Provider Role Attitudes, Marital Satisfaction, Role Overload, and Housework: A Dyadic Approach

Abstract

Treating the marital dyad as the unit of analysis, this study examined the within-couple patterning of 272 dual-earner spouses’ provider role attitudes and their longitudinal associations with marital satisfaction, role overload, and the division of housework. Based on the congruence of husbands’ and wives’ provider role attitudes, couples were classified into one of four types: (1) main-secondary, (2) coprovider, (3) ambivalent coprovider, and (4) mismatched couples. Nearly half of all spouses differed in their attitudes about breadwinning. A series of mixed model ANCOVAs revealed significant between- and within-couple differences in human capital characteristics, spouses’ perceptions of marital satisfaction and role overload, and the division of housework across 3 years of measurement. Coprovider couples reported higher levels of marital satisfaction and a more equitable division of housework than the other couple groups. Wives in the ambivalent coprovider couples’ group reported higher levels of role overload than their husbands to a greater extent than was found in the other couple groups. As the first study to adopt a dyadic approach that considers the meanings that both spouses in dual-earner couples ascribe to paid employment, these findings advance understanding of how dual-earner spouses’ provider role attitudes serve as contexts for marital quality, behavior, and role-related stress.

From 1970 to 2001, the percentage of single-earner, breadwinner-husband families dropped from 56% to 25%, and dual-earner families emerged as the modal family arrangement in the U.S. (Raley, Mattingly, & Bianchi, 2006). Although women’s earnings account for approximately 40% of the family income on average, most wives do not assume the provider role, nor do their husbands relinquish the breadwinning role and the psychological responsibility to provide (Helms-Erikson, Tanner, Crouter, & McHale, 2000; Townsend, 2002). Drawing a distinction between the act of working at a paid job and the psychological responsibility for breadwinning, feminist scholars emphasize the meanings husbands and wives ascribe to paid work and their implications for family life (Ferree, 2010; Haas, 1986; Hood, 1986; Potuchek, 1992). For example, Hood’s qualitative work identified three distinct couple groups. Coproviders saw themselves sharing the breadwinning responsibility equally with their partners. Main-secondary provider couples viewed wives as earners of supplemental income; wives’ income helped the family, but husbands were seen as the primary providers. Ambivalent coproviders described their economic role in contradictory terms. Wives’ contributions were central, and frequently husbands could not fully support the family, but these couples still viewed wives’ breadwinning responsibility as limited. In recent reviews of the work-family literature, scholars have advocated for greater attention to factors that may better explain the links between paid work and marital and individual well-being for dual-earner couples and, in so doing, have emphasized the importance of examining the meanings husbands and wives ascribe to their employment (Bianchi & Milkie, 2010; Ferree; Perry-Jenkins & MacDermid, forthcoming; Warren, 2007).

What is the impact of dual-earner spouses’ attitudes about breadwinning on their marital quality, behavior and perceptions of role related stress? Only four studies have replicated Hood’s provider role types in some form and linked spouses’ provider role attitudes to marital and individual functioning (Helms-Erikson et al., 2000; Loscocco & Spitze, 2007; Perry-Jenkins, Seery, & Crouter, 1992; Potuchek, 1997), and no studies to date have considered the patterning (and possible incompatibilities) of husbands’ and wives’ attitudes in the same family. Results from this small body of research are promising, however, and offer support for theoretical assertions that dual-earner spouses’ provider role attitudes are closely linked to marital satisfaction, role overload and the division of housework. The current study represents a first step in understanding this important and understudied issue in the work-family literature by answering three questions: 1) To what extent do spouses in dual-earner marriages espouse similar provider role attitudes? 2) How do dimensions of human capital typically associated with breadwinning (i.e., education, earnings, hours worked per week, occupational prestige) vary based on the within-couple patterning of couples’ provider role attitudes? and 3) What implications do couples’ patterning provider role attitudes have for husbands’ and wives’ perceptions of marital satisfaction, role overload, and their division of housework?

A Dyadic Conceptualization of Provider Roles

Peplau (1983) conceptualized roles as consistent individual activity patterns composed of behavior, cognition and affect that are developed and maintained within the context of social relationships. She stressed the interdependence of roles, noting that patterns of activity develop in the context of relationships and are influenced not only by the attitudes of the individuals involved but also by cultural norms and partners’ shared relationship goals. The way in which a role is enacted may be incongruent with one’s thoughts or feelings about a role. Thus, there is potential for intra-individual discrepancies in subjective and behavioral dimensions of roles. Furthermore, when discrepancies exist between the cognitive domain of a role (i.e., attitudes or beliefs about the role) and actual role behavior, role-related stress may result and lead to negative feelings (i.e., the affective component of the role) about one’s role. Also, just as intra-individual discrepancies between the various components of a role are possible, within-couple discrepancies may also exist and reflect variation in the degree to which partners endorse particular role-related norms. Peplau suggested that consensus about roles is optimal (although not always achieved) and that within-couple incongruence is likely to result in relationship dissatisfaction, a sentiment reinforced by others theorizing specifically about provider role attitudes (Potuchek, 1992).

Aligning with Peplau’s (1983) theorizing about roles, Hood’s (1983, 1986) distinct provider role types (i.e., main-secondary, coprovider, and ambivalent coprovider) underscore variation in the meaning of work roles among dual-earner couples. Diverging from Peplau’s perspective, however, is Hood’s assumption of within-couple congruence in spouses’ attitudes about breadwinning. Although Hood did acknowledge that within-couple discrepancies in provider role attitudes could exist, her empirical research and theoretical writings focused primarily on couples in which spouses espoused similar attitudes about breadwinning. Furthermore, subsequent research, grounded in Hood’s work, has been characterized by partner-specific studies examining the provider role attitudes of either husbands or wives (e.g., Helms-Erikson et al., 2000; Perry-Jenkins & Crouter, 1990; Perry-Jenkins et al., 1992; Potucheck, 1992). And, in the one study that sampled couples, gender differences were examined at the group level rather than dyadically (Loscocco & Spitze, 2007). We suggest that couples mismatched in provider role attitudes should be examined in concert with Hood’s original “matched” typology to better understand the implications for marital relationships and role overload when spouses endorse divergent versus similar attitudes toward providing.

Provider Role Attitudes and Human Capital Investments

The limited literature that exists suggests that spouses’ provider role attitudes tend to align with their human capital investments and employment behaviors associated with breadwinning—particularly for women for whom breadwinning has historically been considered to be optional and therefore more subject to variation (Crouter & Helms-Erikson, 2000; Nock, 2001; Raley et al., 2006). For example, several studies have found that coprovider wives are more educated and earn more than wives espousing main-secondary attitudes (Helms-Erikson et al., 2000; Perry-Jenkins et al., 1992; Potuchek, 1992; 1997). Although similar patterns emerged in income and education for husbands’ provider role groups, mean differences were not statistically significant in the only study that examined variation based on husbands’ provider role attitudes (Perry-Jenkins & Crouter, 1990). The income and education of the wives in these same families did vary, however, based on their husbands’ provider role attitudes. Wives married to coprovider husbands earned more income and were more highly educated than wives married to main-secondary husbands, offering support for Peplau’s (1983) notion that both spouses play a part in how role-related cognitions are actualized. Although no studies to date have examined how human capital investments and employment behaviors covary with the patterning of spouses’ attitudes towards providing, Helms-Erikson and her colleagues did examine within-couple variations in spouses’ work-related resources based on wives’ provider role attitudes, finding that wives who espoused main-secondary provider role attitudes earned significantly less income and were less educated than their husbands and that coprovider and ambivalent coprovider wives had more similar levels of education and income than main-secondary wives.

Marital and Individual Correlates of Spouses’ Provider Role Attitudes

The scant literature linking spouses’ provider role attitudes to individual and marital functioning provides some support for theoretical suggestions that dual-earner spouses who acknowledge the importance of wives’ economic provision for their families (i.e., coproviders) are more satisfied in their marriages and divide housework more equitably than do spouses who are ambivalent about wives’ economic provision or view wives’ earnings as nonessential (e.g., Perry-Jenkins & Crouter, 1990; Perry-Jenkins et al., 1992). Variation in this general pattern of findings, however, exists based on which spouses’ provider role attitudes are assessed, suggesting that the link between provider role attitudes may be gendered for particular outcomes. Regarding the division of housework, husbands performed a higher percentage of housework in families where either husbands or wives espoused coprovider or ambivalent coprovider attitudes than when either spouse defined their breadwinning arrangement as main-secondary (Perry-Jenkins & Crouter; Perry-Jenkins et al.). Additional results from this same group of studies suggest, however, that wives’ absolute time spent in housework did not vary by their own provider role attitudes lending support for theoretical assertions that wives’ involvement in housework is less voluntary than husbands' (Helms, Walls, & Demo, 2010). Taken together, these results suggest that regardless of whether husbands or their wives embrace or are ambivalent about coproviding, husbands are more likely to share in housework than in marriages in which husbands or their wives espouse main-secondary ideals. For marital satisfaction, ambivalence may be problematic—particularly for wives. For example, Perry-Jenkins et al. found that ambivalent coprovider wives were less maritally satisfied than either coprovider or main-secondary provider wives. The authors suggested that ambivalent coprovider wives may be employed out of necessity, rather than choice or an internal breadwinning obligation, and thus resent their husbands’ inadequacies as family breadwinners. No direct links between husbands’ provider role attitudes and marital satisfaction have been found, but studies of spouses’ gender role attitudes suggest that marital satisfaction is highest when spouses’ own attitudes align with their role behavior and husbands’ and wives’ attitudes align (McHale & Crouter, 1992).

Links between provider role attitudes and spouses’ subjective evaluations of role overload have been mixed, with some findings suggesting that coprovider wives perceive lower levels of overload than other wives (Perry-Jenkins et al., 1992), and others finding no difference in spouses’ perceptions of role overload based on wives’ provider role attitudes (Helms-Erikson et al., 2000). A final set of results suggests that role overload may be most salient for spouses who are ambivalent about their provider role and that the extent to which ambivalent spouses feel overloaded may be gendered (Loscocco & Spitze, 2007). The larger work-family literature offers some support for the gendered nature of role-related stress and suggests that women report higher levels of role overload and daily stress than men (Almeida & Horn, 2004). The inconsistencies in this small body of research may be a result of the partner-specific approaches that do not allow for the examination of spouse by provider role group interactions which would reveal gendered patterns of role overload in particular marital contexts.

Research Goals and Hypotheses

A more holistic and complex understanding of how spouses’ provider role attitudes serve as contexts for their perceptions of marital satisfaction, role overload, and the division of housework requires analytic strategies that enable one to consider the meanings that both spouses in dual-earner couples ascribe to paid employment. After identifying four groups of couples based on the congruence of their coded responses on attitudes towards breadwinning (i.e., mismatched couples and matched coprovider, ambivalent coprovider, and main-secondary couples), the first goal of the study was to describe couple group differences on human capital characteristics (i.e., education, earnings, hours worked per week, occupational prestige). Our second goal was to examine the link between couple group membership and spouses’ reports of marital satisfaction, role overload, and the division of housework across a three-year period.

Drawing from both theoretical and empirical literatures, we formulated several hypotheses. Based on theoretical assertions, we expected husbands and wives in coprovider partnerships who evidenced both intra-individual and within-couple congruence to report the highest levels of marital satisfaction, comparatively. Moreover, we hypothesized that main-secondary couples who shared provider role attitudes and were not ambivalent about their roles as breadwinners to report higher levels of marital satisfaction than mismatched couples whose attitudes did not align. Furthermore, empirical findings linking wives’ ambivalence about providing to lower levels of marital satisfaction led us to expect that, similar to mismatched couples, ambivalent coprovider couples who were internally incongruent would report lower levels of marital satisfaction than either coprovider or main-secondary couples. Findings from the broader work-family literature and theoretical links between intraindividual incongruence and role-related stress led us to expect a gendered pattern of role overload evidenced by wives reporting significantly greater overload than their husbands. We expected this pattern, however, to be most pronounced for couples in which both spouses had ambivalent attitudes about breadwinning. Finally, based on prior research, we expected coprovider couples to divide housework more equitably than other provider role groups. In addition, we hypothesized that ambivalent coprovider couples would share tasks more equitably than those couples who defined their roles as main-secondary or who were mismatched in their provider role attitudes. We used longitudinal data to capitalize on the statistical power derived from multiple occasions of measurement, but, because past empirical work has found relatively stable effects for associations between spouses’ gendered attitudes and personal and marital well-being over relatively short durations of time in the middle years of marriage (Helms, Proulx, Klute, McHale, & Crouter, 2006), we expected to find a consistent pattern of results across time.

Methods

Participants

The sample was drawn from the first 3 phases of a longitudinal study of family relationships in dual-earner households. Four hundred families participated at Time 1, 203 of which had 2 preteen children, and 197 of which had 2 adolescent children. Families were recruited via letters sent home to parents of fourth and fifth or ninth and tenth grade students in 13 rural and small urban school districts in central Pennsylvania. The letter to families described the research effort in general terms, indicated that families would receive a $100 honorarium for each phase of participation and outlined eligibility criteria which included parents being married, employed, and having a second-born child approximately 1 to 3 years younger than the targeted child. Eligible families were invited to return a postcard if interested in participating. Over 90% of those families who returned the card and met the criteria agreed to participate. Analysis of census data along with school enrollment information from the area indicated that 11% of the school population met the study criteria. Dividing the number of families recruited by the number of families estimated to be eligible yielded a response rate of 34% for the participation of four members of each family. This response rate is comparable to the National Survey of Families and Households rate of 37% for only three family members. By Time 3, 6 families declined participation resulting in a 98.5% retention rate across three years of measurement.

Of the 394 families for whom data were available at all three phases, there were 272 couples in which both husbands and wives were employed at least part-time (i.e., 20 hours per week or more) across the three years of measurement – a requirement of the current study. Most spouses were employed full-time; husbands averaged 48 and wives 38 work hours per week, respectively. The 272 participating couples were primarily White (98%), working and middle class, and resided in rural areas (49%), small towns (31%), and small cities (20%). The median family income was $58,000 at Time 1 (1995–1997), and couples had been married an average of 16 years. Husbands and wives were 41 and 38 years old on average, respectively, and both husbands and wives averaged 2 years of education beyond high school. Comparisons of our sample with U.S. Census data on families from the same geographic region suggested that the couples in our sample were representative of the population with two exceptions: spouses were slightly older (2 years) and were more likely to have some post-high school education.

Procedures

The following procedures conformed to the requirements of the Institutional Review Board at the study’s home institution. During each timepoint, data were collected using two procedures. First, husbands and wives were interviewed separately during home interviews that averaged 2 to 3 hours in duration. Prior to interviewing, the study was described in general terms and consent was obtained. Spouses then responded to a series of survey and open-ended questions regarding their work circumstances, provider role attitudes, family relationships and mental health. Second, during the 2 to 3 weeks following the home interview, couples completed seven telephone interviews (five weeknights and two weekend nights). Each spouse was interviewed four times during the course of seven calls (one weeknight call included both spouses). The telephone interviews provided information on (a) the frequency of and time spent on over 50 activities during the day of the call and (b) companions in spouses’ daily activities.

Measures

Provider role attitudes

Derived from Hood’s (1986) recommendations for measuring breadwinning attitudes, Perry-Jenkins et al.’s (1992) interview and coding scheme was used to assess spouses’ provider role attitudes and categorize each spouse’s responses into one of three provider role types (e.g., main-secondary, coprovider, ambivalent coprovider). In separate interviews, husbands and wives were asked a variety of structured and open-ended questions representing global attitudes (e.g., “In general, how important are the roles of parent, spouse, worker, provider/breadwinner for the family, maintainer of the home for women/men in families?), (b) family-specific attitudes (e.g., “With reference to your own family, who do you feel should provide the income?”), and (c) reports of how the wife’s income is actually used (e.g., main provider, wife’s income helps to pay monthly bills, or wife’s income used for “extras” or what she wants). In addition, spouses were asked to share their views about mothers’ employment and the extent to which mothers should be (as compared to fathers) responsible for providing for their families. Two independent raters coded each separate response as representing main-secondary, coprovider, or ambivalent coprovider attitudes; raters coded husbands’ and wives’ responses separately. Final codes for each spouse were derived by examining the pattern of codes across responses and, in cases where inconsistencies were noted, narrative responses were used for clarification. Cohen’s kappa for interrater reliability was .73 indicating that coders were very reliable. Forty-seven percent of wives and 50% of husbands were coded as main-secondary providers (e.g., “A wife should be able to contribute to the family [financially], but I still have to think the husband should be more the provider.”), 34% of wives and 22% of husbands were categorized as coproviders (e.g., “I have a profession. I work. It is important to work to contribute to the family”), and 19% of wives and 28% of husbands were categorized as ambivalent coproviders (e.g., “If there is money pressure, both should work. Nurturing of children cannot be done right by switching male and female roles.”)

Spouses’ gender-typed attitudes

Spouses’ gender-typed attitudes were collected during home interviews at Time 1 and 3 using Spence and Helmreich’s (1972) Attitudes Toward Women Scale. On this 15-item questionnaire, spouses indicate the extent to which they agreed with a variety of statements regarding women’s roles in society (e.g., “If both a husband and wife are working outside the home, they should share equally in the routine household chores, such as washing dishes and doing laundry.”). Response options ranged from 1 (strongly agree) to 4 (strongly disagree), and the data were coded so that high scores indicated more conventional attitudes. Cronbach’s alphas ranged from .71 and .80.

Human capital characteristics

At Time 1, spouses reported the years of education they had completed. Occupation, gross yearly earnings, and work hours were reported annually. Occupational prestige was classified at Time 1 according to the National Opinion Research Council (Nakao & Treas, 1994). Spouses’ primary occupations ranged from 20.05 (maid) to 86.05 (physician), with average occupational prestige scores of 50.24 and 47.86 for wives and husbands, respectively. Positions in the middle range included sales representatives, skilled laborers (e.g., carpenters and electricians), secretaries, and office supervisors.

Marital satisfaction

Perceptions of marital satisfaction were assessed annually via an adaptation of The Aspects of Married Life Questionnaire (Helms, Crouter, & McHale, 2003). Spouses rated their satisfaction with seven domains of married life (e.g., marital communication and decision-making) using a 9-point scale (1 = very dissatisfied; 9 = very satisfied). Higher scores indicated higher levels of marital satisfaction. Cronbach’s alphas ranged from .86 to .89.

Role overload

At each phase of the study, husbands and wives completed Reilly’s (1982) 13-item Role Overload Scale. Using a scale ranging from 1 (strongly agree) to 5 (strongly disagree), spouses indicated the extent to which they felt overwhelmed by multiple role commitments and not having enough time for themselves (e.g., “I can't ever seem to get caught up”). Cronbach's alphas ranged from .87 to .90, and higher scores indicated greater role overload.

Division of housework

Data on the division of housework were collected at each time point during individual telephone interviews using a cued-recall procedure. Husbands and wives were given activity lists at the end of each home interview that they then used for reporting a variety of activities, including household tasks. The current investigation focused on the repetitive, time-consuming tasks traditionally performed by women, typically labeled housework. During each of four calls, spouses were asked (a) how often they performed four traditionally feminine household tasks (i.e., do dishes, prepare a meal or snack, laundry, and vacuum/clean/straighten up), (b) how long each activity lasted (in minutes), and (c) who else had participated in the activity. The division of housework was operationalized as the percentage of total time spent by the couple on traditionally feminine household tasks that was performed by the wife. The telephone interview sequence involved one telephone call when both spouses were interviewed on the same evening. Husbands' and wives' reports of joint household activities from this call were correlated at .93 at Time 1, indicating that the measure is very reliable.

Results

Four groups of couples were observed in the data: (1) a main-secondary group in which both spouses espoused main-secondary provider role attitudes (n = 94); 2) a coprovider group in which both spouses held coprovider attitudes (n = 32); 3) an ambivalent coprovider group was made up of 19 spouses who shared ambivalent views about their coprovider arrangement; and (4) a mismatched group in which spouses had discrepant provider role attitudes (n = 127). A preliminary set of analyses further validated the provider role group memberships and showed that spouses’ gender-stereotypic attitudes (measured only at Time 1 and 3) were highest in the main-secondary group and lowest in the coprovider group as evidenced by a significant Between Groups effect, F(3, 272) = 29.70, p < .0001 and Tukey post-hocs. (See Table 1). A significant overall Spouse effect revealed that regardless of group membership husbands espoused more gender stereotypic attitudes than their wives, F(1, 272) = 22.17, p < .0001.

Table 1.

Means and (Standard Deviations) for Human Capital and Attitudinal Characteristics by Provider Role Group

	Main-secondary (n = 94 couples)	Coprovider (n = 32 couples)	Ambivalent coprovider (n = 19 couples)	Mismatched (n = 127 couples)
Education
Wives	13.81 (1.90)	16.19 (2.16)	15.16 (1.98)	14.66 (2.21)
Husbands	14.07 (2.15)	15.59 (2.03)	15.05 (2.41)	14.06 (2.35)
Couple a	13.94b (1.77)	15.89a (1.94)	15.10a,c (1.93)	14.36b,c (1.96)
H –W b	.27b (1.97)	−.59a (1.60)	−.10a,b (2.16)	−.61a (2.33)
Income
Wives	17,787 (10,438)	36,126 (20,634)	26,819 (11,900)	26,796 (12,472)
Husbands	41,666 (17,704)	36,838 (14,264)	55,568 (45,292)	36,087 (16,335)
Couple a	29,416a (9,909)	36,482b (13,909)	41,193b (20,229)	31,406a (10,893)
H –W b	23,257b (17,981)	712a (22,014)	28,750b (52,432)	9,363a (19,383)
Work Hours
Wives	33.97 (8.43)	40.17 (7.05)	37.74 (8.55)	39.67 (6.30)
Husbands	48.69 (9.16)	45.64 (8.36)	48.18 (5.41)	47.64 (9.21)
Couple a	41.33b (5.82)	42.90a,b (6.03)	42.96a,b (5.69)	43.67a (5.36)
H –W b	14.72b (13.21)	5.47a (9.67)	10.44a,b (8.69)	7.99a (11.65)
Gender-Role Attitudes
Wives	27.68 (5.10)	20.35 (4.24)	23.24 (3.00)	25.21 (5.00)
Husbands	29.33 (5.24)	22.65 (4.43)	25.71 (4.45)	27.97 (4.65)
Couple a	28.50b (3.90)	21.50c (3.73)	24.47a (3.02)	26.60a (3.75)
H –W b	1.66 (6.78)	2.29 (4.44)	2.47 (4.60)	2.76 (6.10)

Note. With the exception of education, scores for husbands and wives are averaged across 3 points of measurement. Group differences for significant effects determined by Tukey HSD tests are indicated by means in the same row that do not share subscripts.

^aScores represent the average of husbands’ and wives’ reports.

^bScores represent husbands’ –wives’ difference.

Our first research goal was to describe group differences in human capital characteristics (i.e., education, earnings, hours worked per week, occupational prestige). To address this goal we conducted a series of 4 (Group) X 2 (Spouse) X 3 (Time) mixed model ANOVAs, treating spouse and time (when applicable) as within-group factors. Sample sizes varied slightly due to missing data on the outcome of interest. Because cell sizes were unequal, we examined Type III sums of squares. Significant univariate findings were followed up using the Tukey HSD test.

Several findings emerged from these descriptive analyses that helped further differentiate the groups (Table 1). First, for education, a significant Between Groups effect for couple group, F(3, 272) = 9.30, p < .0001, showed that coprovider couples were more educated than main-secondary or mismatched couples, but did not differ from ambivalent coprovider couples. Furthermore, ambivalent coprovider couples reported higher levels of education than main-secondary couples. A significant Spouse X Couples Group interaction showed that wives in the main-secondary group were less educated than their husbands, whereas wives in the coprovider and mismatched groups were more educated than their husbands, F(3, 272) = 3.35, p < .02.

Similar to the results for education, a significant Between Groups effect was found for income, F(3, 260) = 6.78, p < .001, indicating that spouses in both the coprovider and ambivalent coprovider groups had higher combined incomes than the main-secondary and mismatched groups (averaged across the three years of measurement). A significant overall Spouse effect, F(1, 260) = 68.71, p < .0001, showed that husbands earned higher incomes than wives, but this finding was qualified by a Spouse X Group interaction, F(3, 260) = 12.26, p < .0001, indicating that this gendered pattern was most apparent in the main-secondary and ambivalent coprovider groups. The findings for income did not coincide with those found for work hours. A significant Between Groups effect, F(3, 271) = 3.13, p < .05 showed that main-secondary spouses worked fewer hours per week than the mismatched couples group. A significant overall Spouse effect, however, was similar to the results found for income in that husbands averaged higher work hours than their wives, F(1, 271) = 103.78, p < .0001. A Spouse X Group interaction, F(3, 271) = 7.74, p < .0001, indicated, however, that main-secondary couples had a larger gap between their work hours (favoring husbands) than either coprovider or mismatched couples. A fourth set of findings showed an overall Between Groups effect for occupational prestige suggesting that spouses in the coprovider group were employed in higher prestige jobs (M = 54.08, SD = 13.68 for wives; M = 53.20, SD = 11.49 for husbands) than either main-secondary (M = 48.08, SD = 12.55 for wives; M = 47.32, SD = 11.49 for husbands) or mismatched couples (M = 50.65, SD = 12.21 for wives; M = 46.51, SD = 12.88 for husbands), F(3, 270) = 3.48, p < .02. Taken together, these findings suggested that mismatched and main-secondary couples worked in jobs lower in occupational prestige and at a lower pay rate than their coprovider counterparts.

Our second goal was to examine marital satisfaction, role overload, and the division of housework over a three-year period for the four couple groups. First, we used a series of 4 (Group) X 2 (Spouse) X 3 (Time) mixed model ANCOVAs with spouses’ reports of marital satisfaction and role overload as dependent variables. To examine group differences in couples’ division of housework, we conducted a 4 (Group) X 3 (Time) mixed model ANCOVA. We treated spouses’ education, an indicator of social class, as a covariate to control for possible social class difference in our outcomes of interest. (See Table 2.) The Benjamini-Hochberg FDR method was used to test hypothesized differences between groups for significant effects.

Table 2.

Means and (Standard Deviations) for Marital Satisfaction and Role Overload by Provider Role Group

	Main-secondary (n = 94 couples)	Coprovider (n = 32 couples)	Ambivalent coprovider (n = 19 couples)	Mismatched (n = 127 couples)
Satisfaction
Wives	46.57 (8.82)	51.84 (6.41)	46.23 (7.89)	45.76 (9.56)
Husbands	48.62 (8.39)	50.66 (7.39)	49.19 (7.97)	49.14 (7.58)
Couple a	47.54a (6.91)	51.37b (5.73)	47.71a,b (6.68)	47.45a (7.55)
H –W b	1.97 (10.31)	−1.43 (7.79)	2.96 (8.55)	3.38 (8.36)
Role Overload
Wives	44.61 (7.74)	45.81 (6.85)	47.72 (7.65)	46.06 (8.24)
Husbands	41.82 (8.92)	42.09 (8.77)	37.40 (8.68)	42.01 (7.73)
Couple a	43.21 (6.65)	43.95 (5.76)	42.56 (6.24)	44.03 (6.13)
H –W b	−2.79a (10.11)	−3.72a (10.71)	−10.32b (10.58)	−4.05a (10.25)

Note. Because no effects for time emerged, all scores for husbands and wives are averaged across the 3 points of measurement for ease of presentation. Group differences for significant effects determined by the Benjamini-Hochberg FDR are indicated by means in the same row that do not share subscripts.

^aScores represent the average of husbands’ and wives’ reports.

^bScores represent husbands’ –wives’ difference.

A between-group effect was found for marital satisfaction, F(3, 267) = 2.68, p < .05. As hypothesized, coprovider couples consistently reported higher levels of marital satisfaction than main-secondary, ambivalent coprovider, and mismatched couples. Contrary to our hypothesis, however, main-secondary couples did not differ from ambivalent coprovider or mismatched couples in reports of marital satisfaction. Further, husbands averaged higher levels of marital satisfaction than wives as indicated by a significant Spouse effect, F(1, 267) = 5.27, p < .05.

Whereas coprovider couples emerged as different from other groups in marital satisfaction, spouses in the ambivalent coprovider group were unique in their reports of role overload. Although, overall, wives reported higher role overload than their husbands, F(1, 272) = 38.85, p < .0001, a significant Spouse X Group interaction, F(3, 272) = 2.72, p < .05, revealed a significantly larger gap (i.e., wives reported higher role overload than their husbands) in role overload for ambivalent coprovider couples, than was found for the other couple groups.

For our final set of analyses, a significant Between Groups effect emerged for the division of housework, F(3, 268) = 17.25, p < .0001. As hypothesized, coprovider couples were significantly different than all other groups and demonstrated the most egalitarian division of housework, with wives completing 59% of the housework on average over the three points of measurement. In contrast to our hypothesis, however, the division of housework was not more equitable for ambivalent coprovider couples than main-secondary or mismatched couples. Indeed, ambivalent coprovider couples and main-secondary couples did not differ from one another and had the most “traditional” divisions of housework with wives completing an average of 81% of the housework for both groups. For mismatched couples, wives completed 73% of the housework and differed significantly from all other couple groups.

Post-hoc Analyses

Because neither previous theorizing or empirical work led us to make predictions regarding the probability of specific types of mismatched couples nor how particular “mismatches” might be related to marital relationships and role overload, all possible configurations of mismatched couples were treated as single group in the substantive analyses reported above. Our provider role typology, however, allowed for six possible mismatched couple configurations, and each was represented in our data (i.e., main-secondary wives and coprovider husbands, n = 14; main-secondary wives and ambivalent coprovider husbands, n = 20; coprovider wives and main-secondary husbands, n = 23; coprovider wives and ambivalent coprovider husbands, n = 37; ambivalent coprovider wives and main-secondary husbands, n = 20; ambivalent coprovider wives and coprovider husbands, n = 13). To explore possible differences, we ran a series of post-hoc 3 (Wives’ Provider Role) X 3 (Husbands’ provider role) X 2 (Spouse) X 3 (Time) mixed model ANCOVAs treating spouse (when applicable) and time as repeated factors and education as a covariate. Results (not shown) from these exploratory post-hoc analyses yielded no significant findings between the six groups. It should be noted, however, that cell sizes were small, potentially limiting our ability to detect effects.

To further probe for potential within group differences for the mismatched couples group, we created subgroups based on two different grouping strategies (i.e., subgroups based on husbands’ provider role attitudes and subgroups based on wives’ provider role attitudes) which each yielded three, mismatched groups, all with adequate cell sizes. Groups based on husbands’ provider role attitudes were distributed as follows: n = 43 for main-secondary husbands married to either coprovider or ambivalent coprovider wives; n = 27 for coprovider husbands and either main-secondary or ambivalent coprovider wives; and n = 57 for ambivalent coprovider husbands and either main-secondary or coprovider wives. Groups based on wives’ provider role attitudes were: n = 34 for main-secondary wives and coprovider or ambivalent coprovider husbands; n = 60 for coprovider wives and main-secondary or ambivalent coprovider husbands; and n = 33 for ambivalent coprovider wives and either main-secondary or coprovider husbands. Next we conducted a series of 3 (Group) X 2 (Spouse) X 3 (Time) mixed model ANCOVAs for marital satisfaction, role overload, and the division of housework separately for the groups defined by husbands and wives. Because these analyses were exploratory, the Tukey HSD test was used to examine group differences. No significant findings emerged from the analyses with mismatched groups based on husbands’ provider role attitudes. In the analyses based on wives’ provider role attitudes, we found a significant Between Groups effect for the division of housework indicating that main-secondary wives in mismatched marriages completed a greater share of housework (i.e., 80%) than coprovider wives in mismatched marriages (72%), F(2, 122) = 3.03, p < .05.

Discussion

The current study advanced our understanding of how spouses’ provider role attitudes serve as contexts for marital satisfaction, role overload, and the division of housework by adopting a dyadic approach that considers the meanings that both spouses in dual-earner couples ascribe to paid employment. This study is the first to build upon the recommendations of work-family scholars (Bianchi & Milkie, 2010; Ferree, 2010; Perry-Jenkins & MacDermid, forthcoming) and move beyond the limitations of earlier partner-specific approaches to better align the dyadic focus of theoretical work with an analytic strategy that captured both spouses’ attitudes toward breadwinning and their links with marital experiences and role-related stress. Furthermore, our dyadic approach captured the inherent complexity of roles (Peplau, 1983) by considering both within-person and within-couple discrepancies in the provider role-attitudes of dual-earner spouses and provided support for previously untested assumptions that the patterning of dual-earner spouses’ provider role attitudes has implications for marital satisfaction, role overload, and the division of housework. Notably, the greatest marital satisfaction and the most equitable division of housework were reported by spouses in the coprovider group for whom partners’ attitudes about breadwinning matched their own employment behavior and one another’s attitudes. Furthermore, ambivalent coprovider couples had the largest gap in spouses’ reports of role overload suggesting that wives in this group experienced higher levels of role-related stress than their husbands to a degree not seen in the other couple groups. These patterns of findings did not vary across the three years of measurement.

We began our investigation by grouping couples based on the match in husbands’ and wives’ provider role attitudes and identified 34% of couples as main-secondary, 12% as coproviders, 7% as ambivalent coproviders, and 47% as mismatched. Results from other studies in which dual-earner husbands’ and wives’ provider role attitudes were examined separately showed that approximately 45% of husbands and wives endorsed coprovider attitudes, whereas approximately one third were ambivalent coproviders, and main-secondary providers represented less than 25% of dual-earner spouses (Loscocco & Spitze, 2007; Perry-Jenkins & Crouter, 1990; Perry-Jenkins et. al., 1992). As the first study to examine the within-couple patterning of provider role attitudes, our results suggest that coprovider couples may be much less prevalent than previously assumed based on studies of individual spouses. In addition, given both theoretical and empirical assumptions regarding within-couple congruence in attitudes toward breadwinning that have characterized the literature to date (Hood, 1983; 1986; Perry-Jenkins et al.), we find it noteworthy that nearly half of the dual-earner couples in our study did not share attitudes about breadwinning. These findings support speculations that in periods of widespread social change in gendered employment patterns, differential rates of change in attitudinal dimensions of breadwinning may result in divergent views of wives’ involvement in paid employment within couples (Ferree, 2010; Nock, 2001; Sullivan, 2004). Furthermore, post-hoc analyses revealed that the largest subgroup of mismatched couples in our sample was comprised of coprovider wives married to either main-secondary or ambivalent coprovider husbands, suggesting that contemporary, dual-earner wives’ provider role attitudes may be more likely to align with their role-related behavior than is the case for their husbands who may feel ambivalent about their wives’ economic provision or prefer an arrangement in which the husband serves as the primary breadwinner. Consistent with earlier work in which contemporary women were found to be much more likely to reject traditional notions of gendered behavior in families than were similar cohorts of men, these findings underscore the “predictably inconsistent” patterns of gendered relations within couples characterized by feminist scholars as a “stalled revolution” or “slow drip” of change (Ferree; Sullivan). Accordingly, an important avenue to pursue in future work is how within-couple patterns in provider role attitudes change over longer stretches of time and covary with changes in spouses’ human capital and gendered behaviors in the home (Sullivan).

To address our first research goal, we examined the human capital characteristics of the four couple types. Findings from these analyses helped further describe the couple groups and enhanced our understanding how spouses’ attitudes about breadwinning align (or fail to align) with their employment behaviors and human capital investments. The pattern of findings from these analyses coincided with findings from previous, partner-specific studies and suggested that main-secondary and coprovider couples’ human capital and employment characteristics align with their provider role attitudes (Helms-Erikson et al., 2000; Perry-Jenkins & Crouter, 1990; Perry-Jenkins et al., 1992; Potuchek, 1992; 1997). Main-secondary couples’ provider role attitudes were consistent with their human capital investments and work-related characteristics in that wives were less educated than their husbands, were employed fewer hours per week, and had significantly lower incomes than their husbands. Husband and wives in the coprovider group, in contrast, earned similar incomes and had similar work hours. Furthermore, ambivalent coproviders’ intra-individual discrepancies regarding attitudes towards breadwinning reflected discrepancies in human capital characteristics. For example, ambivalent coproviders, who were more educated than their main-secondary counterparts and no different than coprovider couples in education, were characterized by wives with higher levels of education than their husbands and a gendered within-couple pattern of income similar to main-secondary couples (i.e., husbands earning significantly more money than their wives).

For our second research goal, we first examined couple group differences in marital satisfaction over a three-year period of time. As hypothesized, spouses in the coprovider couples group reported higher levels of marital satisfaction than mismatched, main-secondary, and ambivalent couples. Our expectation that main-secondary couples who matched in provider role attitudes would be more satisfied than either mismatched couples or matched spouses who were ambivalent about their role was not supported. Taken together, these findings advance theoretical assumptions that within-couple congruence in roles is optimal for marital satisfaction in that they suggest that the type of match is of great importance. That is, for dual-earner couples, marital quality is better when both spouses share coprovider ideologies that acknowledge the contributions of wives’ employment to family breadwinning. This finding is important in that it clarifies results from partner-specific studies of dual-earner wives’ provider role attitudes and further suggests that a marital context in which both spouses endorse coprovider attitudes is optimal for both husbands’ and wives’ marital quality (e.g., Perry-Jenkins & Crouter, 1990; Perry-Jenkins et al., 1992). Furthermore, our results provide an important caveat to a small but growing body of work in which shared breadwinning--operationalized as a relatively equal contribution to the family income within couples--is viewed as a marital risk (Nock, 2001).

Similar to the pattern of results for marital satisfaction and as hypothesized, coprovider couples demonstrated the most equitable division of housework suggesting that equality in the home is most likely to be realized for dual-earner couples in which shared breadwinning is embraced ideologically by both partners. Ambivalent coproviders, however, who were hypothesized to demonstrate a more equal division of labor than main-secondary or mismatched couples based on prior work (Perry-Jenkins et al., 1992), were remarkably traditional in their division of housework and did not differ from main-secondary couples who were ideologically more traditional. Instead, mismatched couples, comprised of a large subgroup of coprovider wives married to either main-secondary or ambivalent coprovider husbands, divided tasks more equitably than either main-secondary or ambivalent coprovider couples. Results from the post-hoc analyses suggest that wives’ attitudes toward breadwinning may be particularly important for the division of housework for couples in which spouses are ideologically incongruent. Although unable to achieve the more equitable division of housework demonstrated by matched coprovider spouses, coprovider wives married to men who were either ambivalent about wives’ employment or preferred a main-secondary arrangement, appeared to have somewhat of an advantage over their main-secondary counterparts in their divisions of housework with their husbands. As others have suggested, it may be that coprovider wives feel entitled to equality and thus demand greater involvement from their husbands who, in turn, may change their housework behavior regardless of their more traditional ideological preferences (Ferree, 2010; Sullivan, 2004).

The inconsistent findings from previous studies regarding the links between provider role attitudes and role-related stress (Helms-Erikson et al., 2000; Hood, 1983; Loscocco & Spitze, 2007; Perry-Jenkins et al., 1992) are further clarified by our dyadic approach which suggests that it may be within-couple patterns in reports of role overload that vary based on provider role group membership rather than simple between-group differences. Specifically, our results showed a gendered pattern in reports of role overload over time (i.e., wives reporting more role overload than their husbands) that was most pronounced for ambivalent coprovider couples. Although this finding should be interpreted tentatively given the small number of ambivalent coprovider couples in our study, results suggest that in a marital context in which both spouses exhibit ambivalent provider role attitudes, wives may feel significantly more burdened and overwhelmed than their husbands, and the extent of this burden is significantly greater than that experienced by wives in other marital provider role configurations. In interpreting this finding, it is important to note that wives in the ambivalent coprovider group were more educated than their husbands but made significantly less money than them and were responsible for more housework (i.e., 81%). It may be that for these couples, role ambivalence results in wives’ greater responsibility for housework by default, and this disproportionate burden is not offset by strongly endorsed and mutually agreed upon views regarding a gendered division of roles in the home or reduced work hours. In this context, these educated and employed wives feel over-burdened with the demands of work and family to an extent not experienced by their husbands.

In interpreting these findings it is important to note several caveats. First, the homogeneity of our sample limits the generalizability of our findings to other groups for whom variations are likely to exist based on geography and sociohistorical factors related to employment patterns and family roles, structural inequalities, cultural prescriptions regarding gender-specific behavior, or immigration history. Regardless of the population under study, however, the dyadic approach adopted in this study, will be important for understanding both between and within-couple differences in the patterning of provider role attitudes and is essential for challenging conceptual models that assume family solidarity (Ferree, 2010; Perry-Jenkins, Repetti, & Crouter, 2000). Second, recent research suggests that the patterning of spouses’ gendered attitudes is likely to create contexts for processes (e.g., displays of warmth) that play an important role in predicting marital and personal well-being (Miller, Caughlin, & Huston, 2003). Longitudinal research focusing on marital processes that potentially mediate the association between spouses’ provider role attitudes and marital quality, behavior, or personal well-being will be important to examine. Finally, this work underscores the continued heterogeneity that exists both between and within dual-earner couples and individuals despite structural level changes in women’s employment (Ferree; Sullivan, 2004). Our work speaks specifically to the importance of the attitudinal dimension of breadwinning which may or may not align within couples or with individual spouses’ employment behavior. Although our work supports and extends earlier theorizing (Peplau, 1983) in suggesting that a combination of within-person and within-couple consensus in provider role attitudes may promote marital harmony, it challenges the assumption in most prior work that the most dual-earner couples have reached such a consensus (Hood, 1983; 1986). Intra-individual ambivalence between attitudes and employment behaviors and within-couple discrepancies in provider role attitudes appears to matter, particularly for wives.