Journal of Marriage and Family 65 (May 2003): 396–403396

EUNJU LEE, GLENNA SPITZE, AND JOHN R. LOGAN

State University of New York at Albany

l

Social Support to Parents-in-Law: The Interplay of

Gender and Kin Hierarchies

Using data from a representative sample of mid-dle-aged married persons, we compare men’s andwomen’s contact and assistance to older parents-in-law and parents. Women have more visits andphone contact with their parents than do men, andmen talk on the phone more with their in-lawsthan do women. There are no gender differencesin assistance patterns. Multivariate analysisshows that women contact and help parents morethan in-laws, whereas for men there are no suchdifferences. There is little direct evidence that thepresence of one set of parents affects relationswith the other. Our findings suggest that althoughwomen clearly give priority to relations with theirown parents, men experience pulls in both direc-tions.

Both parents and parents-in-law are central tomarried people’s family networks for a significantportion of their adult lives. In recent years, how-ever, there has been much more research and pub-lic attention devoted to children’s support to olderparents than to relations with their in-laws (Allen,Blieszner, & Roberto, 2000; Goetting, 1990; Lo-gan & Spitze, 1996; Rossi & Rossi, 1990; Waite& Harrison, 1992). A review of research on in-law relations concluded that, although the targetof many jokes and stereotypes, these relations

Department of Sociology, SUNY Albany, Albany, NY12222 (gds01@albany.edu).

Key Words: adult children, aging parents, in-law relations,parents-in-law.

have been ‘‘underrepresented in sociology jour-nals and family textbooks’’ (Goetting, p. 67). Ros-si and Rossi showed how parents-in-law fit intothe hierarchy of normative obligations, but la-mented that they did not collect more informationabout contact and assistance patterns with in-laws(pp. 502–503). Ten years later, Walker (2000, p.606) added that ’’we sometimes have very littleto say worth noting on subjects of tremendous im-portance to the lived experience of family mem-bers . . . in-laws and in-law relationships, for ex-ample.’’

The purpose of this article is to investigate pat-terns of social support to older parents-in-lawamong middle-aged men and women. Building onwhat we know about family support to aging par-ents, we examine the factors that influence filialrelations with parents-in-law. Two general ques-tions guide our analysis. First, we ask how genderaffects filial relations with in-laws. Gender differ-ences in contact and helping behaviors have beenwidely investigated in the literature on intergen-erational relations (Logan & Spitze, 1996; Rossi& Rossi, 1990), but it is not known how men’sand women’s support to parents-in-law compareand whether there are parallel differences to thosefor parents.

Second, we ask about potentially conflictingobligations to parents-in-law and parents. Waiteand Harrison (1992) argued that family memberscompete for middle-aged women’s limited timeand resources. According to Rossi and Rossi’s(1990) hierarchy of obligations, parents have

397Social Support to Parents-in-Law

higher claims on adult children’s time and re-sources than do in-laws. How does this ranking ofobligations play out in filial relations with bothsets of parents? Do parents receive more supportthan parents-in-law? How does the presence ofone set of parents affect married persons’ relationswith the other set, and does this operate similarlyfor men and women?

To pursue these goals, we analyze data for arepresentative sample of middle-aged marriedmen and women who have parents, parents-in-law, or both. We attempt to determine what factorsshape those filial relations, using as predictorscharacteristics of both the adult child respondentand the parent or parent-in-law. In the section thatfollows, we present what is known about relationswith parents-in-law to place our research ques-tions in context. We also describe other factorsknown to affect relations with parents as back-ground for developing models for relations withparents-in-law.

BACKGROUND AND RESEARCH QUESTIONS

Our first research question concerns the role ofgender in relations with parents-in-law. On onehand, there is a vast literature documenting wom-en’s centrality in kin networks (Allen et al., 2000;Rossi & Rossi, 1990). Women often act as kin-keepers of families, organizing gatherings, makingcontacts, and taking care of the sick and old (DiLeonardo, 1987). Thus, the woman as kinkeeperperspective might suggest that this central rolewould extend to women’s involvement with theirhusbands’ families: Women may have closer tiesthan do men with both their own parents and theirin-laws.

There are, however, reasons to speculate thatthere may be different patterns in filial relationswith in-laws. Caregiving studies suggest that al-though some daughters-in-law may become care-givers to their elderly in-laws, children-in-law aremore often in supporting roles (Goetting, 1990;Kivett, 1989; Merrill, 1993). A study using a rep-resentative sample of middle-aged women foundtheir contact with parents-in-law to be less fre-quent than with their own children and parents(Waite & Harrison, 1992). In part because of theadjustment problems and conflicts that are thesource of popular stereotypes (Fischer, 1983; Ko-marovsky, 1964; Lopata, 1999), it is possible thatthe boundaries of kinkeeping are limited to one’sown children, siblings, and parents. If so, wom-en’s kinkeeping would not bring them closer to

their in-laws but might instead draw their hus-bands toward the wife’s parents during the courseof marriage, in a sort of gravitational pull. Rossiand Rossi (1990, p. 194) hypothesized that ‘‘wom-en’s closer ties to their parents than men’s has theeffect after marriage of involving men in closerand more frequent relations with their parents-in-law than women experience with their husbands’parents.’’ Their data on reported closeness to par-ents-in-law are consistent with this position (seealso Serovich & Price, 1994).

Thus, we know from past research that thereare gender differences in relations with parents,but no clear evidence as to how these might playout in contact with parents-in-law. We test thesedifferences by comparing mean levels of twotypes of contact and assistance given by men andwomen to parents-in-law.

Our second question concerns conditions underwhich parents and parents-in-law may competefor their adult children’s time and attention. Menand women in the middle years face many poten-tially competing responsibilities, including mar-riage, employment, and the needs of dependentchildren living at home, adult children justlaunched, and older parents. Rossi and Rossi(1990, p. 503) described the demands of havingtwo sets of parents: ‘‘Far more prevalent than cop-ing simultaneously with help to elderly parentsand children among middle-aged couples may becoping with crises in the lives of both sets of par-ents . . . or the widowed mothers of both the wifeand the husband.’’ How does the existence of liv-ing parents, or their presence in the local area,affect one’s relations with parents-in-law? Andhow does the presence of living in-laws affect pat-terns of help and contact with one’s own parents?

In a creative study of the structure of norma-tive obligations, Rossi and Rossi (1990) foundparents and children to be in an inner circle towhom the greatest obligations are felt, followedby those removed by one link, including parents-in-law. Married adult children must decide how tobalance their contact with both sets of parents,taking into account any differences that may existin needs for support. This would suggest that, giv-en normal limitations on time and energy, thepresence of one set of parents would decrease sup-port given to the other. The juggling hypothesissuggests that, despite some degree of priority giv-en to parents, there will be competition in bothdirections in these relations when both exist. Al-ternatively, because parents are higher in the kinhierarchy, it is possible that the presence of other

398 Journal of Marriage and Family

kin (including in-laws) will not detract from in-volvement with one’s own parents. But there maybe an effect in the other direction: Having one’sparents living or nearby could reduce involvementwith in-laws. Thus, the kin hierarchy hypothesispredicts an effect of having parents on supportprovided to in-laws but no inverse effect. Thesecontrasting possibilities are investigated in ourmultivariate analysis. Next, we turn to a brief de-scription of other factors to be included in ouranalysis.

OTHER RELEVANT CHARACTERISTICS OF ADULT

CHILDREN AND AGING PARENTS

Other predictors reflect proximity, roles and re-sources of adult children, perceived needs of par-ents, and family structure. One of the most con-sistent determinants of parent–child contact andassistance is proximity (Logan & Spitze, 1996;Rossi & Rossi, 1990; Silverstein & Litwak, 1993;Waite & Harrison, 1992). Thus, we include in ourmodels measures of travel time to parents (and in-laws), as well as whether the other set of parents,if alive, lives in the local area or farther away. Wealso include measures of other roles that may in-fluence availability of middle-aged men’s andwomen’s time, including paid work and parenting,whether of minor or adult children (Rossi & Ros-si, 1990; Stoller, 1983; Waite & Harrison, 1992).Men’s and women’s resources, captured by factorssuch as income, education, and health, may alsoinfluence their availability and willingness tospend time with parents.

Characteristics of older parents such as age,marital status, living arrangements, health, andgender may reflect their needs for support andthus influence contact and support provided byadult children (Silverstein & Litwak, 1993; Stoll-er, 1983). Family structure may also be relevant.Thus, older parents with more children may haveless contact with any one child (Rossi & Rossi,1990). In the same vein, if an older person sharesa residence with sons or daughters, one might ex-pect him or her to need less support from personsoutside the household.

METHOD

Sample

Our data are taken from a personal interview sur-vey conducted in the Albany–Schenectady–Troy,New York, metropolitan area between 1988 and

1989 (see Logan & Spitze, 1996). A probabilitysample of 1,200 persons aged 40 and older wasinterviewed, with an overall response rate of 67%.The sample includes 5.4% African Americans(slightly above 1980 Census figures) and only ahandful of persons in other non-White categories.

For this analysis, we select women and menwho are currently married and have surviving par-ents, surviving parents-in-law, or both. Of these438 respondents, 327 have living parents and 324have living parents-in-law. Because it is not pos-sible to determine the frequency of contact andhelp given to parents coresiding with the respon-dent, those living in the same residence with in-laws or parents (n 5 28) are not included. Wealso exclude data on divorced parents who there-fore do not live with each other (14 cases withdivorced parents, 13 with divorced in-laws, and 3with both sets of parents divorced), because ourmodels require married parents to be treated as asingle unit in terms of contact and assistance.These exclusions result in valid data for 287 re-spondents with parents and 293 with in-laws, ofwhom a few additional cases are lost as a conse-quence of missing data on one or more variables(described later).

Measurement

Dependent variables. Measures of contact includefrequency of visits per month and of telephonecontacts per month. We asked, ‘‘How often do youvisit with [person] in your home or his/hers/theirsor somewhere else?’’ and ‘‘How often do you talkto [person] on the telephone?’’ with eight responsecategories ranging from once a day or more toonce a year or less, to never. Both measures arerecoded to times per month with a maximum of30 (once a day).

Our measure of help was a follow-up to ques-tions asking whether the respondent helped (par-ents/in-laws) during the previous year with vari-ous indoor and outdoor household tasks, errands,or personal/sick care. We then asked, for personshelped, ‘‘Taking all kinds of help together, in anaverage week, how many hours would you sayyou spend helping [person] in one way or anoth-er?’’ We use this measure of weekly hours helpedfor our analysis. It should be noted that althoughparents typically help adult children in variousways for much of their joint lives, the parents (andin-laws) represented here have reached an age(mid- to late 70s) at which they are providing little

399Social Support to Parents-in-Law

TABLE 1. MEANS AND STANDARD DEVIATIONSa FOR INDEPENDENT VARIABLES IN ANALYSIS: RESPONDENTS WITH LIVING

PARENTS, PARENTS-IN-LAW, OR BOTH

Men

M SD

Women

M SD

Key predictorsRelationship (0 5 parents, 1 5 in-laws)Other parent set in areab (1 5 yes)Other parent set lives away (1 5 yes)

.52

.32

.34

.50

.38

.30

Respondent characteristicsEmployed (1 5 yes)Children ,18 at homeAdult child home or in areaEducation (years)Income (in $1,000s)Health (1 to 5, 5 5 excellent)

.86

.61

.5014.1046.424.31

2.7419.18

.70

.69

.48

.5613.8443.534.28

2.4619.27

.67

Parent/in-law characteristicsDistance (hours)Number of adult childrenMother (in-law) only (1 5 yes)Father (in-law) onlyc (1 5 yes)Age (years)Healthd

Coreside with other adult child (1 5 yes)

3.273.59.51.12

74.693.59.16

6.261.86

8.441.01

3.753.47.59.14

76.963.33.10

6.581.94

8.001.02

Measures of contact and assistanceNumber of phone calls per month 7.03 9.55 7.73 10.23Number of visits per monthHours of help per week

4.251.40

7.284.51

5.051.49

8.253.72

N 255 280

aFor continuous variables. bOmitted category: other parents not alive. cOmitted category: both alive and living together.dIf both are alive, age is oldest and health is poorest.

TABLE 2. MEAN CONTACT AND HELP BY GENDER OF RESPONDENT AND RELATIONSHIP

Parents

Men Women

In-Laws

Men Women

Number of phone calls per monthNumber of visits per monthHours of help weekly

6.59a

3.69a

1.66

10.615.862.00

7.43a

4.751.16

4.804.22.97

N 121 141 134 139

aGender difference significant at p , .001 (t test).

instrumental help. Thus, we focus on help in onlyone direction.

Independent variables. Respondent characteristicsinclude employment (1 5 yes, 0 5 no); childrenunder 18 at home (1 5 yes, 0 5 no); adult chil-dren in the home or living in the local area (1 5yes, 0 5 no); years of education; income in thou-sands of dollars; and subjective health (rangingfrom 1 5 very poor to 5 5 excellent).

We also include characteristics of the parent(in-law) or the parent (in-law) couple. We measure

distance from parents as travel time ‘‘using yourusual means of transportation’’ and recode it fromminutes, hours, or days, to hours, for consistency.(We also recode this to a maximum of 1 day toavoid extreme values.) We include a set of dummyvariables indicating whether the mother, father, orboth are alive (the reference category). We includeparent age (in years) and health (as describedabove), coded as oldest parent and worst healthwhen both parents are alive. We include the par-ents’ number of adult children (including the re-spondent or respondent’s spouse), and a dummy

400 Journal of Marriage and Family

TABLE 3. UNSTANDARDIZED REGRESSION COEFFICIENTS FOR VISITING, PHONING, AND WEEKLY HOURS OF HELP TO

PARENTS AND PARENTS-IN-LAW FOR MEN AND WOMEN (STANDARD ERRORS IN PARENTHESES)

Visiting

Men Women

Phoning

Men Women

Key predictorsRelationship (0 5 parents, 1 5 in-laws)Other parent set in areaOther parent set lives away

1.14a (.92).63 (1.14).70 (1.27)

21.79* (.84)1.12 (1.14)1.05 (1.20)

2.14b (1.10)23.90*b (1.44)23.78*b (1.54)

26.30* (1.11)1.84 (1.40).98 (1.58)

Respondent characteristicsEmploymentChildren ,18 at homeAdult children at home or in areaEducationIncome (in $1,000s)Health

2.79 (1.47).54 (1.03).57 (.93)

2.19 (.22)2.29 (.30)

.77 (.62)

2.50 (1.13).54 (.91).06 (1.08)

2.32 (.23)2.36 (.32)2.96 (.81)

22.57 (1.83)21.29 (1.33)2.94 (1.15)2.03 (.22)2.63† (.35)

.77 (.91)

.75 (1.33)

.21 (1.23)2.25 (1.21)2.33 (.27)

.04 (.32)2.66 (.87)

Parent/in-law characteristicsDistanceNumber of adult childrenMother (in-law) onlyFather (in-law) onlyAgeHealthCoreside with other adult child

2.32* (.04)2.25 (.22)1.21 (1.04).01 (1.46).05 (.05)

2.02 (.50)21.92* (.96)

2.30* (.04)2.71* (.22)1.82† (1.00)

2.89 (1.28).05 (.07)

2.12 (.51)21.73 (1.37)

2.34* (.06)2.79* (.29)

.65 (1.22)23.75* (1.26)

.001 (.07)

.37 (.53)2.75† (1.55)

2.40* (.06)21.17* (.28)

.35 (1.32)2.92 (1.69)

.02 (.08)1.02† (.59)3.17 (2.09)

Constant 8.42 (5.77) 14.50 (7.01) 15.22 (7.39) 17.20 (8.81)

N 255 280 255 280

Pseudo d2R .08 .17 .13 .19

aDifference between men and women significant at p 5 .02. bDifference between men and women significant at p , .01.cDifference between men and women significant at p 5 .06. dBased on squared correlation between observed and predictedvalues following generalized estimating equation.

†p , .10. *p , .05.

variable for whether the parent(s) coreside withanother adult child. Finally, we include a set ofdummy variables indicating whether the other setof parents lives in the area, or lives away fromthe area, with other set of parents not alive as thereference category. This will allow us to deter-mine whether there is any sign of competition orhierarchy in contacts with one’s parents comparedto one’s parents-in-law.

Analysis

To test for differences in relationships with parentsand parents-in-law, we create a combined file (n5 535 after exclusion of cases with missing data)that includes one case for each parent (couple) andfor each parent-in-law (couple) that a respondenthas surviving (subject to the exclusions mentionedearlier). Thus, 262 cases represent relations withparents, and 273 represent relations with parents-in-law. Because 166 of the respondents have fulldata for both parents and in-laws and thus con-tribute two cases to the combined file, there is a

potential problem of dependency across cases.Generalized estimating equations (GEEs) providea computationally simple approach for estimatingparameters in such data. Liang and Zeger (1986)introduced GEEs as a method for dealing withcorrelated data within the framework of the gen-eralized linear model. More recently with the GEEapproach, rather than using estimates based ontreating within-cluster responses as independent,one can specify a likely correlation structure (aworking correlation structure), but the standarderrors are adjusted to reflect what actually occursfor the sample data. As a consequence, when cor-relations are modest, all working correlation struc-tures generally lead to similar GEE estimates andstandard errors because the empirical dependencehas a large effect on adjustment of the naı̈ve stan-dard errors (Agresti & Liu, 2001; Spitze, Logan,Deane, & Zerger, 1994). We used the SAS PROCGENMOD to implement the GEE method usingonly the empirical dependence for the sample data(Liang and Zeger’s original derivation of GEE) toadjust the standard errors.

401Social Support to Parents-in-Law

TABLE 3. EXTENDED

Help

Men Women

2.48 (.60).08 (.88).36 (1.44)

21.11* (.38).60 (.48).47 (.49)

21.96 (1.27)2.59 (.77)2.95 (.78)2.02 (.10)

.07 (.17)21.15 (1.04)

.89† (.50)

.06 (.45)2.33 (.56)2.03 (.09)

.004 (.10)2.80† (.46)

2.09† (.05).12c (.18)

1.44* (.61).33 (.51).03 (.03)

2.55* (.28)2.93 (.61)

2.08* (.02)2.28* (.10)

.75* (.34)

.71 (.74)

.003 (.02)

.03 (.23)2.52 (.37)

7.71 (5.06) 6.63 (3.85)

255 280

.08 .07

After discussing mean levels of visiting, tele-phone contact, and help by gender for parents andparents-in-law, we present results of multivariateanalysis for men and women. We test for differ-ences between men’s and women’s coefficients us-ing interaction terms in a combined equation, andsignificant differences are indicated. Descriptivestatistics for the variables in the models are pre-sented in Table 1.

RESULTS

Looking first at gender differences (Table 2), wefind that women talk on the phone and visit withtheir own parents more often than men do, andmen have phone contact with in-laws more oftenthan women do. There are no gender differencesin weekly hours of help. Because we are focusingon a sample that includes many relatively healthyand independent older parents, and the help wemeasure is likely to be occasional (such as tripsto the grocery store or some yard work), it is notsurprising that we fail to find the larger genderdifferences others have reported for caregiving tofrail elderly parents. Thus, we have evidence thatwomen’s kinkeeping role creates more contactwith women’s than men’s parents and that this

kinkeeping does not extend to in-laws. There issome evidence that women may draw husbandsinto contact with their own parents. To test for theexistence of a kin hierarchy, we turn to our mul-tivariate analysis.

Contact

In Table 3 (columns 1–4), we see that women visitwith their parents significantly more often thanthey visit with their parents-in-law. For men, thereis no difference. There is also no effect of thesurvival or local presence of the other set of par-ents on visiting with parents. Other important pre-dictors of frequency of visiting include distance,with similar coefficients for men and women. Foreach 3 hours of travel time, there is one fewervisit per month. Women also see parents less oftenwhen the parent has more adult children, and mensee parents less often when the parent coresideswith another child. Finally, women tend to seewidowed mothers (in-law) more often than mar-ried parent couples.

Turning to frequency of telephone contact (seeTable 3, columns 5–8), we find that women talkwith their own parents significantly more oftenthan with their husbands’ parents, but there is nosuch difference for men. This is consistent withthe pattern of means in Table 2. Another genderedpattern is that for men, having a set of parents inthe local area or even living farther away decreas-es contact with the other set of parents, but thereis no such effect for women. This could suggestthat the presence of parents makes it more difficultfor men to be drawn as closely into their wives’family group and, similarly, that having in-lawsinterferes with the strength of the tie with theirown parents. We tested and found no significantinteraction between these other parent variablesand parent relationship, so it would appear thatboth of these processes may be occurring for men.

We find strong effects of distance on telephonecontact; even though distance might be seen asless of a barrier to telephone contact than to vis-iting in person, the coefficients for both men andwomen are of the same order of magnitude asthose for visiting. Similarly, both men and womentalk with their parents less often when there aremore adult children on that side of the family.Men also talk less on the phone with a widowedfather (in-law), consistent with the notion thatmen are less comfortable with phone contact (seeFischer, 1992). Effects of other control variablesinclude a tendency toward more telephone contact

402 Journal of Marriage and Family

by women with healthier parents and less by high-er income men and those whose parents coresidewith another child.

Assistance Patterns

Turning to patterns of helping (Table 3, columns9–12), we find, again, a significant difference forwomen between hours of help to parents and toin-laws. In contrast, there is no such difference formen (although the difference between these co-efficients is not significant). We find no evidencethat parents and parents-in-law compete for theiradult children’s help. The presence of parents (in-laws) in the area, or elsewhere, has no effect onhelp given to the other set of parents.

As with contact, we find similar effects of dis-tance for men and women (although only margin-ally significant for men). We also find a negativeeffect of sibling size for women, parallel to thosefound for contact. Men provide more help to par-ents in worse health and when a mother (in-law)is the lone survivor, both men and women helpmore hours per week.

DISCUSSION

Previous research has demonstrated that adultchildren often provide crucial social support to ag-ing parents. We have extended the analysis to filialrelations with in-laws by presenting data on twoforms of contact and on weekly hours of assis-tance to parents-in-law. We asked whether thereare gender differences in patterns of intergenera-tional support to parents and to parents-in-law.When we compare mean levels of contact, we findthat women have more contact (both visiting andphone) with their own parents than men have withtheirs. The only significant difference for in-lawsis in the other direction, however: Men have morephone contact with in-laws than do women. Thus,we would conclude that women’s kinkeeping roledoes not extend to relations with their in-laws andthat women’s more central role in intergeneration-al relations is restricted to their own original kin.

Our second question concerned competitionbetween parents and in-laws for married adultchildren’s attention. There are two kinds of evi-dence for men of this kind of competition. First,the fact that men have equal amounts of contactwith parents and parents-in-law suggests that thepull toward the wife’s parents is counterbalancingthe norms that might give higher priority to theirown parents. Second, there is direct evidence of

competition between parents and in-laws formen’s telephone contact; having one set of parentsalive (whether local or not) decreases phone con-tact with the other. Thus, this is further supportthat their own parents do not have the same kindof priority that women’s do in their kin relations.The fact that this effect is specific to phoning mayalso reflect men’s general limitations on willing-ness to spend time on the phone (Fischer, 1992).

These patterns suggest that for women, the re-lationship with their own parents is paramount,close to the top of the hierarchy of obligations,whereas for men this hierarchy is more tenuous.Our data do not tell us through what process thispattern occurs, but they are consistent with Rossiand Rossi’s (1990) expectation that men get drawninto contact with their in-laws through theirwives’ influence. Women may arrange gatheringsto which their husbands accompany them and mayinitiate telephone calls in which they involve theirhusbands. Future work on kinkeeping might morespecifically address whether there are situations inwhich daughters-in-law are more involved in suchactivities, such as in a family with no daughters.It would also be useful to examine these genderdifferences in longitudinal context to determinehow these intergenerational relationships aretransformed with the transition to marriage.

One important limitation of our study involvesracial and ethnic diversity. Although representa-tive of the local area in which it was conducted,there are too few African American respondentsfor separate analysis and even fewer in othergroups. Studies focusing on the relative involve-ment with own and spouse’s kin among membersof different ethnic groups would be a useful con-tribution to the family literature. Furthermore, itwould be interesting to know how Rossi and Ros-si’s (1990) hierarchies of obligations are experi-enced by members of various subgroups in U.S.society.

Finally, these patterns underscore some moregeneral questions regarding the process throughwhich they occur. Researchers tend to view con-tact between older parents and their adult childrenas support from the child to the parent. Althoughthe help we have analyzed certainly could beviewed in that manner, contact may be supportiveor not, and may be initiated by either party. It mayalso be arranged by one family member but in-volve the entire family. Furthermore, parents mayhave a variety of ways to communicate their ex-pectations to adult children regarding contact orhelp and thus may influence even the help they

403Social Support to Parents-in-Law

receive. Our data do not tell us about the contextin which these contacts occur, nor their meaningfor the people involved. Thus, we hope that futureresearchers on these relationships will gather suchinformation through both more detailed surveyquestions and in-depth interviews about the pro-cess through which family members become in-volved in various forms of intergenerational con-tact and assistance. It would be useful to knowhow women experience their role of kinkeeperand to learn how men experience the tugs andpulls that are only hinted at by the kind of analysiswe describe here.

NOTE

Data used in this article were gathered by John Loganand Glenna Spitze under Grant AG-06831–01 from theNational Institute on Aging. We are deeply grateful toGlenn Deane for statistical advice.

REFERENCES

Agresti, A., & Liu, I. (2001). Strategies for modeling acategorical variable allowing multiple category choic-es. Sociological Methods & Research, 29, 403–434.

Allen, K. R., Blieszner, R., & Roberto, K. A. (2000).Families in the middle and later years: A review andcritique of research in the 1990s. Journal of Marriageand the Family, 62, 911–926.

Di Leonardo, M. (1987). The female world of cards andholidays: Women, families, and the work of kinship.Signs: Journal of Culture and Society, 12, 440–453.

Fischer, C. (1992). America calling: A social history ofthe telephone to 1940. Berkeley: University of Cali-fornia Press.

Fischer, L. R. (1983). Mothers and mothers-in-law.Journal of Marriage and the Family, 45, 187–192.

Goetting, A. (1990). Patterns of support among in-lawsin the United States: A review of research. Journalof Family Issues, 11, 67–90.

Kivett, V. R. (1989). Mother-in-law and daughter-in-lawrelations. In J. A. Mancini (Ed.), Aging parents andadult children (pp. 17–32). Lexington, MA: Lexing-ton Books.

Komarovsky, M. (1964). Blue-collar marriage. NewYork: Random House.

Liang, K., & Zeger, S. L. (1986). Longitudinal dataanalysis using generalized linear models. Biometrika,73, 13–22.

Logan, J. R., & Spitze, G. D. (1996). Family ties: En-during relations between parents and their grownchildren. Philadelphia: Temple University Press.

Lopata, H. Z. (1999). In-laws and the concept of family.Marriage and Family Review, 28, 161–172.

Merrill, D. M. (1993). Daughters-in-law as caregiversto the elderly. Research on Aging, 15, 70–91.

Rossi, A. S., & Rossi, P. H. (1990). Of human bonding:Parent–child relations across the life course. NewYork: Aldine de Gruyter.

Serovich, J. M., & Price, S. J. (1994). In-law relation-ships: A role theory perspective. International Jour-nal of Sociology of the Family, 24, 127–146.

Silverstein, M., & Litwak, E. (1993). A task-specifictypology of intergenerational family structure in laterlife. The Gerontologist, 33, 258–264.

Spitze, G., Logan, J., Deane, G., & Zerger, S. (1994).Adult children’s divorce and intergenerational rela-tionships, Journal of Marriage and the Family, 56,279–293.

Stoller, E. P. (1983). Parental caregiving by adult chil-dren. Journal of Marriage and the Family, 45, 851–858.

Waite, L. J., & Harrison, S. C. (1992). Keeping in touch:How women in mid-life allocate social contactsamong kith and kin. Social Forces, 70, 637–655.

Walker, A. J. (2000). Refracted knowledge: Viewingfamilies through the prism of social science. Journalof Marriage and Family, 62, 595–608.