12
Fathers’ Caregiving and Breadwinning: A Gender Congruence Analysis Trent W. Maurer and Joseph H. Pleck Georgia Southern University Maurer, Pleck, and Rane’s Gender Congruence Theory was further expanded via Bandura’s Social Cognitive Theory and tested to predict parenting behaviors in fathers and mothers. Results provided some support for the revised theory for fathers: Both perceived expectations from their wives for caregiving behavior and the perceived caregiving behaviors of other fathers positively predicted their own caregiving behav- ior, whereas no independent variables were able to predict their breadwinning behavior. Results provided little support for the revised theory for mothers: Only the perceived caregiving behaviors of other mothers positively predicted their own caregiving be- havior, and no independent variables predicted their breadwinning behavior. Role Identity was not a significant predictor of behavior in either role for fathers or mothers. Keywords: fathers, parenting, caregiving, Gender Congruence Theory, Social Cogni- tive Theory The roles, responsibilities, and functions of fatherhood are changing (Furstenberg, 1995; Palkovitz, 2002). Although financially provid- ing for one’s family (i.e., breadwinning) has traditionally been a task and responsibility as- signed to fathers (Aryee & Luk, 1996), mothers have been increasingly sharing in this responsi- bility (Hochschild & Machung, 1989). Con- versely, providing child care (i.e., caregiving) has traditionally been a task and responsibility assigned to mothers (Aryee & Luk, 1996), but fathers have been providing an increasing pro- portion of the family caregiving (see Pleck, 1997, for a review). Further, many fathers re- port that although they feel social pressure to be the family breadwinner (Doucet, 2004), they personally want to be more involved with care- giving and perceive increasing sociocultural ex- pectations to be involved with caregiving (Hen- wood & Procter, 2003). The changing nature of fatherhood is further complicated by two issues. First, although many fathers both want to be involved with caregiving and perceive a social expectation to be in- volved, caregiving is still primarily defined as woman’s work or feminine, which means that many fathers need to cognitively redefine those tasks that are nontraditional for men as still somehow being masculine to reduce the threat to their own masculinity (Doucet, 2004). That is, because the caregiving role is not tradition- ally defined as masculine, men are less likely to socially or personally identify with it as long as they perceive it as feminine and are more likely to participate in it because of extrinsic motiva- tions (e.g., because they think it will make their wives happy). Second, and largely as a product of the first issue, providing caregiving is still more discre- tionary for fathers than it is for mothers (Riggs, 2005), particularly when it comes to choosing between breadwinning and caregiving responsi- bilities (Singley & Hynes, 2005). This idea that fathers’ caregiving responsibilities are more voluntary and negotiable than mothers’ can even be seen in media portrayals of family life (Fleming & Tobin, 2005), which provide fa- Trent W. Maurer, Department of Hospitality, Tourism, and Family & Consumer Sciences, Georgia Southern Uni- versity; Joseph H. Pleck, Department of Human & Com- munity Development, University of Illinois at Urbana– Champaign. This manuscript is based on the first author’s doctoral dissertation. The work reported here was supported by the Cooperative State Research, Education and Extension Ser- vice, U.S. Department of Agriculture, under Project ILLU- 45-0366 to Joseph H. Pleck. Correspondence concerning this article should be ad- dressed to Trent W. Maurer, Department of Hospitality, Tourism, and Family & Consumer Sciences, P.O. Box 8021, Georgia Southern University, Statesboro, GA 30460. E-mail: [email protected] Psychology of Men & Masculinity Copyright 2006 by the American Psychological Association 2006, Vol. 7, No. 2, 101–112 1524-9220/06/$12.00 DOI: 10.1037/1524-9220.7.2.101 101

Fathers' caregiving and breadwinning: A gender congruence analysis

Embed Size (px)

Citation preview

Fathers’ Caregiving and Breadwinning:A Gender Congruence Analysis

Trent W. Maurer and Joseph H. PleckGeorgia Southern University

Maurer, Pleck, and Rane’s Gender Congruence Theory was further expanded viaBandura’s Social Cognitive Theory and tested to predict parenting behaviors in fathersand mothers. Results provided some support for the revised theory for fathers: Bothperceived expectations from their wives for caregiving behavior and the perceivedcaregiving behaviors of other fathers positively predicted their own caregiving behav-ior, whereas no independent variables were able to predict their breadwinning behavior.Results provided little support for the revised theory for mothers: Only the perceivedcaregiving behaviors of other mothers positively predicted their own caregiving be-havior, and no independent variables predicted their breadwinning behavior. RoleIdentity was not a significant predictor of behavior in either role for fathers or mothers.

Keywords: fathers, parenting, caregiving, Gender Congruence Theory, Social Cogni-tive Theory

The roles, responsibilities, and functions offatherhood are changing (Furstenberg, 1995;Palkovitz, 2002). Although financially provid-ing for one’s family (i.e., breadwinning) hastraditionally been a task and responsibility as-signed to fathers (Aryee & Luk, 1996), mothershave been increasingly sharing in this responsi-bility (Hochschild & Machung, 1989). Con-versely, providing child care (i.e., caregiving)has traditionally been a task and responsibilityassigned to mothers (Aryee & Luk, 1996), butfathers have been providing an increasing pro-portion of the family caregiving (see Pleck,1997, for a review). Further, many fathers re-port that although they feel social pressure to bethe family breadwinner (Doucet, 2004), they

personally want to be more involved with care-giving and perceive increasing sociocultural ex-pectations to be involved with caregiving (Hen-wood & Procter, 2003).

The changing nature of fatherhood is furthercomplicated by two issues. First, although manyfathers both want to be involved with caregivingand perceive a social expectation to be in-volved, caregiving is still primarily defined aswoman’s work or feminine, which means thatmany fathers need to cognitively redefine thosetasks that are nontraditional for men as stillsomehow being masculine to reduce the threatto their own masculinity (Doucet, 2004). Thatis, because the caregiving role is not tradition-ally defined as masculine, men are less likely tosocially or personally identify with it as long asthey perceive it as feminine and are more likelyto participate in it because of extrinsic motiva-tions (e.g., because they think it will make theirwives happy).

Second, and largely as a product of the firstissue, providing caregiving is still more discre-tionary for fathers than it is for mothers (Riggs,2005), particularly when it comes to choosingbetween breadwinning and caregiving responsi-bilities (Singley & Hynes, 2005). This idea thatfathers’ caregiving responsibilities are morevoluntary and negotiable than mothers’ caneven be seen in media portrayals of family life(Fleming & Tobin, 2005), which provide fa-

Trent W. Maurer, Department of Hospitality, Tourism,and Family & Consumer Sciences, Georgia Southern Uni-versity; Joseph H. Pleck, Department of Human & Com-munity Development, University of Illinois at Urbana–Champaign.

This manuscript is based on the first author’s doctoraldissertation. The work reported here was supported by theCooperative State Research, Education and Extension Ser-vice, U.S. Department of Agriculture, under Project ILLU-45-0366 to Joseph H. Pleck.

Correspondence concerning this article should be ad-dressed to Trent W. Maurer, Department of Hospitality,Tourism, and Family & Consumer Sciences, P.O. Box8021, Georgia Southern University, Statesboro, GA 30460.E-mail: [email protected]

Psychology of Men & Masculinity Copyright 2006 by the American Psychological Association2006, Vol. 7, No. 2, 101–112 1524-9220/06/$12.00 DOI: 10.1037/1524-9220.7.2.101

101

thers with ample examples that mothers are theprimary caregivers in society and that fathers’contributions to family caregiving are both min-imal and ancillary. So the question remains,Why do some fathers become more involvedwith caregiving, and how does their gender in-fluence the factors that encourage or discouragetheir involvement?

Although significant research on fatherhoodhas been conducted in the past quarter centuryand a substantial amount of that research hasbeen directed to attempting to explain why fa-thers choose to get involved with caregiving(see Marsiglio, Amato, Day, & Lamb, 2000, fora review; cf. Pleck, 1997), there have been fewattempts to identify how gendered expectationsfor parental behavior may influence fathers’(and mothers’) behavior. This fact is particu-larly striking given the increasing interestamong some researchers in the subjective expe-riences of fathers qua fathers from a symbolicinteractionist perspective, more generally, andan identity theory perspective, more specifically(e.g., Fox & Bruce, 1999; Ihinger-Tallman, Pas-ley, & Buehler, 1993; Marsiglio & Cohan,2000; Maurer, Pleck, & Rane, 2001; McBride &Rane, 1997; Minton & Pasley, 1996). Theseresearchers have attempted to illuminate theprocesses by which men identify with the socialrole of father, the meaning of that identity, andthe relationship between father identity andmen’s behaviors, particularly as they concerninvolvement with their children. Certainly, animportant part of not only identity, but also ofsocial roles and expectations is gender, yetmany of these researchers have failed to exam-ine how gendered expectations may help defineidentity and social expectations or even moder-ate the relationship between identity, social ex-pectations, and behavior.

However, this shortcoming may be explainedby the fact that this line of research has beenlargely atheoretical (outside of the general guid-ing principles of symbolic interactionism andidentity theory), and what theory has emergedhas generally been limited to explaining specificresults or justifying limited hypotheses (seeDoherty, Kouneski, & Erikson, 1998, for a re-view). Further, given that much of the researchon fatherhood has been done exclusively in menwithout comparison to women (e.g., Ihinger-Tallman et al., 1993; Minton & Pasley, 1996), itwould be difficult to develop a theoretical

model that adequately hypothesizes the effectgender would have on involvement.

Review of Gender Congruence Theory

There is one notable exception to the afore-mentioned pattern: Maurer et al.’s (2001) Gen-der Congruence Theory, which postulates aconnection between identity, the feedback indi-viduals receive from others about their identityand behavior (termed reflected-appraisals), andtheir behavior. Here, identification with a role(e.g., breadwinner) influences behavior in thatrole if the role is gender-traditional (e.g., bread-winning for men). Reflected-appraisals only in-fluence gender-nontraditional behaviors (e.g.,caregiving for men). The reason for this, ac-cording to Maurer et al. (2001), is that “becausethe traditional role expectations . . . are clearlydefined, women’s and men’s identities . . .should be clearly established independent ofperceived reflected-appraisals” (p. 312).

This theory is supported by numerous find-ings in the literature. Ihinger-Tallman et al.(1993) reported father parenting identity to besignificantly correlated with paying child sup-port among divorced fathers. McBride and Rane(1997) reported finding no relationship betweenfathers’ parental identity and paternal involve-ment, but mothers’ parental identity was signif-icantly correlated with mothers’ involvement. Asubsequent investigation by Rane and McBride(2000) focusing exclusively on fathers repli-cated those findings.

Pasley, Futris, and Skinner (2002) reportedsignificant positive correlations between fa-thers’ perceptions of their wives’ perceptions oftheir worth and competence as fathers and thosefathers’ performance of various child-relatedbehaviors. Fox and Bruce (2001) reported sig-nificant regression coefficients for perceived re-flected-appraisals on paternal behavior, paternalresponsiveness, and a composite measure ofpaternal involvement. Further, the effect of per-ceived reflected-appraisals on the compositemeasure remained significant even after con-trolling for the effect of paternal identity, whichis precisely what is predicted by Gender Con-gruence Theory for a gender-nontraditionalrole.

Maurer et al.’s (2001) own findings largelysupported their Gender Congruence Theory: Fa-thers’ caregiving behaviors were predicted not

102 MAURER AND PLECK

by their own caregiving identities, but by theirperceived reflected-appraisals about caregivingfrom their wives; fathers’ breadwinning behav-iors were predicted by their breadwinning iden-tities, but not by their perceived reflected-ap-praisals. However, the theory was only sup-ported for mothers’ caregiving: Mothers’caregiving behaviors were predicted by theircaregiving identities, not by their caregivingperceived reflected-appraisals. Mothers’ bread-winning behaviors were predicted by theirbreadwinning identities and not by their per-ceived reflected-appraisals (the opposite ofwhat was predicted).

Revision of Gender Congruence Theory

Given the ambiguous nature of expectationsfor gender–nontraditional behaviors, the need tosocially reference those behaviors in the ab-sence of a clear identity in those roles, and thechanging social expectations for fathers (Hen-wood & Procter, 2003), which may create a newsocial context for fathering, it may be beneficialto incorporate aspects of Bandura’s (1989) So-cial Cognitive Theory into Gender CongruenceTheory. Bandura’s theory suggests that individ-uals may model behavior they observe, partic-ularly in ambiguous situations and especially ifthat behavior is enacted by multiple members ofthe individual’s gender. Fathers may look toother fathers to acquire needed skills throughmodeling and determine what behaviors are ex-pected of them.

Some prior researchers have investigated theinfluence of fathers’ social networks on theirfathering. These studies have primarily focusedon the size of fathers’ social networks comparedwith those of mothers, focusing especially onthe balance between kin and nonkin networkmembers (Munch, McPherson, & Smith-Lovin,1997; Riley, 1990) and the encouragement andresources provided by fathers’ social networks(Riley, 1990; Russell, 1983, 1986).

Daly (1993) reported that the fathers he in-terviewed deliberately observed other fathers tolearn desired parenting skills or techniques andeven sought out “specific instances of goodfathering behavior among [their] peers” (p.523). Although Daly (1993) did not quantita-tively test a relationship between the perceivedbehavior of same-sex other parents and one’sown parenting behavior, his study clearly estab-

lished that fathers can and do model their fa-thering behavior after what they observe inother fathers. As a result, Gender CongruenceTheory will be expanded here to include thebroader social context and accommodate theperceived behavior of other parents of the samegender in both traditional and nontraditionalroles.

We will also attempt to control for parents’behavior in the other role, which was a missingelement in Maurer et al.’s (2001) original test-ing of the theory. Given the substantial evidenceto suggest that breadwinning (measured ashours spent working) is negatively related topaternal involvement (Ishii-Kuntz, Makino,Kato, & Tsuchiya, 2004) and positively relatedto paternal perceptions of not spending enough“quality time” (which could be interpreted as adimension of involvement) with children(Milkie, Mattingly, Nomaguchi, Bianchi, &Robinson, 2004), breadwinning hours should beentered as an independent variable in the care-giving models. Likewise, it is necessary to con-trol for the potential impact of caregiving be-havior on breadwinning behavior, so caregivingbehavior should be entered as an independentvariable in the breadwinning models. Using thisapproach, it will be possible to determine theinfluence that identity, perceived reflected-ap-praisals, and perceived behavior of other par-ents have on parents’ own gender-traditionaland gender-nontraditional behaviors indepen-dent of the influence of their behavior in theother role.

Testing our revision of Gender CongruenceTheory will require separate analyses for bothmothers and fathers for both caregiving andbreadwinning behaviors. This investigation isguided by three hypotheses:

Hypothesis 1: Parents’ behavior will bepositively predicted by their identity in thegender-traditional role, but not in the gen-der-nontraditional role.

Hypothesis 2: Parents’ behavior will bepositively predicted by perceived reflect-ed-appraisals from their partners in thegender-nontraditional role, but not in thegender-traditional role.

Hypothesis 3: Parents’ behavior will bepositively predicted by the perceived be-havior in that role of same-sex other par-

103FATHERS’ CAREGIVING AND BREADWINNING

ents in the gender-nontraditional role, butnot in the gender-traditional role.

Methods

In this investigation, the target populationwas parents in two-parent families who had atleast one preschool-aged child (2–5 years old).The strategy we chose to locate and recruit theseparents was to contact local child-care centersand laboratory preschools and use them as basesof recruitment. In exchange for filling out thequestionnaire and returning it to their child-carefacility, parents received a children’s bookwhich they selected from a provided list.

Parents in two-parent families with at leastone child aged 2–5 were recruited at seven sites.The first site was an on-campus laboratory pre-school serving 2- to 5-year-olds in both univer-sity and community families. The second sitewas a similar facility at a local communitycollege. The remaining five sites were state-licensed child-care facilities in the same com-munity. The sites were selected based on thewillingness of their directors to assist in datacollection.

Participants

At the first site, 33 of 92 eligible fathers(35.9%) and 39 of 92 eligible mothers (42.4%)returned completed questionnaires. At the re-maining sites, administrative data on parents’marital status were unavailable, so an exactnumber of eligible parents and thus precise par-ticipation rates, could not be determined. Theremaining 14 fathers and 17 mothers were dis-tributed among the other six sites as follows:from the second site, 3 fathers and 4 mothers;from the third site, 1 father and 1 mother; fromthe fourth site, 2 fathers and 2 mothers; from thefifth site, 1 father and 2 mothers; from the sixthsite, 6 fathers and 6 mothers; and from theseventh site, 1 father and 2 mothers. The finalsample was 47 fathers and 56 mothers. Onefather and 1 mother were unmarried cohabitors;the remaining 46 fathers and 55 mothers weremarried. (One father and 1 mother listed theirmarital status as “separated,” and given that thefocus of this investigation on marital dynamics,it was decided not to include them in the anal-yses, nor are they represented in the numbersabove.)

For the first site, administrative data availablefor all families using the facility made it possi-ble to compare the demographic characteristicsof the 33 fathers and 39 mothers from that sitein the analysis sample with the 59 eligible fa-thers and 53 eligible mothers who did not re-turnquestionnaires. The results of these analysesindicated that there were no significant differ-ences between the groups. For example, samplemothers and nonsample mothers did not signif-icantly differ in their ages, levels of education,or ethnic backgrounds.

Because nearly two thirds of the total samplecame from this first site, additional analyseswere conducted to determine whether the par-ticipants from this site were different from par-ticipants at the other sites. The results of theseanalyses also revealed that there were no sig-nificant differences between the groups. For ex-ample, fathers at the first site did not differsignificantly from fathers at the other sites in thefrequency with which they had male versusfemale children, the ages of their children, ortheir own levels of education.

Of the fathers, 80.85% (38 of 47) and 83.93%of mothers (47 of 56) were White. For fathers,the ethnicities of the remaining nine were Afri-can American, 1; Asian, 3; Hispanic, 3; andother, 2. For mothers, the ethnicities of theremaining nine were: African American, 1;Asian, 4; Hispanic, 3; and other, 1. Because ofthe small number of minority participants, eth-nicity was dichotomously coded into White (1)and other (0). Average age of fathers was 38.20years (SD � 5.90), with a mean education levelof 18.30 years (SD � 2.06). Average age ofmothers was 36.20 years (SD � 5.50), with amean education level of 18.02 years(SD � 2.14). Of the fathers, 91.49% (43 of 47)were employed, with 1 father’s status beingunreported. Of the mothers, 71.43% (40 of 56)were employed. The mean number of the sumof weekly paid and unpaid work hourswas 51.62 for fathers (SD � 30.38) and 24.49for mothers (SD � 20.72).

Questionnaires asked parents to think of theiractivities with reference to their child; if parentshad more than one child enrolled at their child-care facility, questionnaires instructed parentsto think of their oldest child served by theirchild care facility (“target child”). With this inmind, of the fathers’ children, 46.81% (22) wereboys and 53.19% (25) were girls, with an aver-

104 MAURER AND PLECK

age age of 4.54 years (SD � .83). Of the moth-ers’ target children, 42.86% (24) were boysand 57.14% (32) were girls, with an average ageof 4.56 years (SD � .76). All of these samplecharacteristics were used as control variables.

Measures

Caregiving and Breadwinning Identity andReflected-Appraisal Inventory. Parents’ care-giving and breadwinning identities and per-ceived reflected-appraisals were assessed usingMaurer et al.’s (2001) Caregiving and Bread-winning Identity and Reflected-Appraisal In-ventory (CBIRAI). The CBIRAI measures theextent to which parents see themselves (iden-tity), see their partners (reflected-appraisal), andbelieve they are seen by their partners (per-ceived reflected-appraisal), as caregivers andbreadwinners. Most responses are scored on a5-point agree or disagree Likert scale, with afew items asking participants to indicate theimportance of an item (not at all to extremely).A mean score is calculated for each subscalewith higher scores indicating more of the con-struct (e.g., a mean score of 4 on the caregivingidentity subscale means a participant identifiesmore with the role of caregiver than a partici-pant who has a mean score of 2). A sample itemfor the caregiving identity subscale is “I wouldlike to be remembered for the quality of care Igave my child.”

Because analyses were not restricted to thecouple level (which is necessary to use thereflected-appraisal subscale), only the identityand perceived reflected-appraisal subscales ofthe CBIRAI were used. Alphas for all measuresappear in Table 1.

Caregiving Involvement Scale. Parents’caregiving behavior was assessed with the Care-giving Involvement Scale (CIS), adapted fromBruce and Fox’s (1997) Father InvolvementScale (FIS). The FIS is a 21-item measure, withfour subscales (caregiving, socioemotional, ex-ecutive function, and teaching). The FIS wasadapted for use in this study by rewording thedirections to be applicable to both fathers andmothers (i.e., asking participants to think aboutwhat they do as “parents” rather than as “fa-thers”) and using only the subsection of the FISdirectly applicable to caregiving behaviors (sixitems). Items in the CIS ask how frequentlyparents engage in specific caregiving activities.

Items are rated on a 4-point scale with responsesranging from never or hardly ever, sometimes,often, to almost every day. A sample item asksparents how often they “Assist or superviseyour child in bathing (or personal hygiene).”

Caregiving Involvement Scale–Other Par-ents. The perceived caregiving behavior ofother same-gender parents was assessed withthe Caregiving Involvement Scale–Other Par-ents (CIS–OP), which was developed for use inthis study from the CIS, using the same re-sponse format. The six-item CIS–OP asks howfrequently other same-gender parents engage inspecific caregiving activities. A sample itemasks parents how often they believe other moth-ers or fathers “Assist or supervise their child inbathing (or personal hygiene).” The scale wascreated as a mean of all the scale’s items.

Breadwinning Involvement Scale. Parents’breadwinning behavior was assessed with theMaurer et al. (2001) Breadwinning InvolvementScale (BIS). The BIS is a two-item free-re-sponse scale, asking parents to report their totalweekly paid and unpaid employment hours. Thetwo numbers are then added to create a total ofweekly employment hours.

Breadwinning Involvement Scale–Other Par-ents (BIS–OP). The perceived breadwinningbehavior of other same-gender parents wasassessed with the Breadwinning InvolvementScale–Other Parents (BIS–OP), which wasdeveloped for use in this study from the BIS,using the same response format. The two-itemscale asks parents how many paid and unpaidhours they think other same-gender parentswork in a week. This scale is a sum of the twoitems.

Results

A correlational matrix was created to deter-mine the relationship between the control, inde-pendent, and dependent variables for both fa-thers and mothers. Few correlations between thecontrol and dependent variables reached signif-icance, although many of the correlations be-tween independent and dependent variablesreached significance. These results are presentedin Table 1. Because parent’s ethnicity and childgender are dichotomous variables, correlationswith those variables are point-biserial.

105FATHERS’ CAREGIVING AND BREADWINNING

Tab

le1

Cor

rela

tion

sB

etw

een

Pro

ject

Var

iabl

es

Var

iabl

ePA

GE

PED

ET

HC

HA

GE

GE

ND

ER

CG

IDC

GPR

AC

ISO

PC

ISB

WID

BW

PRA

BIS

OP

BIS

PAG

E—

.35*

.16

�.1

4�

.01

.27

.26

�.2

3.0

2�

.15

�.2

0.1

2.1

1—

PED

.29*

—�

.01

�.0

7.4

1**

.24

.39*

*�

.10

.23

�.2

2�

.13

.04

�.0

9—

ET

H.1

7.1

0—

�.0

7�

.02

�.3

1*�

.18

.17

�.0

8.2

8.4

9**

.11

.44*

*—

CH

AG

E.0

1.0

3.1

8—

.12

�.0

7�

.02

�.0

4.0

4�

.18

�.1

7.0

2�

.07

—G

EN

DE

R.1

5.2

5�

.08

.10

—.0

8�

.01

�.0

7.0

8�

.06

�.1

3.0

2.1

2—

CG

ID�

.17

�.4

0**

�.1

6�

.07

�.1

6—

.65*

**.2

5.5

1***

�.3

2*�

.22

�.0

7�

.10

.74

CG

PRA

�.2

9*�

.38*

*�

.25

�.0

9�

.25

.65*

**—

.24

.57*

**�

.24

�.1

7�

.06

�.1

7.6

3C

ISO

P�

.12

.12

.01

�.1

7.0

1.0

6.1

3—

.47*

**.1

0.3

2*�

.18

�.1

1.9

5C

IS�

.22

�.3

4**

�.1

6�

.06

�.1

6.3

3*.3

3*.3

6**

—�

.38*

*�

.03

�.2

0�

.12

.79

BW

ID.0

5.1

6.0

9�

.18

.12

�.1

8�

.11

.36*

*—

.62*

**�

.06

.17

.76

BW

PRA

.11

.02

.08

.09

.15

�.1

4�

.13

.19

�.0

8.5

9***

—�

.11

.20

.86

BIS

OP

.10

.16

.17

�.0

3.1

7�

.09

.40*

*�

.08

�.1

1.2

2.1

9—

.25

—B

IS.3

9**

.30*

.12

�.1

7.1

2�

.18

�.2

0.0

7�

.42*

**.3

5**

.33*

.19

——

Not

e.N

umbe

rsbe

low

the

diag

onal

are

for

mot

hers

(N�

56),

num

bers

abov

eth

edi

agon

alar

efo

rfa

ther

s(N

�47

).A

lpha

valu

esar

efo

rth

eco

mbi

ned

sam

ple.

PAG

E�

pare

nt’s

age;

PED

�pa

rent

’sed

ucat

ion;

ET

H�

pare

nt’s

ethn

icity

(1�

whi

te,0

�ot

her)

;CH

AG

E�

child

age;

GE

ND

ER

�ch

ildge

nder

(1�

girl

,0�

boy)

;CG

ID�

care

givi

ngid

entit

y;C

GPR

A�

care

givi

ngpe

rcei

ved

refle

cted

-app

rais

al;

CIS

OP

�pe

rcei

ved

care

givi

ngbe

havi

orof

othe

rm

othe

rs/f

athe

rs;

CIS

�ow

nca

regi

ving

beha

vior

;B

WID

�br

eadw

inni

ngid

entit

y;B

WPR

A�

brea

dwin

ning

perc

eive

dre

flect

ed-a

ppra

isal

;B

ISO

P�

perc

eive

dbr

eadw

inni

ngbe

havi

orof

othe

rm

othe

rs/f

athe

rs;

BIS

�ow

nbr

eadw

inni

ngbe

havi

or.

*p

�.0

5.**

p�

.01.

***

p�

.001

.

106 MAURER AND PLECK

Multiple Regression Analyses

In both regressions (caregiving and bread-winning) for both mothers and fathers, controlvariables that were significantly correlated withthe dependent variable were forward entered inthe first block (using the standard default crite-ria of p � .05 to enter and p � .10 to remove).For fathers’ caregiving, no control variableswere entered. For fathers’ breadwinning, fa-thers’ ethnicity (dummy coded as White or non-White) was entered. For mothers’ caregiving,mothers’ education was entered. For mothers’breadwinning, both mothers’ education andmothers’ age were entered. Forward entry waschosen for the control variables because of thealready limited power to detect significant ef-fects for the independent variables, because ofthe small sample size. Independent variableswere force entered into regression. Two-tailedtests of significance were chosen to maintain aconservative estimate. Multicollinearity was as-sessed for all regression models, but none wasevident.

For fathers, the caregiving model was signif-icant, F(4, 41) � 8.88, p � .001, R2 � .41. Onlytwo of the independent variables achieved sig-nificance: caregiving perceived reflected-ap-praisals, b � 0.68, SE � 0.28, p � .05, andperceived caregiving behavior of other fathers,b � 0.38, SE � 0.13, p � .01. Neither caregiv-ing identity (b � 0.42, SE � 0.32, ns) norbreadwinning behavior (b � 0.00, SE � .00, ns)achieved significance. Results are presented inTable 2.

In the fathers’ breadwinning model, only fa-thers’ ethnicity emerged as a significant predic-

tor, b � 23.50, SE � 9.35, p � .05, and theoverall model was significant, F(5, 40) � 2.52,p � .05, R2 � .14. None of the independentvariables emerged as significant: breadwinningidentity, b � 2.05, SE � 9.85, ns; breadwinningperceived reflected-appraisals, b � �0.02,SE � 9.28, ns; perceived breadwinning of otherfathers, b � 0.41, SE � 0.29, ns; and caregivingbehavior, b � �1.08, SE � 5.77, ns. Results arepresented in Table 3.

For mothers, the caregiving model was sig-nificant, F(5, 49) � 7.21, p � .01, R2 � .37.Mothers’ education emerged as a significantpredictor in the model, but it dropped below thep � .05 level of significance in the final model,b � �0.43, SE � 0.03, ns. (It was retainedbecause the level of significance was between.05 and .10.) Two of the independent variablesachieved significance: breadwinning behavior,b � �0.01, SE � 0.00, p � .01, and perceivedcaregiving behavior of other mothers, b � 0.40,SE � 0.11, p � .01. Neither caregiving identity,b � .18, SE � .18, ns, nor caregiving perceivedreflected-appraisals, b � 0.04, SE � 0.16, ns,achieved significance. Results are presented inTable 2.

In the mothers’ breadwinning model, the modelwas again significant, F(5, 45) � 5.62, p � .01,R2 � .32. Mothers’ age reached significance,b � 6.00, SE � 2.24, p � .05. Of the independentvariables, only caregiving behavior, b � �17.42,SE � 5.90, p � .01, reached significance. Bread-winning identity, b � 12.99, SE � 7.02, ns, bread-winning perceived reflected-appraisals, b � 2.20,SE � 4.98, ns, and perceived breadwinning be-havior of other mothers, b � 0.03, SE � 0.16, ns,

Table 2Regression Analysis Predicting Fathers’ and Mothers’ Caregiving Behaviors

Variable

Fathers(N � 47)

Mothers(N � 56)

b � b �

Breadwinning behavior 0.00 �0.01 �0.01 �.35**Caregiving identity 0.42 0.20 0.18 .15Caregiving perceived reflected-appraisal 0.68 0.36* 0.04 .04Perceived caregiving behavior of other

same-sex parents 0.38 0.34** 0.40 .39**Education — — �0.04 �.21Adjusted R2 .41*** .37***

Note: Model significance, reported on model adjusted R2, was assessed with an F test.* p � .05. ** p � .01. *** p � .001.

107FATHERS’ CAREGIVING AND BREADWINNING

all failed to reach significance. Results are pre-sented in Table 3.

Discussion

For fathers, the expanded Gender Congru-ence Theory model was almost fully supported.For five of the six predictions, results were ashypothesized. In the gender-nontraditional roleof caregiving, fathers’ behavior was predictedonly by perceived reflected-appraisals and theperceived caregiving behavior of other fathers,but not by their own identity. In the gender-traditional role of breadwinning, fathers’ behav-ior was not significantly predicted by eitherperceived reflected-appraisals or the perceivedbreadwinning behavior of other fathers, but nei-ther was it predicted by fathers’ own breadwin-ning identity (contrary to the predictions of Hy-pothesis 1).

Given the prior findings by Maurer et al.(2001) that fathers’ breadwinning identity was asignificant predictor of their breadwinning be-havior, this is somewhat surprising. The firstinterpretation of this deviation would be to at-tribute it to the fact that in this investigation, wecontrolled for caregiving behavior and fathers’ethnicity in the breadwinning equation, whereasMaurer et al. did not. However, given the factthat the zero-order correlation between bread-winning identity and behavior was not signifi-cant to begin with, that interpretation is notlikely. Rather, it may be that identity is irrele-vant to breadwinning behavior for fathers be-cause low levels of self-identification with thebreadwinner role do not excuse men from the

social obligation to be a breadwinner (Doucet,2004). As this interpretation would still not ex-plain why Maurer et al. were able to documenta connection between breadwinning identityand behavior, it is apparent that further researchon this topic is required to more fully explorethe nature of the (possible) connection.

Despite this deviation from predictions, thedata gave strong support to the theory. Thissupport for the revised Gender CongruenceTheory for fathers provides further support forthe conceptualization of paternal caregiving be-haviors as a function of social expectations.Although this conceptualization is traditionallydiscussed in the literature in terms of mothers’negative influences upon paternal caregiving (i.e.,“maternal gatekeeping”; Allen & Hawkins,1999; McBride et al., 2005), it may be benefi-cial to expand this discussion to include theways in which positive expectations can en-courage paternal caregiving, particularly giventhe significantly positive effect of perceived re-flected-appraisals on fathers’ caregiving. Thatis, it may be more appropriate, particularly froman intervention perspective, to consider theways in which maternal expectations, or lackthereof, for paternal caregiving influence fathersto become more involved with caregiving, ratherthan more narrowly focusing on the ways mothersmay restrict paternal involvement.

Further, this support suggests that some so-cial modeling may be occurring among fathers.That is, the more involved fathers perceiveother fathers to be, the more they attempt tomodel the level of that involvement (and themore models they may have), which may be

Table 3Regression Analysis Predicting Fathers’ and Mothers’ Breadwinning Behaviors

Variable

Fathers(N � 47)

Mothers(N � 56)

b � b �

Caregiving behavior �1.08 �.03 �17.42 �.36**Breadwinning identity 2.05 .04 12.99 .29Breadwinning perceived reflected-appraisal �0.02 .00 2.20 .07Perceived breadwinning behavior of other

same-sex-parents 0.41 .29 0.03 .02Ethnicity 23.50 .40* — —Age — — 6.00 .32*Adjusted R2 .14* .32***

Note: Model significance, reported on model adjusted R2, was assessed with an F test.* p � .05. ** p � .01. *** p � .001.

108 MAURER AND PLECK

directly assisted by observing practical skillsfrom those parents necessary for that involvement.

This finding may be the most important con-tribution of this investigation. Prior researchershave not assessed or even explored this variableas a potential predictor of fathers’ caregivingbehavior (outside of Daly’s, 1993, observations,as noted in the literature review), which makesthe fact that it emerged as a significant predictorin these analyses more important, especiallyafter we had already accounted for the effect ofcaregiving perceived reflected-appraisals (whichprior research has established to be a highly sig-nificant influence on fathers’ caregiving). Futureresearchers should include this potential influenceand more fully explore the effect it may have onfathers’ caregiving behaviors.

Both perceived reflected-appraisals and theperceived caregiving behaviors of other fathersare effectively products of fathers’ social cog-nition. That is, fathers’ perceptions of theirwives’ expectations for caregiving and fathers’perceptions of the caregiving behaviors of otherfathers, both of which are fathers’ cognitiveinterpretations and assessments of their socialenvironments, are highly significant positivepredictors of their own caregiving behavior.This suggests that fathers actively define theirpaternal role as caregivers in response to theirsocial perceptions, as postulated in our GenderCongruence Model.

For mothers, the data did not generally sup-port the revised Gender Congruence Theory.Only two of the six predictions were supported:Mothers’ breadwinning behavior was not pre-dicted by their breadwinning identities, andmothers’ caregiving behavior was not predictedby their perceived reflected-appraisals for care-giving. This finding is problematic for threereasons. First, the fact that only two of sixpredictions were supported by the data suggeststhat although Gender Congruence Theory maybe useful in predicting fathers’ behaviors, itprobably has limited usefulness for mothers.This conclusion is further suggested by the sec-ond problem, which is that both supported pre-dictions are null predictions and, as such, arenot meaningful empirically. Third, these resultscontradict the findings of Maurer et al. (2001),which indicated that for both behaviors, identitywas a significant predictor. However, the reasonfor the deviation from the Maurer et al. findingsmay be a product of controlling for the other

behavior in the regression equations. This the-ory is particularly likely given the significant(yet relatively small) correlations between iden-tity and behavior for both caregiving and bread-winning for mothers. It may be that given thesmall sample size in this investigation, therewas insufficient power to detect a relationshipbetween identity and behavior after controllingfor the larger effect of the other behavior.

Three unanticipated significant findingsemerged for mothers. First, contrary to the pre-dictions of Hypothesis 3, mothers’ caregivingbehavior was significantly predicted by the per-ceived caregiving behaviors of other mothers.Although this result was discussed above forfathers, the same implications hold for mothers:As this is the first direct testing of this variablefor mothers and it emerged as significant overprior successful predictors, future researcherswill need to investigate the effect that this per-ception may have on mothers’ caregivingbehavior.

Second, mothers’ age was a significant posi-tive predictor of mothers’ breadwinning behav-ior. The most parsimonious explanation for thisfinding is that older mothers are more likely tobe further along in their careers and thus moreestablished in them by the time they have chil-dren, which would make them less likely toleave those careers.

Third, mothers’ breadwinning behavior was astrong negative predictor of mothers’ caregiv-ing behavior (and vice versa). This finding isconsistent with prior literature on mothers’ em-ployment, which suggests that the more amother is involved with the worker role, the lesstime she has to enact the mother role (Pleck,1985), and may be particularly clear in thisinvestigation since breadwinning behavior wasoperationalized as the number of paid and un-paid hours at work.

Additionally, this finding is interesting be-cause a similar relationship did not appear forfathers, despite the fact that fathers reportedworking twice as many hours on average asmothers did. Although a few studies have doc-umented a negative association between fa-thers’ work hours and caregiving behaviors(Ishii-Kuntz et al., 2004; Milkie et al., 2004),the overall conclusion in the literature on thesubject is that fathers’ work hours are not con-sistently related to caregiving behaviors (seePleck, 1997, for a review). These findings sug-

109FATHERS’ CAREGIVING AND BREADWINNING

gest that although caregiving and breadwinningbehaviors may be competitively organized formothers, they are not for fathers, which mayalso help to explain why the revised GenderCongruence Theory explained paternal behav-iors better than maternal behaviors. That is,Gender Congruence Theory conceptualizes theroles of caregiver and breadwinner in a non-competitive orthogonal way in terms of tradi-tional gender assignment rather than as compet-ing roles that individual parents must choosebetween. For use in understanding mothers’ be-havior, Gender Congruence Theory may need toinclude the competitive organization of caregiv-ing and breadwinning for mothers and also bemodified in other ways.

Limitations of the Study

Arguably the most significant limitation inthis project is that although a project goal was topredict parental behavior, the data were neitherlongitudinal nor experimental. As a result, thedirectionality of the relationships between thepredictor variables and parental behavior cannotconclusively be established nor can causality beconfirmed.

Second, the CIS defines caregiving behaviorsin a relatively narrow way, which may not cap-ture the extent to which some parents are in-volved with their children. For example, a fathermay decide that the mother is better at providingphysical care to their child than he is, and in thebest interests of the child (as he perceives it), herefrains from providing caregiving so that shecan provide it all. Clearly, his choice could beconceptualized as involvement, but it would notappear as caregiving behavior on his part if theCIS were used. Further, in this investigation, thereliability of the CIS was slightly lower thanthat reported in the literature, particularly formothers. Although the lower reliability ismostly likely a product of a smaller sample size,it is possible that the measure does not capturecaregiving behavior as well for mothers as itdoes for fathers. Additionally, the CIS measurescaregiving behaviors on an entirely differentscale than the BIS uses to measure breadwin-ning behaviors (never to almost every day vs.number of hours). Because the scales are notparallel, any comparisons between caregivingand breadwinning must be made with caution.

Third, there are numerous issues with oursample. This sample was relatively small (47fathers and 56 mothers), and, thus, there waslimited power to detect significant relationships,particularly when multiple predictors were si-multaneously entered in the same regressionmodels. Additionally, this sample was fairlyhomogeneous; ethnic and educational diversitywere limited, which compromises the externalvalidity of this study. Ironically, however, thislack of diversity facilitates comparison betweenthis study and prior ones using these measures,because such convenience samples have beentypical (Maurer et al., 2001; Maurer, Pleck, &Rane, 2003).

Further, as in many convenience samples,issues of self-selection need to be considered.Although our analyses revealed no differencebetween participants and nonparticipant s (orbetween participants at different sites) on thedemographic variables, there is no way to knowwhether there would have been differences be-tween the groups on the project variables. Fur-ther, rejection rates at the site for which theycould be calculated exceeded 50%. Althoughthis rejection rate further compromises the gen-eralizability of this investigation, it is not at allsurprising. Prior investigations of a similar na-ture have had rejection rates of approximately50%–66% (Bruce & Fox, 1999; Maurer et al.,2001). Recruiting fathers to participate in re-search, even when one is paying substantialincentives, is exceptionally difficult (Cook,Jones, Dick, & Singh, 2005).

Future Directions

However, despite these limitations the re-vised Gender Congruence Theory offers apromising new approach to explaining fathers’caregiving and breadwinning behaviors as aproduct of their identities, perceived reflected-appraisals, and perceived behaviors of other fa-thers. Future researchers could attempt to ad-dress some of the shortcomings of this investi-gation by recruiting more representativesamples of fathers (and mothers) or by explor-ing other ways of operationalizing caregivinginvolvement. In addition, future researcherscould attempt to further refine the Gender Con-gruence Theory to more adequately predictmothers’ parenting behaviors, perhaps by ac-counting for the competitive organization of

110 MAURER AND PLECK

roles among mothers or by assessing the degreeto which mothers perceive a given role as gen-der-congruent or incongruent.

References

Allen, S. M., & Hawkins, A. J. (1999). Maternalgatekeeping: Mothers’ beliefs and behaviors thatinhibit greater father involvement in family work.Journal of Marriage & the Family, 61, 199–212.

Aryee, S., & Luk, V. (1996). Balancing two majorparts of adult life experience: Work and familyidentity among dual-earner couples. Human Rela-tions, 49, 465–487.

Bandura, A. (1989). Social cognitive theory. In R.Vasta (Ed.), Annals of child development, Vol. 6:Six theories of child development: Revised formu-lations and current issues (pp. 1–60). Greenwich,CT: JAI Press.

Bruce, C., & Fox, G. L. (1997, November). Measur-ing father involvement among lower-income whiteand African-American populations. Poster sessionpresented at the annual meeting of the NationalCouncil on Family Relations, Arlington, VA.

Bruce, C., & Fox, G. L. (1999). Accounting forpatterns of father involvement: Age of child, fa-ther-child coresidence, and father role salience.Sociological Inquiry, 69, 458–476.

Cook, J. L., Jones, R. M., Dick, A. J., & Singh, A.(2005). Revisiting men’s role in father involve-ment: The importance of personal expectations.Fathering, 3, 165–178.

Daly, K. (1993). Reshaping fatherhood: Finding themodels. Journal of Family Issues, 14, 510–530.

Doherty, W. J., Kouneski, E. F., & Erickson, M. F.(1998). Responsible fathering: An overview andconceptual framework. Journal of Marriage andFamily, 60, 277–292.

Doucet, A. (2004). “It’s almost like I have a job, butI don’t get paid”: Fathers at home reconfiguringwork, care, and masculinity. Fathering, 2, 277–302.

Fleming, L. M., & Tobin, D. J. (2005). Popularchild-rearing books: Where is daddy? Psychologyof Men & Masculinity, 6, 18–24.

Fox, G. L., & Bruce, C. (1999). The anticipation ofsingle parenthood: A profile of men’s concerns.Journal of Family Issues, 20, 485–506.

Fox, G. L., & Bruce, C. (2001). Conditional father-hood: Identity theory and parental investment the-ory as alternative sources of explanation of father-ing. Journal of Marriage and Family, 63, 394–403.

Furstenberg, F. F. (1995). Dealing with dads: Thechanging roles of fathers. In P. L. Chase-Lansdale& J. Brooks-Gunn (Eds.), Escape from poverty:

What makes a difference for children? (pp. 189–210). NY: Cambridge University Press.

Henwood, K., & Procter, J. (2003). The “good fa-ther”: Reading men’s accounts of paternal involve-ment during the transition to first-time fatherhood.British Journal of Social Psychology, 42, 337–355.

Hochschild, A., & Machung, A. (1989). The secondshift: Working parents and the revolution at home.New York: Viking.

Ihinger-Tallmann, M., Pasley, K., & Buehler, C.(1993). Developing a middle range theory of fatherinvolvement postdivorce. Journal of Family Is-sues, 14, 550–571.

Ishii-Kuntz, M., Makino, K., Kato, K., & Tsuchiya,M. (2004). Japanese fathers of preschoolers andtheir involvement in child care. Journal of Mar-riage & Family, 66, 779–791.

Marsiglio, W., Amato, P., Day, R. D., & Lamb, M. E.(2000). Scholarship on fatherhood in the 1990sand beyond. Journal of Marriage and Family, 62,1173–1191.

Marsiglio, W., & Cohan, M. (2000). Contextualizingfather involvement and paternal influence: Socio-logical and qualitative themes. Marriage & FamilyReview, 29, 75–95.

Maurer, T. W., Pleck, J. H., & Rane, T. R. (2001).Parental identity and reflected-appraisals: Mea-surement and gender dynamics. Journal of Mar-riage and Family, 63, 309–321.

Maurer, T. W., Pleck, J. H., & Rane, T. R. (2003).Methodological considerations in measuring pater-nal identity. Fathering, 1, 117–129.

McBride, B. A., Brown, G. L., Bost, K. K., Shin, N.,Vaughn, B., & Korth, B. (2005). Paternal identity,maternal gatekeeping, and father involvement.Family Relations, 54, 360–372.

McBride, B. A., & Rane, T. R. (1997). Role identity,role investments, and paternal involvement: Impli-cations for parenting programs for men. EarlyChildhood Research Quarterly, 12, 173–197.

Milkie, M. A., Mattingly, M. J., Nomaguchi, K. M.,Bianchi, S. M., & Robinson, J. P. (2004). The timesqueeze: Parental statuses and feelings about timewith children. Journal of Marriage & Family, 66,739–761.

Minton, C., & Pasley, K. (1996). Fathers’ parentingrole identity and father involvement: A compari-son of nondivorced and divorced, nonresident fa-thers. Journal of Family Issues, 17, 26–45.

Munch, A., McPherson, J. M., & Smith-Lovin, L.(1997). Gender, children, and social contact: Theeffects of childrearing for men and women. Amer-ican Sociological Review, 62, 509–520.

Palkovitz, R. (2002). Involved fathering and men’sadult development: Provisional balances. Mah-wah, NJ: Erlbaum.

Pasley, K., Futris, T. G., & Skinner, M. L. (2002).Effects of commitment and psychological central-

111FATHERS’ CAREGIVING AND BREADWINNING

ity on fathering. Journal of Marriage and Fam-ily, 64, 130–138.

Pleck, J. H. (1985). Working wives, working hus-bands. Beverly Hills, CA: Sage.

Pleck, J. H. (1997). Paternal involvement: Levels,sources, and consequences. In M. E. Lamb, (Ed.),The role of the father in child development (3rded., pp. 66–103). New York: Wiley.

Rane, T. R., & McBride, B. A. (2000). Identitytheory as a guide to understanding fathers’ in-volvement with their children. Journal of FamilyIssues, 21, 347–366.

Riggs, J. M. (2005). Impressions of mothers andfathers on the periphery of child care. Psychologyof Women Quarterly, 29, 58–62.

Riley, D. (1990). Network influence on father in-volvement in childrearing. In M. Cochran, M.Larner, D. Riley, L. Gunnarsson, & C. R. Hender-

son, Jr. (Eds.), Extending families: The social net-works of parents and their children (pp. 131–153).Cambridge, England: Cambridge University Press.

Russell, G. (1983). The changing role of fathers. St.Lucia, Queensland, Australia: University ofQueensland Press.

Russell, G. (1986). Primary caretakers and role shar-ing fathers. In M. E. Lamb (Ed.), The father’s role:Applied perspectives (pp. 29–60). New York:Wiley.

Singley, S., & Hynes, K. (2005). Transitions to par-enthood: Work-family policies, gender, and thecouple context. Gender & Society, 19, 376–397.

Received January 31, 2005Revision received December 22, 2005

Accepted December 29, 2005 �

112 MAURER AND PLECK