .,G5 M p , .,!. "AA C-8LS>J( K-CM-8( -JD W=LL=-MS To further e&amine *ender differences( t tests were conducted on the level of e&perienced conflict across all si& dimensions. >n four of the si& dimensions of conflict si*nificant differences were found. More specifically( females were found to e&perience more conflict than men in terms of all three family interference with wor$ forms of conflict time( strain( 'ehavior# as well as strain 'ased wor$ interference with family conflict. =t is possi'le that the inconsistent findin*s in past research on *ender differences La*le( Miles( 2 =ceno*le( !447( :rone et al.( !44"( .lec$( !477( Williams 2 -lli*er( !44D# may 'e e&plained 'y the fact that females are li$ely to e&perience more conflict than men on only some( not all( forms of conflict. Thus( the way in which conflict was measured may e&plain whether *ender differences were found. Differential &elationships Differential relationships 'etween the dimensions of wor$%family conflict and several antecedent and outcome measures were e&amined. To e&amine the differences in the wor$%family conflict dimensions two models were tested( one for each direction( which included relevant antecedents and conse)uences. The first model included the three forms of W=: conflict. =n addition( the antecedents of wor$Hrole conflict( wor$Hrole am'i*uity( wor$ involvement( and wor$ social support were included since they all represent the wor$ domain. This model is consistent with past research that showed domainHspecific antecedents were related to different directions of wor$%family conflict i.e.( -dams et al.( !44A; :rone et al.( !447; Thomas 2 1anster( !446#. The four outcomes 9o' satisfaction( family satisfaction( life satisfaction( and or*anizational commitment# also were included in the model. :inally( consistent with past research( direct paths from the antecedents to outcomes were included. The second model e&amined was similar to the first e&cept that it included the three :=W forms of conflict and family specific antecedents and conse)uences. The model approach descri'ed a'ove was chosen 'ecause of its advanta*es relative to a more traditional correlational analysis e.*.( accounts for measurement error( omni'us statistical test#. To determine if the dimensions of conflict were differentially related to the antecedents and outcomes considered here( the si*nificance of the path coefficients from the model were e&amined. These path coefficients appear in Ta'le 7. The three forms of W=: conflict have differential relationships such that three of the four antecedents role conflict( am'i*uity( and involvement# were si*nificantly related to strain 'ased conflict( two am'i*uity and involvement# were related to 'ehavior 'ased( and only one involvement# to time 'ased. The three forms of W=: conflict also differentially predicted the three types of satisfaction and commitment. More specifically( two of the forms of conflict strain and 'ehavior# were si*nificantly related to the outcomes of family and life satisfaction. <owever( timeH'ased conflict was not si*nificantly related to any of the outcomes of interest. Similar findin*s of differential relationships were found for the :=W varia'les. -ll four of the family domain antecedents si*nificantly predicted 'ehaviorH'ased W>8K%:-M=LK C>J:L=CT "A7 conflict 'ut only two role conflict and social support# predicted timeH and strainH'ased conflict. :urthermore( while family role conflict had similar relations to all three forms of conflict( social support was more hi*hly related to time and strain conflict than 'ehavior conflict. The strainH'ased form of conflict si*nificantly predicted three of the four outcome varia'les not predicted 'y the other two forms of conflict. =n addition( or*anizational commitment was si*nificantly related to the 'ehaviorH'ased form of conflict 'ut not the other two forms of :=W. These findin*s would su**est that the si& dimensions of wor$%family conflict are differentially related to various antecedents and outcomes commonly found in the wor$%family conflict literature. T-3LL 7 Completely Standardized .ath Loadin*s Measure TimeH'ased wor$ interference with family StrainH'ased wor$ interference with family 3ehaviorH'ased wor$ interference with family Three forms of wor$ interference with family conflict -ntecedents Wor$ role conflict 1# .!! ."4M ."! Wor$ role am'i*uity 1# .!7 ."DM .""M Wor$ social support 2# .,, 2.,G 2.,4 Wor$ involvement 1# .G7M .G7M ."!M >utcomesE Jo' satisfaction 2# .,D 2.,G .,, :amily satisfaction 2# .,7 2."6M 2.G4M Life satisfaction 2# .!G 2."DM 2.GAM >r*anizational commitment 2# .,D .,G 2.,A Measure TimeH'ased family interference with wor$ StrainH'ased family interference with wor$ 3ehaviorH'ased family interference with wor$ Three forms of family interference with wor$ conflict -ntecedents :amily role conflict 1# ."6M ."7M ."7M :amily role am'i*uity 1# 2.,4 .," .",M :amily social support 2# 2.G5M 2.G6M 2."GM :amily involvement 1# .,, 2.," .!"M >utcomesE Jo' satisfaction 2# 2.,G 2."DM 2.!D :amily satisfaction 2# .," 2.""M .!! Life satisfaction 2# .,7 2."GM .,4 >r*anizational commitment 2# 2.!6 2.!! 2.""M !ote+ ! 5 ""6. M p , .,6. "A5 C-8LS>J( K-CM-8( -JD W=LL=-MS DISC'SSION The present research constructed and initially validated a comprehensive scale of wor$%family conflict that incorporated the multiple dimensions of the construct. The items composin* the scale are a com'ination of items from previous wor$ and new items developed specifically for this study. Content ade)uacy( content analysis( e&ploratory and confirmatory factor analyses( and correlation analyses were performed on these items. The end result was an !5Hitem scale with si& different su'scales that measured the si& dimensions of wor$%family conflictE timeH'ased W=:( timeH'ased :=W( strainH'ased W=:( strainH'ased :=W( 'ehaviorH'ased W=:( and 'ehaviorH'ased :=W. Lach of the scales in the si&dimensional model showed discriminant validity( internal consistency( and invariance of the factor structure across samples. =n addition( each of the scales differentially related to various antecedents and conse)uences of wor$%family conflict( further su**estin* the potential predictive validity of the scales. >ther scales e&ist that measure wor$%family conflict i.e.( :rone et al.( !44"; 1ute$ et al.( !44!#( and some have even 'een su'9ected to su'stantial validation efforts Jetemeyer et al.( !44A; Stephens 2 Sommer( !44A#. <owever( none of the e&istin* scales provide a way to measure each of the si& dimensions of conflict. =n fact( Jetemeyer et al. !44A# stated that their scale was /not as useful as scales that use a multidimensional approach0 p. D,5# to measure wor$%family conflict. Stephens and Sommer !44A#( whose measure consisted of W=: items( ac$nowled*ed that /further study is necessary to ade)uately measure family to wor$ conflict0 p. D56#. The scale developed in the present study overcomes 'oth limitations of previous scale development efforts and answers the call for a measure that considers the importance of 'oth nature and direction of conflict Kosse$ 2 >ze$i( !445#. The multidimensional measure of the concept of wor$%family conflict developed in the present study is a more accurate depiction of the construct as it allows each of the si& dimensions to 'e e&amined. :uture use of this scale should provide a *reater understandin* re*ardin* how the separate wor$%family conflict dimensions relate to attitudes and 'ehaviors of interest. Strengths, 6imitations, and Future &esearch The research performed to construct and validate this scale has several stren*ths. :irst( the research consisted of three different studies that to*ether provide a very thorou*h scaleHdevelopment effort. <ence( the resultin* scale has 'een su'9ected to ri*orous development and validation procedures. -nother stren*th of this research is that it incorporated five different samples. Thus( the potential for sample specific 'ias has 'een reduced 'y usin* uni)ue and independent samples for each phase of the pro9ect. @sin* multiple samples also allowed us to e&amine the invariance of the final scale across samples. :urthermore( the new scale includes each of the si& dimension of wor$%family conflict( some of which have 'een missin* in previous measures. :inally( the scale measures all of the dimensions of wor$%family conflict usin* only !5 items. W>8K%:-M=LK C>J:L=CT "A4 <owever( the study is not without limitations. :irst( the scale was validated on only two samples. -dditional validation of the scale across or*anizations and occupations is needed to further esta'lish the scale and provide *eneraliza'ility. Second( we did not incorporate all of the items from the Jetemeyer et al. !44A# scale( as this scale was not pu'lished while the current research was underway. :uture research should include the Jetemeyer scale and the one developed here in one study to determine the de*ree of difference or overlap 'etween them. :inally( only ei*ht constructs were used to e&amine the differential relations of the wor$%family conflict scales. :uture research should incorporate additional constructs thou*ht to 'e uni)uely related to different dimensions of wor$%family conflict. While the current research included traditional varia'les found in the wor$%family conflict literature( it would 'e useful to e&amine differential predictions with additional antecedents and conse)uences. While each of these limitations provides an opportunity for future research( there is also the need for more research on 'ehaviorH'ased conflict. More research is needed to clarify the meanin* of 'ehaviorH'ased conflict and su'se)uently its measurement. =t has 'een historically considered as the reco*nition that different 'ehaviors are necessary at wor$ and at home( which in and of itself does not reflect conflict. <owever( the ina'ility of the individual to ad9ust that 'ehavior from one role to the other more clearly represents the construct. =n fact( in the current study the results from the C:- Ta'le D# su**est the dimensions of 'ehaviorH'ased conflict are hi*hly correlated .5G#. This correlation( however( could 'e inflated due to the restrictive assumptions of confirmatory factor analysis that all secondary factor loadin*s are zeros. Thus( an e&ploratory factor analysis was conducted and the factor correlation 'etween the two 'ehavior factors was si*nificantly lower .D"#. =n addition( the factor loadin*s demonstrated an appropriate simple structure. Thus( while the e&istin* factors do discriminate( further research also may 'e needed to provide additional conceptual distinction. :inally( further research is needed on the uni)ue antecedents and outcomes for each of the dimensions of wor$%family conflict measured 'y this scale. While a *reat deal is $nown a'out wor$%family conflict in *eneral( very little is $nown a'out the stren*th of the relationships of the si& dimensions of wor$%family conflict with other varia'les. Does each uni)uely e&plain different outcomesN Does each have uni)ue predictorsN :u rthermore( different )uestions need to 'e as$ed a'out the directions of wor$%family conflict. Most research su**ests that W=: conflict is *reater than :=W conflict 1ute$ et al.( !44!; Jud*e( 3oudreau( 2 3retz( !44D; Jetemeyer et al.( !44A#. <owever( little is $now a'out when the forms of wor$%family conflict are com'ined with the directions. That is( are all forms of conflict time( strain( 'ehavior# *reater from the W=: direction than from the :=W directionN -ll of these )uestions and more 'e* to 'e answered. =t is hoped that when researchers set out to e&plore these issues in the future( the scale developed and validated in the present study will 'e employed to measure the comple& nature of wor$%family conflict. "7, C-8LS>J( K-CM-8( -JD W=LL=-MS ""ENDIX Initial ,( Items from E/istin0 Scales =tem Source !. -fter wor$( = come home too tired to do some of the thin*s =?d li$e to do. 1ute$ et al. !44!#; Stephens and Sommer !44G# ". = feel = have more to do than = can comforta'ly handle. Du&'ury et al. !44"# G. My wor$ $eeps me from my family activities more than = would li$e.a Du&'ury et al. !44"#; Stephens and Sommer !44G( !44A# D. >n the 9o' = have so much wor$ to do that it ta$es away from my personal interests.a 1ute$ et al. !44!# 6. = feel physically drained when = *et home from wor$. Du&'ury et al. !44"#; Stephens and Sommer !44G# A. The tensions and an&ieties = feel from my family and wor$ responsi'ilities often 'ecome so *reat that my efforts to cope suffer. Stephens and Sommer !44G# 7. My familyFfriends disli$e how often = am preoccupied with my wor$ while = am at home. 1ute$ et al. !44!#; Du&'ury et al. !44"#; Stephens and Sommer !44G# 5. = feel emotionally drained when = *et home from wor$. Du&'ury et al. !44"# 4. The demands of my 9o' ma$e it difficult for me to maintain the $ind of relationship with my spouse and children that = would li$e.a Du&'ury et al. !44"#; Stephens and Sommer !44G( !44A# !,. My wor$ ta$es up time that =?d li$e to spend with familyFfriends.a :rone et al. !44"a#; 1ute$ et al. !44!#; Stephens and Sommer !44G( !44A# !!. = feel = have to rush to *et everythin* done each day. Du&'ury et al. !44"# !". My wor$ often interferes with my family responsi'ilities.a :rone et al. !44"a# !G. 3ecause my wor$ is so demandin*( at times = am irrita'le at home. Du&'ury et al. !44"#; Stephens and Sommer !44A# !D. =?m often too tired at wor$ 'ecause of the thin*s = have to do at home.a 1ute$ et al. !44!# !6. = feel = don?t have enou*h time for myself. Du&'ury et al. !44"# !A. =t is difficult for me to rela& when = am away from my wor$.a Stephens and Sommer !44G# !7. My personal demands are so *reat that it ta$es away from my wor$.a 1ute$ et al. !44!#; Du&'ury et al. !44"#; Stephens and Sommer !44G# !5. = often 'rin* wor$ home to do on the evenin*s and wee$ends.a Stephens and Sommer !44G# !4. = *enerally do not seem to have enou*h time to fulfill my potential 'oth in my career and as a spouse or parent. Stephens and Sommer !44G( !44A# ",. My superiors and peers disli$e how often = am preoccupied with my personal life while at wor$.a 1ute$ et al. !44!#; Du&'ury et al. !44"# W>8K%:-M=LK C>J:L=CT "7! ""ENDIX 1 #e2ised Items for Second #ound of Content de3uacy TimeH'ased wor$ interference with family !, items# L&istin* =tems G. My wor$ $eeps me from my family activities more than = would li$e.a D. >n the 9o' = have so much wor$ to do that it ta$es away from my personal interests. 4. The demands of my 9o' ma$e it difficult for me to maintain the $ind of relationship with my spouse and children that = would li$e. !,. My wor$ ta$es up time that =?d li$e to spend with familyFfriends.a !". My wor$ often interferes with my family responsi'ilities. !5. = often 'rin* wor$ home to do on the evenin*s and wee$ends. "". The time = must devote to my 9o' $eeps me from participatin* e)ually in household responsi'ilities and activities.a Jew =tems !. = feel = don?t have enou*h time to fulfill my responsi'ilities at home due to time = have to spend on my career.a ". = feel *uilty for spendin* too much time at wor$ and not enou*h time with my family. TimeH'ased wor$ interference with family !, items# -..LJD=B -*Continued =tem Source "!. My personal life ta$es up time that =?d li$e to spend at wor$.a :rone et al. !44"a#; 1ute$ et al. !44!#; Stephens and Sommer !44G# "". The time = must devote to my 9o' $eeps me from participatin* e)ually in household responsi'ilities and activities.a Stephens and Sommer !44G( !44A# "G. My family life often interferes with my responsi'ilities at wor$.a :rone et al. !44"a# "D. = am not a'le to act the same way at home as = do at wor$.a Stephens and Sommer !44G( !44A# "6. The pro'lemHsolvin* approaches = use in my 9o' are not effective in resolvin* pro'lems at home.a Stephens and Sommer !44G( !44A# "A. = act differently in respondin* to interpersonal pro'lems at wor$ than = do at home.a Stephens and Sommer !44G( !44A# "7. 3ehavior that is effective and necessary for me at wor$ would 'e counterproductive at home.a Stephens and Sommer !44G( !44A# "5. The thin*s = do that ma$e me effective at wor$ do not help me to 'e a 'etter parent and spouse.a Stephens and Sommer !44G( !44A# "4. What wor$s for me at home does not seem to 'e effective at wor$ as well( and vice versa.a Stephens and Sommer !44G( !44A# G,. =n order for me to succeed at wor$( = must 'e a different person than = can 'e at home.a Stephens and Sommer !44G# G!. = often feel the strain of attemptin* to 'alance my responsi'ilities at wor$ and home. Stephens and Sommer !44G( !44A# a =tems retained for Study ". "7" C-8LS>J( K-CM-8( -JD W=LL=-MS -..LJD=B 3*Continued G. = have to miss family activities due to the amount of time = must spend on wor$ responsi'ilities.a TimeH'ased family interference with wor$ !, items# L&istin* =tems ",. My superiors and peers disli$e how often = am preoccupied with my personal life while at wor$. "!. My personal life ta$es up time that =?d li$e to spend at wor$.a Jew =tems !. The time = spend on family responsi'ilities often interfere with my wor$ responsi'ilities.a ". My family responsi'ilities prevent me from effectively performin* my 9o'. G. = find myself ma$in* family related phone calls or runnin* personal errands durin* wor$ time. D. The demands of my family life prevent me from developin* important career relationships. 6. The time = spend with my family often causes me to not spend time in activities at wor$ that could 'e helpful to my career.a A. = feel *uilty for spendin* time with my family when = $now = should 'e concentratin* on wor$. 7. = have to miss wor$ activities due to amount of time = must spend on family responsi'ilities.a 5. = feel = don?t have enou*h time to fulfill my potential in my career 'ecause = need to spend time with my family and friends.a StrainH'ased wor$ interference with family !, items# L&istin* =tem !A. =t is difficult for me to rela& when = am away from my wor$. Jew =tems !. The stress from my 9o' often ma$es me irrita'le when = *et home.a ". When = *et home from wor$ = am often too physically tired to participate in family activitiesFresponsi'ilities.a G. Tension and an&iety from wor$ often creep into my family life.a D. = often feel = am rushin* to *et my nonwor$ responsi'ilities ta$en care of in order to *et 'ac$ to wor$. 6. = am often stressed tryin* to 'alance my responsi'ilities when wor$ interferes with the rest of my life. A. = am often so emotionally drained when = *et home from wor$ that it prevents me from contri'utin* to my family.a 7. = am often preoccupied with wor$ while = am at home. 5. Due to all the pressures at wor$( sometimes when = come home = am too stressed to do the thin*s = en9oy.a 4. Sometimes = feel overwhelmed 'y all of my responsi'ilities at wor$. StrainH'ased family interference with wor$ !, items# L&istin* =tems !D. =?m often too tired at wor$ 'ecause of the thin*s = have to do at home. !7. My personal demands are so *reat that it ta$es away from my wor$. "G. My family life often interferes with my responsi'ilities at wor$. Jew =tems !. Due to stress at home( = am often preoccupied with family matters at wor$.a W>8K%:-M=LK C>J:L=CT "7G #EFE#ENCES -dams( 1. -.( Kin*( L. -.( 2 Kin*( D. W. !44A#. 8elationships of 9o' and family involvement( family social support( and wor$%family conflict with 9o' and life satisfaction. %ournal of $pplied 'sychology, 81, D!!%D",. -ryee( S. !44"#. -ntecedents and outcomes of wor$%family conflict amon* married professional womenE Lvidence from Sin*apore. Human &elations, 45, 5!G%5G7. -ryee( S.( 2 Lu$( C. !44A#. Wor$ and nonwor$ influences on the career satisfaction of dualHearner couples. %ournal of Vocational Behavior, 49, G5%6". 3acharach( S. 3.( 3am'er*er( ..( 2 Conley( S. !44!#. Wor$%home conflict amon* nurses and en*ineersE Mediatin* the impact of role stress on 'urnout and satisfaction at wor$. %ournal of (rgani)ational Behavior, 12, G4%6G. 3a*ozzi( 8. ..( 2 Ki( K. !455#. >n the evaluation of structural e)uation models. %ournal of the $cademy of Marketing Science, 16, 7D%4D. 3alfour( D. L.( 2 Wechsler( 3. !44A#. >r*anizational commitmentE -ntecedents and outcomes in pu'lic or*anizations. 'u#lic 'roductivity and Management &evie, 29, "6A%"77. -..LJD=B 3*Continued TimeH'ased wor$ interference with family !, items# ". Due to my family responsi'ilities( sometime others in the or*anization have to pic$ up the slac$ i.e.( stay late( travel#. G. The stress from my family life interferes with my wor$ life.a D. = feel rushed at wor$ so that = can *o home to my family. 6. 3ecause = am often stressed from family responsi'ilities( = have a hart time concentratin* on my wor$.a A. Tension and an&iety from my nonwor$ life often e&tend into my 9o'.a 7. Due to all the pressures at home( sometimes it is hard for me to do my 9o' well.a 3ehavior wor$ interference with family 7 itemsRall e&istin* items# "D. = am not a'le to act the same way at home as = do at wor$.a "6. The pro'lemHsolvin* approaches = use in my 9o' are not effective in resolvin* pro'lems at home.a "A. = act differently in respondin* to interpersonal pro'lems at wor$ than = do at home. "7. 3ehavior that is effective and necessary for me at wor$ would 'e counterproductive at home.a "5. The 'ehaviors = perform that ma$e me effective at wor$ do not help me to 'e a 'etter parent and spouse.a "4. What wor$s for me at home does not seem to 'e effective at wor$ as well. G,. =n order for me to 'e as successful at home as = am at wor$( = must 'ehave differently.a 3ehavior family interference with wor$ 7 itemsRall new items# !. The 'ehaviors that wor$ for me at home do not seem to 'e effective at wor$.a ". 3ehavior that is effective and necessary for me at home would 'e counterproductive at wor$.a G. The thin*s = do that ma$e me effective at home help me to 'e more successful at my 9o'. D. The pro'lem solvin* 'ehavior that wor$ for me at home does not seem to 'e as useful at wor$.a 6. =n order for me to succeed at wor$( = must 'e a different person than = can 'e at home. A. The 'ehaviors = use to respond to interpersonal pro'lems at wor$ perform 'etter at home than at wor$.a 7. = do not succeed at wor$ when = use the same 'ehaviors that are effective at home.a a =tems retained for Study G. "7D C-8LS>J( K-CM-8( -JD W=LL=-MS 3edeian( -. 1.( 3ur$e( 3. 1.( 2 Moffett( 8. 1. !455#. >utcomes of wor$%family conflict amon* married male and female professionals. %ournal of Management, 14G#( D76%D4!. 3ohen( <. C.( 2 CiverosHLon*( -. !45!#. Balancing :o#s and family life" Do fle;i#le ork schedules help< .hiladelphiaE Temple @niv. .ress. 3ollen( K. !454# Structural e=uations ith latent varia#les+ Jew Kor$E Wiley. 3uchanan( 3. !47D#. 3uildin* or*anizational commitmentE The socialization of mana*ers in wor$ or*anizations. $dministrative Science >uarterly, 19, 6GG%6DA. 3urley( K. !454#. ?ork@family conflict and marital ad:ustment in dual career couples" $ comparison of three time models+ @npu'lished doctoral dissertation( Claremont 1raduate School( Claremont( C-. Cammann( C.( :ichman( M.( Jen$ins( D.( 2 Klesh( J. !474#. 8he Michigan (rgani)ational $ssessment >uestionnaire+ -nn -r'or( Michi*anE @niv. of Michi*an. Cortina( J. M. !44G#. What is coefficient alphaN -n e&amination of theory and application. %ournal of $pplied 'sychology, 78, 45%!,D. DeCillis( 8. :. !44!#. Scale development" 8heory and applications+ Jew'ury .ar$( C-E Sa*e. Diener( L.( Lmmons( 8. -.( Larsen( 8. J.( 2 1riffin( S. !456#. The satisfaction with life scale. %ournal of 'ersonality $ssessment, 49!#( 7!%76. Du&'ury( L. L.( <i**ins( C. -.( 2 Mills( S. !44"#. -fterHhours telecommutin* and wor$%family conflictE - comparative analysis. 5nformation Systems &esearch, 3, !7G%!4,. La*le( 3. W.( Miles( L. W.( 2 =ceno*le( M. L. !447#. =nterrole conflicts and permea'ility of wor$ and family domainsE -re there *ender differencesN %ournal of Vocational Behavior, 50, !A5%!5D. Lisen'er*er( 8.( <untin*ton( 8.( <utchison( S.( 2 Sowa( D. !45A#. .erceived or*anizational support. %ournal of $pplied 'sychology, 71, 6,,%6,7. :ord( J. K.( MacCallum( 8. C.( 2 Tait( M. !45A#. The application of e&ploratory factor analysis in applied psycholo*yE - critical review and analysis. 'ersonnel 'sychology, 39, "4!%G!D. :rone( M. 8.( 8ussell( M.( 2 Cooper( M .L. !44"#. -ntecedents and outcomes of wor$%family conflictE Testin* a model of the wor$%family interface. %ournal of $pplied 'sychology, 77!#( A6%76. :rone( M. 8.( Kardley( J.( 2 Mar$el( K. S. !447#. Developin* and testin* an inte*ative model of the wor$%family interface. %ournal of Vocational Behavior, 50, !D6%!A7. 1reenhaus( J. <.( 3edeian( -. 1.( 2 Mossholder( K. W. !457#. Wor$ e&periences( 9o' performance( and feelin*s of personal and family wellH'ein*. %ournal of Vocational Behavior, 31, ",,%"!6. 1reenhaus( J. <.( 2 3eutell( J. J. !456#. Sources of conflict 'etween wor$ and family roles. $cademy of Management &evie, 10!#( 7A%55. 1reenhaus( J. <.( .arasuraman( S.( 1ranrose( C. S.( 8a'inowitz( S.( 2 3eutell( J. J. !454#. Sources of wor$%family conflict amon* two career couples. %ournal of Vocational Behavior, 34, !GG%!6G. 1ute$( 3.( Searle( S.( 2 Klepa( L. !44!#. 8ational versus *ender roleHe&planations for wor$%family conflict. %ournal of $pplied 'sychology, 76, 6A,%6A5. <i**ins( C.-.( Du&'ury( L. L.( 2 =rvin*( 8. <. !44"#. Wor$%family conflict in the dualHcareer family. (rgani)ational Behavior and Human Decision 'rocesses, 51, 6!%76. JoPres$o*( K.( 2 SoPr'om( D. !44G#. 65S&A6 ," Structural e=uation modeling ith the S5M'65S command language+ <illsdale( JJE Lrl'aum. Jud*e( T. -.( 3oudreau( J. W.( 2 3retz( 8. D. !44D#. Jo' and life attitudes of male e&ecutives. %ournal of $pplied 'sychology, 79, 7A7%75". Kim( J. >.( 2 Mueller( C. W. !475#. Factor analysis" Statistical methods and practical issues+ Jew .ar$( C-E Sa*e. Kopelman( 8. L.( 1reenhaus( J. <.( 2 Connolly( T. :. !45G#. - model of wor$( family( and interrole conflictE - construct validation study. (rgani)ational Behavior and Human 'erformance, 32, !45%"!6. Kosse$( L. L.( 2 >ze$i( C. !445#. Wor$%family conflict( policies( and the 9o'%life satisfaction relationshipE - review and directions for or*anizational 'ehavior%human resources research. %ournal of $pplied 'sychology, 83, !G4%!D4. W>8K%:-M=LK C>J:L=CT "76 Loerch( K. J.( 8ussell( J. L.( 2 8ush( M. C. !454#. The relationship amon* family domain varia'les and wor$%family conflict for men and women( %ournal of Vocational Behavior, 35, "55%G,5. Marsh( <. W. !446#. Confirmatory factor analysis models of factorial invarianceE - multifaceted approach. Structural A=uation Modeling, 1, 6%GD. Jetemeyer( 8. 1.( 3oles( J. S.( 2 McMurrian( 8. !44A#. Development and validation of wor$%family conflict and family%wor$ conflict scales. %ournal of $pplied 'sychology, 81, D,,%D!,. Junnally( J. C. !475#. 'sychometric theory .-nd ed+/+ Jew Kor$E Mc1raw%<ill. >?Driscoll( M. ..( =l*en( D. 8.( 2 <ildreth( K. !44"#. Time devoted to 9o' and offH9o' activities( interrole conflict( and affective e&periences. %ournal of $pplied 'sychology, 77, "7"%"74. .arasuraman( S.( 1reenhaus( J. <.( 8a'inowitz( S. 8.( 3edeian( -. 1.( 2 Mossholder( K. W. !454#. Wor$ and family varia'les as mediators of the relationship 'etween wives? employment and hus'ands? wellH'ein*. $cademy of Management %ournal, 32, !56%",!. .lec$( J. <. !477#. The wor$%family role system. Social 'ro#lems, 24, D!7%D"7. .lec$( J. <. !475#. ?ork@family conflict" $ national assessment+ .resented at the -nnual meetin* of the Society for the Study of Social .ro'lems( 3oston( M-. Quinn( 8.( 2 Staines( 1. !474#. 8he 4B22 =uality of employment survey+ -nn -r'orE @niv. of Michi*anE Survey 8esearch Center. 8ice( 8. W.( :rone( M. 8.( 2 Mc:arlin( D. 3. !44"#. Wor$%nonwor$ conflict and the perceived )uality of life. %ournal of (rgani)ational Behavior, 13, !66%!A5. 8izzo( J. 8.( <ouse( 8. J.( 2 Lirtzman( S. =. !47,#. 8ole conflict and am'i*uity in comple& or*anizations. $dministrative Science >uarterly, 15, !6,%!AG. Seashore( S. L.( Lawler( L. L.( Mirvis( ..( 2 Cammann( C. !45"#. (#serving and measuring organi)ational change" $ guide to field practice+ Jew Kor$E Wiley. Schau'roec$( J.( Cotton( J. L.( 2 Jennin*s( K. 8. !454#. -ntecedents and conse)uences of role stressE - covariance structure analysis. %ournal of (rgani)ational Behavior, 10, G6%65. Schriesheim( C. -.( 2 <in$in( T. 8. !44,#. =nfluence tactics used 'y su'ordinatesE - theoretical and empirical analysis and refinement of the Kipnis( Schmidt( and Wil$inson Su'scales. %ournal of $pplied 'sychology, 75, "DA%"67. Schriesheim( C. -.( .owers( K. J.( Scandura( T. -.( 1ardiner( C. C.( 2 Lan$au( M. J. !44G#. =mprovin* construct measurement in mana*ement researchE Comments and )uantitative approach for assessin* the theoretical content ade)uacy of paperHandHpencil surveyHtype instruments. %ournal of Management, 19, G56%D!7. Staines( 1. L.( 2 .lec$( J. <. !45G#. 8he impact of ork schedules on the family+ =nstitute for Social 8esearch( -nn -r'or( M=. Stephens( 1. K.( 2 Sommer( S. M. !44G#. ?ork@family conflict, :o# attitudes, and orkplace social support" 5nvestigations of measurement and moderation+ .resented at the meetin* of -cademy of Mana*ement( -tlanta( 1eor*ia. Stephens( 1. K.( 2 Sommer( S. M. !44A#. The measurement of wor$ to family conflict. Aducational and 'sychological Measurement, 56, D76%D5A. Stevens( J. !44"#. $pplied multivariate statistics for the social sciences+ Jew JerseyE Lrl'aum. Thomas( L. T.( 2 1anster( D. C. !446#. =mpact of familyHsupportive wor$ varia'les on wor$%family conflict and strainE - control perspective. %ournal of $pplied 'sychology, 80, A%!6. Wiley( D. L. !457#. The relationship 'etween wor$Fnonwor$ role conflict and 9o'Hrelated outcomesE Some unanticipated findin*s. %ournal of Management, 13, DA7%D7". Williams( K. J.( 2 -lli*er( 1. M. !44D#. 8ole stresssors( mood spillover( and perceptions of wor$%family conflict in employed parents. $cademy of Management %ournal, 37, 5G7%5A5. Williams( L. J.( 3ozdo*an( <.( 2 -imanHSmith( L. !44A#. =nference pro'lems with e)uivalent models. =n 1. -. Marcoulides 2 8. L. Schumac$er Lds.#( $dvanced structural e=uations modeling+ Mahah, !%" Arl#aum+ 8eceivedE Jovem'er ",( !445 "7A C-8LS>J( K-CM-8( -JD W=LL=-MS