Women's Housework: New Tests of Time and Money

• DOI: http://doi.org/10.1111/jomf.12351
• Author: Jennifer L. Hook
• Published date: 01 February 2017
• Date: 01 February 2017

J L. H University of Southern California

Women’s Housework: New Tests of Time

and Money

The author uses variation by the day of

week—comparing weekdays to weekends—to

reconsider threemain explanations for variation

in women’s housework time. The author pre-

dicts that though evidence of gender deviance

neutralization (GDN) should be evident across

the days of the week, evidence of time con-

straints and absolute earnings should be most

apparent on weekdays. The author tests these

hypotheses with the largest sample to date

(American Time Use Survey 2003–2012) and

careful consideration of the functional form

between resources/constraints and housework

time. The author nds that all three measures of

resources/constraints—relative earnings, abso-

lute earnings, and employment hours—perform

as poor predictors of women’s housework on

weekends. Weekends are when women, regard-

less of employment status, do gender, but not in

the way hypothesized by GDN. On weekdays,

women’s own employment hours and earnings

have negative, but diminishing, effects on their

housework time. GDN is not supported.

In the 1990s and early 2000s, several scholars

found that women who substantially out-earned

their husbands performed more housework

Department of Sociology, Universityof Southern

California, 851 Downey Way, Hazel Stanley Hall 314, Los

Angeles, CA 90089-1059 (hook@usc.edu).

This article was edited by Dr. Jennifer Glass.

Key Words: families and work, gender, housework/division

of labor,time use, wives’ employment, work hours.

(Bittman, England, Folbre, Sayer, & Mathe-

son, 2003 in Australia; Evertsson & Nermo,

2004 in the United States) or a greater share

of housework (Greenstein, 2000 in the United

States) than other women. That is, relative

income had a curvilinear (quadratic) relation-

ship with housework hours and share. This

nding was interpreted as evidence of gender

deviance neutralization (GDN) or compen-

satory gender display, essentially that “gender

trumps money” in some situations (Bittman

et al., 2003; Greenstein, 2000). This built on

earlier work that found husbands who were

long-term economically dependent retreated

from housework (Brines, 1994). These ndings

indicate that there are conditions under which

popular economic-based explanations, such as

bargaining models, fail to explain the division

of household labor and thus provide support

for the importance of “doing gender” (West &

Zimmerman, 1987).

Later work, however,casts doubt on this nd-

ing. Gupta (2007) examined the importance of

women’s absolute income versus her relative

income, nding that absolute income was a bet-

ter predictor of housework and nding no evi-

dence of GDN. He argued for an autonomy, or

“her money, her time,” model (see also Gupta,

2006; Gupta & Ash, 2008). Building on this

work, Killewald and Gough (2010) argued that

absolute income had diminishing returns as the

easiest and cheapest tasks to outsource or forgo

would be jettisoned as income rose, but house-

work would plateau as women found it more

difcult to nd acceptable substitutes. They

found support for this hypothesis and that when

Journal of Marriage and Family 79 (February 2017): 179–198 179

DOI:10.1111/jomf.12351

180 Journal of Marriage and Family

accounting for the relationship between income

and housework with splines, the quadratic rela-

tionship between housework and relative earn-

ings disappeared. Once Greenstein’s (2000) and

Evertsson and Nermo’s (2004) ndings with

National Survey of Families and Households

(NSFH) and Panel Study of Income Dynamics

(PSID) data were revisited by Gupta (2007) and

Killewald and Gough (2010), respectively, there

was no compelling evidence of gender display

for U.S. women. A review by Sullivan (2011)

largely concluded that the debate was over.

Prior research using the NSFH and PSID,

however,is limited in two key respects. First, the

NSFH and the PSID ask respondents to estimate

how much time they spend on housework per

week. The NSFH gauges time in specic tasks

such as cooking, washing dishes, cleaning, and

laundry, whereas the PSID does not impose a

specic denition of housework. Beyond the

difculty of accurately estimating how much

time you spend on housework during a week,

Kan (2008b) argued that evidence of neutraliz-

ing gender deviance may simply reect social

desirability bias. That is, “gender deviants” do

not perform more or less housework than others;

they simply report in ways that are consistent

with neutralizing deviance. She found support

for this hypothesis among men with gender

traditional attitudes; they perform more house-

work, according to their time diary records, than

they report in an accompanying survey.Unfortu-

nately, Kan (2008b) did not investigate women’s

housework reporting by relative earnings. Sec-

ond, the GDN hypothesis concerns the right tail

of the distribution. Previous research indicates

that the upward curve in women’s housework

corresponds to when women earn approximately

70% to 75% of couple earnings (Evertsson &

Nermo, 2004, see their Figure 3; Greenstein,

2000, see that Figure 4; Schneider, 2011, see

that Figure 1), which corresponds to about

the 90th percentile among dual earners. Thus,

sample size is important. Previous studies using

the NSFH and the PSID have analyzed around

2,000 couples (Evertsson & Nermo, 2004;

Gupta, 2007). Thus, only about 200 women are

in the tail as measured by the 90th percentile.

Given these limitations, recent evidence of

GDN using the American Time Use Study

(ATUS) is compelling. The ATUS has a larger

sample than the NSFH or PSID, and time diary

data is argued to be better than survey questions.

Using the ATUS, Schneider (2011) found a

quadratic relationship between women’s share

of earnings and housework time that is robust to

absolute income and the exclusion of outliers;

he found no evidence of a curvilinear relation-

ship for men. In a second article, Schneider

(2012) argued that occupation is a more salient

and visible marker of gender deviance than

income and found evidence of a quadratic

relationship between occupational sex compo-

sition and women’s housework. That is, women

in predominately male occupations do more

routine housework. These new ndings, with

arguably the best data on housework time, have

reopened the debate about GDN, and it is once

again a compelling explanation of women’s

housework time.

Although one group of scholars has been

deeply engaged in understanding the rela-

tionship between earnings and housework,

another has focused on time constraints, or

“gendered” time constraints (Bianchi, Milkie,

Sayer, & Robinson, 2000). The time constraints

perspective proposes that individuals pragmat-

ically respond to demands to do housework

given their availability to respond and that

partners distribute workloads toward equilib-

rium (Blood & Wolfe, 1960; Coverman, 1985).

Research generally has found support for this

approach—spousal employment hours and

the presence of children increase individuals’

housework time, and individuals’ employment

hours decrease their housework time. Time con-

straints, however, do not affect men and women

equally—women’s time is more responsive to

their own employment hours and to children

than is men’s (Bianchi et al., 2000). The per-

spective, on its own, does not explain why time

constraints are gendered.

Overall, sociologists agree that housework is

deeply gendered and that women have retained

responsibility for housework despite their

economic gains. There are three key points

of contention, however. First, do women in

gender-deviant positions compensate by per-

forming more housework than other women?

Second, are absolute earnings a better predictor

of women’s housework than relative earnings?

Third, is time more important than (relative

or absolute) money? Part of the challenge

for scholars is disentangling the relationships

among relative earnings, absolute earnings,

and hours. Using the largest sample to date

and exploiting variation by the day of week, I

carefully consider the characteristics of women