Help to Family and Friends: Are There Gender Differences at Older Ages?

Abstract

This paper uses recent data from the Wisconsin Longitudinal Study (N = 5,220) to explore gender differences in the extent to which adults in their 50s and 60s provide informal help to their adult children, elderly parents and friends We find that both men and women report very high levels of helping kin and nonkin alike, though women do more to assist elderly parents and women provide much more emotional support to others than do men. Men provide more assistance than do women with “housework, yard work and repairs.” As they retire from the workforce, married men become significantly more involved in the care of their grandchildren, virtually eliminating any gender difference by the time they are in their 60s.

Despite considerable changes in men’s and women’s work and family roles in recent decades, men still spend more hours in the paid workforce than women, and women perform more unpaid labor than men. Studies have shown repeatedly that compared to men, women do more housework, childcare, and care of kin than men (Bianchi, Robinson, & Milkie, 2006; Gerstel, 2000; Hochschild, 1989). However, most studies focus on adults in their prime working ages when work-family conflicts are greatest.

By the time men and women reach retirement age, women have typically spent 20 or 30 years meeting children’s day to day needs, including emotional needs. Men have concentrated on providing financially for family members, with their interaction with their children and other kin often mediated through their wife (Lareau, 2000). If these early life patterns endure, older women will be more heavily engaged than older men in helping other relatives such as aging parents and grandchildren.

Yet, the gender gap in providing assistance to kin is likely to vary across the life course, reflecting the changing demands of work and family life (Rexroat & Shehan, 1987; Solomon, Acock, & Walker, 2004; Szinovacz, 2000). On one hand, as children age into adulthood and leave the parental home, housework and childcare demands are greatly reduced for both men and women, though probably much more so for women. On the other hand, at retirement, men’s need or desire to reorient their time toward other meaningful activities may actually be greater than for women because they invested so heavily in paid work earlier in life. One area of meaningful activity is helping family members and rekindling connections with friends.

The present paper examines changes in the gender gap in informal helping behaviors for a cohort of U.S. adults as they age from their 50s into their mid-60s. During this period, many “empty the nest,” retire from full-time employment, and make new decisions about how to spend their time. We look specifically at gender differences in the extent to which adults report helping others, both within and outside of their families. Our measures of “help work” incorporate activities ranging from providing practical support such as transportation or help with housework and yard work, to caring for grandchildren, to providing emotional support to family and friends. The importance of helping others in later life, particularly others outside one’s own household, derives in part from the social support networks that such behaviors help establish and maintain (Bengtson, 2001; Swartz, 2009). By caring for their aging parents, older adults demonstrate to their children, their own expectations for the obligations one should have to kin (Cox and Stark, 2005). By assisting adult children, parents potentially obligate those children to help them in the future when they need assistance. And, the help they give to friends and neighbors solidifies social support networks later in life, particularly when children live far away or are focused on their own families and unavailable to provide assistance.

In this study, we explore gender differences in providing help to others. We do not seek to evaluate the gender specialization within the home; we know from other research that older women do more housework within their own home than older men (Szinovacz, 2000). Nor do we focus on the intensive caregiving for a disabled spouse or parent that often entails coresidence and a large time commitment. Here also, the extant literature suggests that women do more of this type of care (Bengtson, Rosenthal, & Burton, 1990). Instead, we focus on something in the middle – the helping behaviors that connect adults to family members and friends outside their own household. Men as well as women have powerful incentives to participate in these types of activities because they counter social isolation and build the social networks that can be activated in times of need. Older married men may rely on their wives to orchestrate these opportunities to help others (Lareau 2000), but no matter who initiates these activities, the mere act of doing them connects older men and women to others in a web of exchanges and likely provides them with a sense of purpose in life.

BACKGROUND

With improvements in health and life expectancy, many adults are enjoying more years of vitality prior to experiencing the serious health declines of old age. During this new “Third Age” of the life course, defined by Moen and Spencer (2006) as the 50s, 60s, and 70s, healthy adults face new opportunities for ‘post-career’ productive activities including part-time employment, volunteering, and informal support to family and friends. Some researchers predict a ‘blurring’ of gender differences during the Third Age, reflecting the growing heterogeneity in men’s and women’s work and family roles, behaviors, and relationships throughout the life course (Moen & Spencer, 2006). Men and women may develop more androgynous traits and qualities later in life that may have been suppressed earlier in the life course, when men were more focused on achieving outside the home and women on nurturing within the home (Gutmann, 1987; James & Lewkowicz, 1997; Sorensen, 1991). Moreover, with the dramatic increases in women’s labor force activity in recent decades, women’s and men’s life course patterns have become more similar.

The few studies that have focused on age and gender differences in housework have shown a narrowing of the gender gap at older ages, with older husbands doing more housework than younger ones and older wives doing less (Gauthier & Smeeding, 2003; Rexroat & Shehan, 1987). But these studies based on cross-sectional data are unable to test whether patterns actually change as individuals age towards retirement. Two studies that followed the same individuals through the retirement transition reported mixed results: one found that men did more housework after retiring (Szinovacz, 2000), whereas another found that men did no more housework after retirement than before (Solomon et al., 2004).

Findings on the gender gap in other forms of unpaid labor, often conceptualized as “care work” or “informal support,” have also been mixed. Gerstel (2000) showed that women did considerably more “care work” for both kin and nonkin than did men, amounting to what she refers to as a “third shift” of labor. Gerstel and colleagues also showed that the gender gap in informal caregiving was closely related to employment status and job characteristics (Gerstel & Gallagher, 1994, 2001; Sarkisian & Gerstel, 2004). Hook (2004), using Australian time use data, found that women spent more time than men providing informal support and volunteering, but the overall time in these activities was relatively small and did not amount to a burdensome “third shift” for women. Chesley and Poppie (2009) found persistent gender differences in emotional support to parents and in-laws, but no gender differences in unpaid task assistance or financial support. Finally, older men were less engaged in informal support for others than were older women, but overall levels of support were high for both genders (Dentinger & Clarkberg, 2002; Liebler & Sandefur, 2002).

Marital status has been found to condition the provision of help to kin and nonkin, as well as shape gender differences in informal support. Married adults have been found to provide less support to parents and nonkin than unmarried adults, a pattern that may be especially true for women (Bracke, 2008; Laditka & Laditka, 2001; Liebler & Sandefur, 2002; Sarkisian & Gerstel, 2004). Men provided more care when they were married to women who were providing support to others (Gerstel, 2000; Gerstel & Gallagher, 2001). Divorce detached men from others, especially their adult children, and remarriage exacerbated this tendency (Amato et al., 1995; Cooney & Uhlenberg, 1990; Kalmijn, 2007; Pezzin & Schone, 1999). Divorce and remarriage did not, however, weaken the ties between mothers and adult children (Amato et al., 1995; Pezzin & Schone, 1999).

Other important determinants of informal support fit into a supply and demand framework. Help is typically only provided when there are needs to be met. Elderly parents received more assistance when they were in poor health and lived in their own residence (as opposed to a senior residence) (Choi, 2003). On one hand, when adult children had families of their own, they were more likely to ask their older parents to provide childcare for grandchildren (Wang & Marcotte, 2007). On the other hand, coresident adult children who were unable to establish an independent life tended to reduce the time and energy that older parents had for helping children who had already left the parental home (Aquilino, 2006).

Supply factors also condition support. Those who were geographically more proximate to kin provided more help (Compton & Pollak, 2009), although the direction of causality is unclear because a change in need may result in a geographic move. Factors that affect the availability of time and other valued resources such as education, income, and good health have been shown to be important controls when assessing who gives help to others (Couch, Daly, & Wolf, 1999). Those with more resources tend to provide more support to others, though patterns often vary by the type of help provided. For example, more affluent people provided the money to purchase assistance whereas those with more time provided more hands-on instrumental support (Boaz, Hu, & Ye, 1999; Zissimopoulos, 2001). Support for elderly parents is often shared among siblings and hence family size affects the provision of help. The more children an elderly person has, the less likely any given child provided help and the more likely that some of the parent-child ties were weak or estranged (Ward, Spitze, & Deane, 2009).

To summarize, large cohorts of men and women have begun to move into older ages, increasing the need to assess how connected these individuals are to friends and family. In this study, we use the most recent waves of the Wisconsin Longitudinal Study (WLS) to assess gender differences in help given to others among adults as they age from their 50s into their 60s, the ages at which many transition from full-time work to other arrangements and when most experience the departure of all children from the parental home. The goal is to illuminate gender gaps in helping others in later life, to describe the web of social connections older adults have to family and friends, and to assess changes for men and women as they transition into retirement and the “empty nest.”

METHOD

Data

The WLS has followed a representative sample of men and women who graduated from Wisconsin high schools in 1957 and has become a long-term study of aging and the life course. Data were collected in 1957 from an original sample of 10,317 graduates, and respondents have been re-interviewed about once per decade, using phone interviews and mail questionnaires. In addition to extensive measures of social background, family characteristics, and employment patterns, these data included detailed reports in 1993 and 2004 of help provided to, or received from, family members (other than the spouse) and nonkin in the past month. This information provided a unique opportunity to examine gender differences in helping behaviors in late middle age. Because the sample was restricted to the 1957 cohort of high school graduates in the state of Wisconsin, the WLS did not permit analysis of either high school dropouts or ethnic minorities. The surviving WLS cohort resembled about two thirds of the national cohort: that is, about two thirds of Americans 60 – 64 in 2000 were non-Hispanic white men and women with at least 12 years of education (Sewell, Hauser, Springer, & Hauser, 2004).

Analytic sample

Our analysis utilized data from the 1993 and 2004 graduate surveys. About 75% of the original 10,317 graduates remained in the sample in 2004 by completing either a mail questionnaire or telephone survey. Because our key variables came from both data collection instruments, we restricted our analytic sample to those 5,465 graduates who completed both a mail questionnaire and a telephone survey in both 1993 and 2004. We eliminated 245 cases, or 4.5% of the sample, with missing values on one or more variables. Because most of the missing information was on the dependent variables (167 of 245 cases or 68%), we chose not to impute missing data for the remaining small number of cases. We thus analyzed a sample of 5,220 graduates with no missing data on the variables used in the analysis.

Our analysis was organized around help to different types of recipients (e.g., parents, adult children, nonkin). We created subsamples of respondents who were in a position to provide informal care to each category of recipient. The “assistance to parents” sample included the 1,294 respondents who had at least one living parent in each year. This restriction allowed us to observe changes over time in helping behaviors as both respondents and their elderly parent(s) grew older. The sample for the “assistance to adult children” models included those respondents with at least one child over the age of 18 in each year (N = 4,747 in 1993 and N = 4,850 in 2004). The nonkin sample included all 5,220 respondents, as all were assumed to have friends, neighbors, or coworkers.

Measures

Dependent variables

Our measures of help work were constructed from the following question asked in both the 1993 and 2004 mail surveys: “…we are interested in the help and support that you receive from or give to people (other than a spouse). We are interested here in help that is not paid for. During the past month have you GIVEN the following kinds of help?” The question was followed by a grid of four types of help by six categories of kin and nonkin who might need help. Respondents were asked to check the relevant boxes.

From these data, we first created dichotomous measures for each type of help: (a) transportation, errands, or shopping, (b) housework, yard work, repairs, or other work around the house, (c) advice, encouragement, or moral or emotional support, and (d) babysitting or childcare. Then, each of these measures was divided according to the recipient of the help. We focused only on help to parents, adult children, and nonkin (friends, neighbors, and co-workers) and created three summary measures reflecting the provision of “any kind of help” to parents, adult children, and nonkin. Because babysitting and childcare was most relevant to respondents with grandchildren, and helping grandchildren was only possible for respondents with adult children, we only examined babysitting help to adult children. Unfortunately, we did not have information about help given specifically to grandchildren in 1993, so we inferred this based on information about childcare support provided to adult children. In order to maintain comparability across years, we adjusted the 2004 data by combining childcare provided to adult children and to grandchildren into one measure.

We recognize several limitations to these measures: first, the data only allowed us to look at whether help was provided and not the amount or intensity of assistance. This likely reduced the gender differences in help provided since women have often been shown to spend more time than men helping others (Gerstel, 2000; Gerstel & Gallagher, 2001; Liebler & Sandefur, 2002). In addition, the categories of assistance combined the traditionally “female” gender-based task of housework with the traditionally “male” activities of yard work and repairs. Finally, very few respondents reported giving help to siblings or other relatives and we were unable to investigate help to these groups of kin. Despite these limitations, we feel the data provide a rich picture of the social fabric connecting older adults to others.

Independent variables

Each of the multivariate models included controls for the respondent’s marital status, employment status, educational attainment, household income, and health status; we also controlled for whether a son or daughter resided with the respondent, and the number of married adult children. Unfortunately, we did not know the number of grandchildren, so we relied on the number of married adult children as a proxy. The models examining help provided to parents also included measures of parental health, whether the parents lived in their own home, the proximity of the respondent to the parents residence, and whether the respondent had a living sibling (e.g., who might share in parental care). We realized that geographic proximity might be endogenous to the provision of help (e.g., if parents and adult children move closer together in order to facilitate the exchange of assistance), but we included it as an important control for accessibility.

Table 1 presents means and distributions of each of the independent variables for the full sample. Between the 1993 and 2004 interviews, the WLS cohort experienced declines in marriage, increases in retirement, more empty-nesters, worsening health, and rising frailty of elderly parents.

Table 1.

Sample Characteristics (%) by Year, Wisconsin Longitudinal Study Graduates, 1993 and 2004

Variables	1993	2004
All WLS Graduates (n = 5,220)
Gender: Male	45.2	45.2
Marital status
Currently married, first marriage	71.4	65.0
Currently married, second or higher marriage	13.5	15.0
Previously married: Divorced/widowed/separated	11.3	16.3
Never married	3.8	3.7
Employment status
Not employed	14.1	57.0
Currently employed part-time	12.5	22.6
Currently employed full-time	73.4	20.5
Education: Years of education based on highest degree
High school grad or less than 1 year college	55.1	55.0
1–3 years college	15.8	15.6
Bachelor’s degree	14.4	14.3
More than bachelor’s degree	14.7	15.1
Household income in last 12 months
<=$25,000	15.5	24.6
$25,001–$50,000	27.3	27.1
$50,001–$75,000	26.5	20.5
$75,001–$100,000	13.4	11.0
>$100,000	17.3	16.8
Health status: Poor/very poor (0 = Excellent/good/fair)	1.1	8.5
Any coresident children (0 = No coresident children)	36.7	9.0
Number of married children
0	32.9	15.4
1	28.7	22.6
2	23.1	29.7
3+	15.3	32.4
Has at least one living parent	63.9	24.8
Has at least one adult child	90.9	92.9
WLS Graduates with at least one living parent in both 1993 and 2004 (n = 1,294)
Parent’s residence
In own home (0 = With graduate or other living situation)	95.1	64.1
Proximity to parent’s residence
<5 miles (including coresidence)	26.8	27.2
6–20 miles	17.0	18.0
21–120 miles	19.2	18.3
121–1,000 miles	22.0	19.9
>1,000 miles	15.6	15.2
Parental health: Poor/very poor (0 = Excellent/good/fair)	9.9	24.5
Living siblings (0 = No living siblings)	93.2	92.4

Note: Characteristics shown for full sample for individual/household and children characteristics. Parent and sibling characteristics only shown for the subset of respondents who had a living parent in both survey waves (the analytic sample for the paren

Analytic Strategy

We examined bivariate gender differences in the provision of help in order to answer three questions: (a) At older ages, do women provide significantly more help to others than do men? (b) Do patterns vary by the type of help provided and the type of recipient? and (c) Does the gender gap decline as individuals age from their 50s (in 1993) into their 60s (in 2004)? We then used logistic regression to examine gender differences over time in the provision of different types of help to parents, adult children, and nonkin, net of a range of supply and demand factors, including employment status. Our final analysis tested specifically whether adults were more likely to help others after transitioning out of the labor force. In all models, we paid particular attention to differences by marital status and incorporated gender-marital status interactions.

RESULTS

Bivariate results

Table 2 gives an overview of gender and marital status differences in help provided to parents, adult children, and nonkin in 1993 and 2004. In addition to showing percentages by gender in each year, the table shows the gender gap, calculated as the arithmetic difference between the women’s and men’s percentages, as well as differences by marital status with significance noted in bold. Overall, the vast majority (well over 90%) of both women and men reported providing some kind of informal help to either friends or family (other than their spouse) in the past month, with only a small 2 – 3 percentage point decline over the decade as they aged from their 50s into their 60s. Women were consistently more likely than men to help others (p < .001), though the overall gender gap remained surprisingly small at 3 – 4 percentage points.

Table 2.

Provision of Help (%) by Gender, Year, and Marital Status, Wisconsin Longitudinal Study Graduates 1993 and 2004

	N		1993			2004			Gender Gap (Women - Men)
	1993	2004	Women	Men	sig.a	Women	Men	sig.a	1993	2004	sig.b
Any help to any recipient	5,220	5,220	97.7	94.2	***	95.6	91.6	***	3.5	4.0
Any help to parentsc	1,294	1,294	60.9	50.9	***	63.8	53.3	***	10.0	10.5
Any help to adult childrend	4,747	4,850	82.7	73.0	***	80.0	75.2	***	9.8	4.8	**
Any help to friends/neighbors/coworkerse	5,220	5,220	70.4	68.3		59.8	56.1	**	2.1	3.7
By Marital Status
Any help to parentsc
Currently married, first marriage	922	836	61.8	53.2	**	66.0	57.5	*	8.6	8.4
Currently married, second or higher marriage	179	206	58.2	44.0		55.2	44.5		14.2	10.6
Previously married: Divorced/widowed/separated	161	221	57.6	43.6		62.0	40.0	***	13.9	22.0
Never married	32	31	--f	--f		--f	--f		--f	--f
Any help to adult childrend
Currently married, first marriage	3,523	3,276	82.8	75.7	***	81.7	78.4	*	7.1	3.3
Currently married, second or higher marriage	664	763	80.8	65.6	***	75.2	71.0		15.2	4.2	*
Previously married: Divorced/widowed/separated	551	801	84.8	63.7	***	78.2	59.7	**	21.1	18.4
Never married	9	10	--f	--f		--f	--f		--f	--f
Any help to friends/neighbors/coworkerse
Currently married, first marriage	3,726	3,393	67.3	67.2		56.6	53.9		0.1	2.7
Currently married, second or higher marriage	702	781	73.4	69.8		61.1	57.9		3.6	3.3
Previously married: Divorced/widowed/separated	592	851	79.4	75.4		65.8	62.7		4.0	3.0
Never married	200	195	84.2	67.4	**	73.2	73.5		16.8	−0.3	*

Note: Significance of difference between married and unmarried individuals’ provision of help in each year is determined by a two-tailed t -test. Bolded figures indicate significant difference (p < .05 or lower) from currently married (first marriage) ind

^aSignificance of difference between women’s and men’s provision of help in each year is determined by a two-tailed t -test.

^bSignificance of difference between the gender gaps in 1993 and 2004 is determined by the significance of the interaction term in a model regressing help onto year, gender, and year*gender.

^cSample includes all respondents with a living parent in both 1993 and 2004.

^dSample includes all respondents with an adult child in each year.

^eSample includes all respondents.

^fCell size too small for reliable estimate.

^*p < .05.

^**p < .01.

^***p < .001.

Over half of all women and men with a living parent helped that parent in the past month, though women were consistently 10 percentage points more likely than men to do so in both years (p < .001). More than three quarters of men and women with adult children assisted their children in some capacity, with women again more likely than men to do so (p < .001). Over time, the gender differences in help to adult children grew significantly narrower, from almost 10 percentage points in 1993 to less than 5 percentage points in 2004 (p < .01). Over half of all women and men in both years reported helping nonkin in the past month, though this declined by more than 10 percentage points for both men and women between 1993 and 2004. The gender gap was quite small (about 2 – 4 percentage points in both years).

Helping others differed substantially by whether respondents had ever been married, whether their first marriage was still intact, and if not, whether they had remarried. In terms of help to aging parents, the gender gap favoring women was apparent regardless of marital status, though it was statistically significant only for women and men who were still in their first marriages (p < .01 in 1993 and p < .05 in 2004), and for previously married women and men in 2004 only (p < .001). Continuously married men were significantly more likely than other men to help their parents in 2004 (p < .05).

Women of all marital statuses were significantly more likely than men to help their adult children, though the gap narrowed over time as more married (and especially remarried) men helped adult children in 2004 than in 1993. For remarried adults, the gender gap narrowed from 15 percentage points in 1993 to 4 percentage points in 2004 (p < .05). In both years, men who were no longer in their first marriage were significantly less likely than continuously married men to help their adult children (p < .05). Both women and men who were no longer married became less likely to help their adult children over time and the gender gap remained quite large.

The marital status differences in help to nonkin were quite different than the patterns for help to family members, with the never married the most likely and the continuously married the least likely to assist nonkin. Over time, women of all marital statuses became less engaged in supporting nonkin, and the same is true for men except for the never married: by 2004, never married men had increased their support of nonkin, from 67% to 73%. The decline for women and increase for men eliminated the gender gap (17 percentage points to 0 percentage points (p < .05)) in help to nonkin for never married adults.

Table 3 shows the specific types of help that women and men provided to parents, adult children, and nonkin. As in Table 2, we present the proportion of women and men who provided different types of help in 1993 and 2004, the significance of the gender gap for each year, the change over time in the gap, and the significance of differences by marital status (in bold). Unlike in Table 2, we collapsed marital status into a dichotomy (currently married vs. not currently married), to simplify the presentation of results.

Table 3.

Provision of Detailed Types of Help (%) by Gender, Year, and Marital Status, Wisconsin Longitudinal Study Graduates, 1993 and 2004

	1993			2004			Gender Gap (Women - Men)
	Women	Men	sig.a	Women	Men	sig.a	1993	2004	sig.b
Help to Parentsc
Currently Married (1993 n = 1101, 2004 n = 1042)
Any Help to Parents	61.3	51.4	***	64.2	54.6	**	9.9	9.6
Any Transportation/Errands/Shopping	39.9	31.3	**	48.6	37.5	***	8.7	11.1
Any Housework/Yard work/Repairs	22.1	29.3	**	28.2	22.8	*	−7.2	5.4	***
Any Advice/Encouragement/Emotional Support	44.0	29.5	***	44.4	34.8	**	14.5	9.6
Not Currently Married (1993 n = 193, 2004 n = 252)
Any Help to Parents	59.1	47.0		62.8	42.6	**	12.1	20.2
Any Transportation/Errands/Shopping	35.4	28.8		47.6	37.7		6.6	9.9
Any Housework/Yard work/Repairs	28.4	28.8		25.1	21.3		−0.4	3.8
Any Advice/Encouragement/Emotional Support	38.6	27.3		48.2	23.0	***	11.3	25.2
Help to Adult Childrend
Currently Married (1993 n = 4187, 2004 n = 4039)
Any Help to Adult Children	82.5	73.9	***	80.7	76.8	**	8.6	3.8	**
Any Transportation/Errands/Shopping	35.0	34.8		28.0	30.2		0.2	−2.3
Any Housework/Yard work/Repairs	19.6	19.8		18.9	26.0	***	−0.2	−7.1	*
Any Advice/Encouragement/Emotional Support	67.9	58.3	***	62.8	54.6	***	9.6	8.2
Any Child Care/Babysitting	46.5	28.3	***	54.0	49.4	**	18.3	4.6	***
Not Currently Married (1993 n = 560, 2004 n = 811)
Any Help to Adult Children	84.0	63.2	***	77.6	59.6	***	20.8	18.0
Any Transportation/Errands/Shopping	38.4	27.6	*	25.9	26.0		10.8	−0.1
Any Housework/Yard work/Repairs	20.3	11.9	*	19.6	20.2		8.4	−0.6	***
Any Advice/Encouragement/Emotional Support	70.4	55.1	***	61.2	46.2	***	15.3	15.0
Any Child Care/Babysitting	36.0	12.4	***	47.1	23.1	***	23.6	24.0
Help to Friends/Neighbors/Coworkerse
Currently Married (1993 n = 4428, 2004 n = 4174)
Any Help to Nonkin	68.2	67.7		57.3	54.7		0.5	2.6
Any Transportation/Errands/Shopping	28.2	29.7		25.0	24.6		−1.5	0.4
Any Housework/Yard work/Repairs	10.1	23.6	***	8.9	23.2	***	−13.5	−14.3
Any Advice/Encouragement/Emotional Support	61.0	52.6	***	49.2	37.4	***	8.4	11.8
Not Currently Married (1993 n = 792, 2004 n = 1046)
Any Help to Nonkin	80.4	73.0	*	66.9	65.7		7.5	1.2
Any Transportation/Errands/Shopping	39.9	43.1		33.1	38.0		−3.1	−4.8
Any Housework/Yard work/Repairs	18.4	37.4	***	10.8	32.7	***	−19.0	−21.9
Any Advice/Encouragement/Emotional Support	75.3	61.6	***	58.0	49.8	*	13.8	8.2

Note: Significance of difference between married and unmarried individuals’ provision of help in each year is determined by a two-tailed t -test. Bolded figures indicate significant difference (p <.05 or lower) from currently married individuals of the sam

^aSignificance of difference between women’s and men’s provision of help in each year is determined by a two-tailed t -test.

^bSignificance of difference between the gender gaps in 1993 and 2004 is determined by the significance of the interaction term in a model regressing help onto year, gender, and year*gender.

^cSample includes all respondents with a living parent in both 1993 and 2004.

^dSample includes all respondents with an adult child in each year.

^eSample includes all respondents.

^*p < .05.

^**p < .01.

^***p < .001.

Whereas Table 2 showed that, regardless of marital status, women were more likely than men to provide “any help” to parents, adult children, and nonkin, Table 3 showed much greater gender variation across different types of assistance. For example, in six of the twelve comparisons, men were more likely than women to provide assistance with “housework, yard work, or repairs,” and in four other comparisons, there was no gender difference.

Women were much more likely than men to provide emotional support to both family and friends. Unmarried men became less engaged emotionally with their parents over time, in contrast to married men and both married and unmarried women, all of whom provided more emotional support as their parents grew older. Although neither change reached statistical significance, the gender gap in emotional support to parents grew narrower over time for currently married adults (from 15% to 10%) but wider for unmarried adults (from 11% to 25%).

Table 3 shows a dramatic increase over time in men’s provision of childcare or babysitting to adult children. The proportion of married men who provided childcare almost doubled from 28% in 1993 to 49% in 2004. Because the proportion of married women who did childcare only increased from 46% to 54%, the overall gender gap dropped from 18% to less than 5% (p < .001). In results not shown, we found similar gender differences for continuously married and remarried women and men, though the likelihood of providing childcare was significantly higher for the continuously married. Unmarried men also increased their role in childcare from 12% in 1993 to 23% in 2004; however, because unmarried women also increased their levels of childcare, the gender gap remained quite large at about 24%.

Multivariate Results

Tables 2 and 3 showed that informal support depended not only on gender but also on marital status, and changed over time as individuals aged into their 60s. With a series of logistic regressions predicting “any help” (Table 4) and detailed “types of help” (Table 5) to parents, adult children, and nonkin, we examined whether these gender and marital status differences could be explained by personal resources and constraints (e.g., employment), as opposed to the needs and demands of others.

Table 4.

Logistic Coefficients Predicting the Provision of Any Help to Parents, Adult Children or Nonkin During the Past Month, Wisconsin Longitudinal Study Graduates, 1993 and 2004

	Parentsa		Adult Childrenb		Nonkinc
	1993	2004	1993	2004	1993	2004
Gender: Male	−0.37*	−0.46**	−0.44***	−0.21*	−0.18*	−0.25***
Marital status
Currently married, first marriage	---	---	---	---	---	---
Currently married, second or later marriage	−0.03	−0.53*	−0.25	−0.49***	0.26	0.16
Previously married: Divorced/widowed/separated	−0.11	−0.14	0.39*	−0.16	0.58***	0.37***
Never married	−0.09	0.09	−1.92*	−1.61*	0.74**	0.47*
Male * Currently married, second or later marriage	−0.17	0.25	−0.28	−0.02	−0.14	0.02
Male * Previously married: Divorced/widowed/separated	−0.21	−0.46	−0.73**	−0.65**	−0.12	0.06
Male * Never married	0.11	−0.33	0.35	0.48	−0.76*	0.31
Employment status
Not currently employed	---	---	---	---	---	---
Currently employed part-time	−0.22	−0.02	−0.13	0.04	0.21	0.15*
Currently employed full-time	−0.44*	−0.29	−0.19	−0.11	0.19*	0.09
Education: Years of education based on highest degree
High school grad or less than 1 year college	---	---	---	---	---	---
1–3 years college	0.36*	0.24	0.27*	0.30**	0.34***	0.40***
Bachelor’s degree	0.86***	0.57**	0.28*	0.35**	0.51***	0.46***
More than bachelor’s degree	0.65**	0.56**	0.53***	0.38**	0.76***	0.68***
Household income in last 12 months
<=$25,000	---	---	---	---	---	---
$25,001–$50,000	−0.10	−0.06	0.22	−0.11	0.23*	0.02
$50,001–$75,000	−0.04	0.20	0.41***	0.04	0.26**	0.02
$75,001–$100,000	−0.18	0.23	0.52***	−0.04	0.21	0.28**
>$100,000	0.15	0.39	0.59***	0.23	0.33**	0.04
Health status: Poor/very poor (0 = Excellent/good/fair)	−0.63	−0.14	−0.78**	−0.23	0.09	−0.20*
Coresident children (0 = No children or no coresident children)	−0.21	−0.29	0.30***	0.48***	−0.03	−0.18
Number of married children	−0.04	0.10*	0.33***	0.24***	−0.03	−0.05*
Parent’s residence
In own home (0 = With graduate or other living situation)	0.08	0.64***	--	--	--	--
Proximity to parent’s residence
<5 miles (including coresidence)	---	---	---	---	---	---
6–20 miles	−0.14	−0.11	--	--	--	--
21–120 miles	−0.62***	−0.85***	--	--	--	--
121–1,000 miles	−1.37***	−1.64***	--	--	--	--
>1,000 miles	−1.42***	−1.61***	--	--	--	--
Parental health: Poor/very poor (0 = Excellent/good/fair)	0.49*	−0.10	--	--	--	--
Living siblings (0 = No living siblings)	−0.20	−0.10***	--	--	--	--
−2 log L	1638	1540	4706	4943	6261	6916
df	26	26	19	19	19	19
n	1,294	1,294	4,747	4,850	5,220	5,220

^aSample includes all respondents with a living parent in both 1993 and 2004.

^bSample includes all respondents with an adult child.

^cSample includes all respondents.

^*p < .05.

^**p < .01.

^***p < .001.

Table 5.

Logistic Coefficients Predicting Provision of Specified Type of Help During the Past Month,^a Wisconsin Longitudinal Study Graduates 1993 and 2004

	Transportation, Errands, Shopping		Housework, Yardwork, Repairs		Advice, Encouragement, Emotional support		Child care, Babysitting
	1993	2004	1993	2004	1993	2004	1993	2004
Help to Parentsb
Gender: Male	−0.34*	−0.50**	0.37*	0.09	−0.73***	−0.58***	---	---
Marital status
Currently married, first marriage	---	---	---	---	---	---	---	---
Currently married, second or later marriage	−0.42	−0.49	−0.15	0.32	0.16	−0.40	---	---
Previously married: Divorced/widowed/separated	−0.14	−0.18	0.31	−0.20	−0.32	0.10	---	---
Never married	−0.60	0.28	0.40	0.74	−0.86	0.20	---	---
Male * Currently married, second or later marriage	0.32	0.37	0.12	−1.00*	−0.31	0.34	---	---
Male * Previously married: Divorced/widowed/separated	−0.04	0.01	−0.46	−0.19	−0.02	−0.69	---	---
Male * Never married	0.23	−0.06	0.38	−0.75	1.33	−0.10	---	---
Help to Adult Childrenc
Gender: Male	0.08	0.14	0.26**	0.55***	−0.45***	−0.48***	−0.55***	−0.04
Marital status
Currently married, first marriage	---	---	---	---	---	---	---	---
Currently married, second or later marriage	0.02	−0.20	−0.33*	−0.36*	0.02	0.00	−0.72***	−0.96***
Previously married: Divorced/widowed/separated	0.25*	−0.12	0.17	0.05	0.25	0.00	−0.17	−0.25*
Never married	−1.01	−0.03	−0.04	−0.21	−1.38	−1.46	−0.55	−1.46
Male * Currently married, second or later marriage	−0.41*	−0.30	−0.29	−0.26	−0.31	0.08	0.05	0.05
Male * Previously married: Divorced/widowed/separated	−0.54*	−0.12	−0.86**	−0.47*	−0.28	−0.17	−0.84**	−1.08***
Male * Never married	0.77	1.10	0.69	−10.36	0.49	1.44	−10.31	1.55
Help to Friends/Neighbors/Co-workersd
Gender: Male	0.04	−0.05	1.18***	1.28***	−0.55***	−0.70***	---	---
Marital status
Currently married, first marriage	---	---	---	---	---	---	---	---
Currently married, second or later marriage	0.12	0.24	0.29	0.35	0.29*	0.22	---	---
Previously married: Divorced/widowed/separated	0.48***	0.37***	0.58***	0.19	0.63***	0.32**	---	---
Never married	0.59**	0.77***	1.02***	0.95***	0.69**	0.02	---	---
Male * Currently married, second or later marriage	0.11	0.01	−0.40	−0.34	−0.20	−0.08	---	---
Male * Previously married: Divorced/widowed/separated	0.26	0.25	0.12	0.30	−0.14	0.21	---	---
Male * Never married	−0.43	−0.06	−1.07**	−0.73*	−0.36	0.62*	---	---

Note: The omitted marital status variable is “currently married in first marriage.”

^aThese models also include all of the covariates included in Table 3. Full model results can be found online in Appendix Tables 1a, 1b, and 1c.

^bSample includes all respondents with a living parent in both 1993 and 2004.

^cSample includes all respondents with an adult child.

^dSample includes all respondents.

^*p < .05.

^**p < .01.

^***p < .001.

Table 4 shows that net of a wide range of covariates, men were still significantly less likely than women to provide “any help” to parents, adult children, and nonkin in both 1993 and 2004. Because the models also included interactions between gender and marital status, the main gender coefficients represented the gender effects for continuously married adults in a first marriage (the omitted category on marital status). In models for all three types of recipients, continuously married men were significantly less likely than married women to provide any kind of help in both 1993 and 2004.

The gender*marital status interactions differed for the three types of recipients. Net of the covariates in the model, the gender gap in help to parents did not differ significantly by marital status: whether married or not, men were consistently less likely than women to help their parents in both 1993 and 2004. In terms of help to adult children, the negative gender*divorced/widowed/separated interaction implied that the gender gap favoring women among the previously married was considerably larger in both years than the gap among the continuously married. Finally, with regard to help to nonkin, in 1993, but not in 2004, the gender gap for the never married was significantly larger than the gap for the continuously married, consistent with the narrowing differences over time by marital status seen in Table 2.

Table 4 also highlighted the association between helping others and a number of covariates. Looking first at the parent models, in 1993, when respondents were in their 50s (and most were still working), those employed full-time were significantly less likely to help parents compared with those not working at all. By 2004, however, when the majority of respondents had retired, current employment status no longer mattered. We tested whether the impact of employment status differed for women and men, but the interactions were not statistically significant.

Whereas employment was associated with giving less help to parents, it was associated with giving more help to nonkin; we interpret this to reflect the support given to colleagues and coworkers. As expected, people with more resources were more likely to help family and friends than people with fewer resources, with education a stronger and more consistent predictor than income in both years. Respondents in poor health were significantly less likely to help adult children (in 1993) and nonkin (in 2004), though the negative effects were not statistically significant in the help to parents models.

Parent characteristics were important determinants of whether their adult children provided them with any help. By 2004, when most surviving parents were quite old, those who were still living in their own homes were significantly more likely to get support. Respondents who lived closer to their parents gave significantly more help than did those who lived further away. In 1993, parents in poor health got significantly more help than healthy parents, but this difference disappeared by 2004.

When adult children lived in the parent’s home, the parent was much more likely to provide adult children with instrumental and emotional support. Parents with more married children were significantly more likely to help adult children in both 1993 and 2004, but were less likely to help nonkin (in 2004).

Table 5 shows results from models predicting the provision of different types of help. Because the effects of the covariates in these models were quite similar to those in the “Any Help” models in Table 4, Table 5 only presents the effects of gender, marital status, and the gender*marital status interactions. Models included all of the covariates in Table 4, with complete results available in online Appendix Tables 1a – 1c. In Table 5, as in Table 4, the main gender effects reflect the gender gap for continuously married respondents (the omitted category on marital status). The implied gender effects for the other marital status groups require adding the main effect to each of the interaction effects. Similarly, the main marital status effects reflect the marital status effects for women, whereas the effects for men require summing the main and interaction effects.

Unlike the results from Table 4, which showed a consistent and significant gender gap favoring women, men outperformed women in the provision of “housework, yard work, and repairs” and there were other cases where there were no gender differences (e.g., in transportation help to adult children and nonkin and in babysitting help to adult children in 2004). Continuously married men were significantly more likely than continuously married women to provide housework, yard work, or repairs to all three sets of recipients, though this pattern did not hold for other marital status categories. Whereas men exceeded women in providing instrumental support to friends and family, women dominated in the realm of emotional support. Women of all marital statuses were consistently more likely than men to provide advice, encouragement, and emotional support to their parents, adult children, and nonkin. The one exception to this pattern was in 2004 when never married men appeared to have significantly narrowed the gap with women in this kind of support to nonkin.

Although most gender differences persisted over time, the most dramatic exception was the provision of childcare or babysitting assistance to adult children. In 1993, married women were significantly more likely than married men to provide childcare or babysitting (p < .001). By 2004, the gender difference had virtually disappeared due to the dramatic increases in childcare among married men (see Table 3). Although divorced men increased their provision of childcare over time, divorced women were much more likely to care for grandchildren, even more so at older ages; the net result was an even bigger gender gap by 2004 among the previously married.

The final analysis asked whether adults who transitioned out of the labor force, either by retiring or by reducing to part-time hours, were more likely to help others than those who continued to work full-time. In order to observe people who cut back their work commitments, we limited the sample to women and men who were working full-time in 1993, and we then examined the effects of their work status in 2004. We assumed those who were not working at all in 2004 had retired, and those working part-time had cut back their work hours. Our models, which are run separately for men and women, examined the impact of work status in 2004 (retired and part-time vs. full-time) on the likelihood of helping others in 2004, controlling for whether they provided assistance in 1993, plus all the controls in the previous models. For simplicity, we only present the employment status effects in Table 6 (full models can be found in online Appendix Tables 2a – 2c). These effects can be interpreted as the impact of reducing employment from full-time to part-time, or from full-time to retired, relative to remaining employed full-time throughout the period (the omitted category).

Table 6.

Impact of Reducing Employment from Fulltime to either Part-time or Retired between 1993 and 2004 on the Likelihood of Helping Parents, Adult Children, and Nonkin in 2004, by Gender, ^a (Logistic Coefficients ^b), Wisconsin Longitudinal Study Graduates, 1993

	Men		Women
	Part-time	Retired	Part-time	Retired
Help to Parentsc
Any Help	0.53	0.40	−0.01	−0.11
Transportation, Errands, Shopping	0.41	0.34	0.23	0.16
Housework, Yardwork, Repairs	0.33	0.47	0.43	0.30
Advice, Encouragement, Emotional Support	0.37	0.13	−0.46	−0.23
Help to Adult Childrend
Any Help	0.24	0.30*	0.19	−0.11
Transportation, Errands, Shopping	0.21	0.31*	0.43*	−0.04
Housework, Yardwork, Repairs	0.14	0.26*	−0.06	0.08
Advice, Encouragement, Emotional Support	0.36*	0.20	0.10	−0.16
Child care, Babysitting	0.11	0.28*	0.36*	0.08
Help to Nonkine
Any Help	0.13	−0.01	−0.02	−0.16
Transportation, Errands, Shopping	0.12	0.12	−0.02	0.22
Housework, Yardwork, Repairs	0.24	0.57***	0.10	0.35
Advice, Encouragement, Emotional Support	−0.07	−0.34**	−0.05	−0.26*

^aSample is restricted to all respondents who were working full time in 1993.

^bLogistic coefficients compare either Part-time or Retired respondents in 2004 to those who are still working full time in models that also control for all the covariates in Table 3 plus a control for whether assistance was given in 1993. Full model res

^cSample includes all respondents with a living parent in both 1993 and 2004.

^dSample includes all respondents with an adult child.

^eSample includes all respondents.

^*p < .05.

^**p < .01.

^***p < .001.

The positive and significant effects for men in Table 6, especially retired men, suggest that when men cut back from full-time employment, they became significantly more likely to help their family and friends. Men were most inclined to help their adult children with a range of tasks, but they also helped nonkin with housework, yard work, and repairs (p < .001). The lack of significant effects of employment change on help to parents suggests that both men and women helped their elderly parents regardless of whether they were still employed. The biggest exception was for providing emotional support to nonkin: men (and women) who had retired were significantly less likely to do this compared to those who were still employed full-time.

DISCUSSION

In this study, we focused on the helping behaviors that connect older adults to a wide set of kin and nonkin. We argued that these connections are important for both men and women because they help solidify the informal support system that these adults may need to activate as they grow older. Women are likely more involved in providing informal support than men, given their greater investment in childcare and kin relationships earlier in life. Nonetheless, as retirement approaches, men more than women may have the opportunity to reallocate time to helping others. Major unanswered questions are: Do men become more involved in kin work as they age? Are gender differences in helping others narrower in retirement than earlier in life? Our answer to these questions is affirmative: men do become more involved in kin work as they age, although this involvement is more apparent when the recipient of help is adult children or friends than elderly parents. Gender differences narrow in some cases, particularly for help to adult children and among continuously married men.

Consistent with prior research, we found that older women provided more help to others than did older men but the gender gap was often modest. Similar to findings reported by others (e.g., Dentinger & Clarkberg, 2002; Liebler & Sandefur, 2002), we found high levels of involvement for both men and women in assisting elderly parents, adult children, and nonkin. Women were far more likely than men to provide emotional support to others. Chesley and Poppie (2009), who used the MIDUS data set, reported a similar gender difference in the provision of emotional support to elderly parents and parents-in-law. Whereas much of the literature on informal support focuses on the provision of help to elderly parents, our wider focus including adult children and friends showed less traditional gender differentiation than patterns of assistance to elderly parents. For example, we found that men provided more help than women with housework, yard work, and repairs, particularly for friends and neighbors and for adult children (in 2004). Help with yard work and repairs, activities men typically do, may be more easily given and more often needed across households than is traditional housework. If we had been able to disaggregate this category, we would likely have seen women doing more housework for others and men doing more yard work and repairs.

Our expectation that we might see relatively greater increases over time in help work of men than women was only borne out in a few cases. As they aged from their 50s to their 60s, men most clearly increased the assistance they gave to their adult children, especially with childcare, but also with help doing housework, yard work and repairs. As they retired from the workforce – and perhaps also as their adult children established families and households of their own – men became much more involved in the lives of their children and grandchildren, virtually eliminating any gender difference by the time they were in their 60s.

Our disaggregation of helping behaviors by marital status provided further insight into the gender differences in connections to kin and nonkin. Consistent with the work of Lareau (2000) and others, being married encourages men to remain supportive of their parents and adult children, though wives in first marriages seem to have a more positive impact than wives in second or later marriages. We found that, consistent with previous studies, men who were no longer in their first marriage were significantly less likely than continuously married men to help their adult children, perhaps reflecting strained relations following a divorce, or weak ties to step children (Amato et al., 1995; Cooney & Uhlenberg, 1990; Pezzin & Schone, 1999). However, remarriage seemed to encourage stronger ties between fathers and their adult children when compared to divorced or widowed men. Although men who were no longer married were less often engaged with their parents and adult children than married men, they were more likely than their married counterparts to help their friends, neighbors, and coworkers. Never married men were also more engaged in helping nonkin than married men, perhaps substituting friendship for family ties.

As with any study, this one has a number of limitations. First, the WLS measures do not show the intensity of helping others, just whether or not any help was given. Without measures of time spent helping others, we are unable to assess the overall burden of help provided to others and are likely underestimating the actual gender differences. Further research utilizing time diary data could shed light on these issues in the cross-section. Second, we do not assess help given to other family members, such as siblings or spouses, who may also be recipients of informal help. Third, our window on helping others captures a relatively small slice of the life course (individuals aging form their 50s to their 60s). We therefore do not assess gendered behaviors earlier in these individuals lives, nor do we yet know how things may change after the entire cohort retires and ages into their 70s and beyond. Finally, there are limitations to our sample – it is a mostly White, Midwestern cohort with at least a high school level of education - and hence further research on different samples will be needed to assess the wider generalizability of our findings.

Despite these limitations, the analysis reported here offers unique insights into the social connections of older adults. Although older women are more engaged in helping others than are men, older men nonetheless seem very engaged in a variety of ways with others. The results suggest an optimistic picture of the social fabric in later life, good news as we enter a period of rapid increase in the size of older cohorts in the United States. Most older adults – men and women - have an active web of obligations to others that they fill on a regular basis. This likely contributes not only to the well-being of others but also to their own sense of worth, connectedness to others, and their own future health and well-being.