INTL. J. AGING AND HUMAN DEVELOPMENT, Vol.

58(2) 127-146, 2004

REVISITING THE GENERATION GAP: EXPLORING THE RELATIONSHIPS OF PARENT/ADULT-CHILD DYADS*

ADAM SHAPIRO University of North Florida

ABSTRACT

There is some evidence that older parents and their adult children may evaluate their relationships with each other in different ways. To date, we know little about what may account for these discrepancies. This investigation compares the perceptions of intergenerational solidarity among 2,590 adultchild/older-parent dyads from the National Survey of Families and Households. Further, this study examines a social structural model to test the relative contribution of individuals sociodemographic and social structural positions to the correspondence between generational perceptions of solidarity. The findings indicate that there is a high degree of disagreement between how adult children and their parents view their relationship. Parents are more likely to report greater relationship quality, while children report greater contact and exchanges of assistance. While a number of variables influence the correspondence between generations, the strongest and most consistent predictors are sex, age, childs marital status, and residential proximity.

INTRODUCTION Parent-child relationships are considered to be among the most important relationships individuals have. These relationships are a primary contributor to the psychological well-being of both generations (Umberson, 1992); they are

*A previous version of this article was presented at the 1998 American Sociological Association Annual Meeting in San Francisco. 127 2004, Baywood Publishing Co., Inc.

128 / SHAPIRO

associated with the exchange of various forms of assistance (Hogan, Eggebeen, & Clogg, 1993); and they predict the likelihood of a childs assumption of caregiving duties (Ingersoll-Dayton & Antonucci, 1988). Two fundamental demographic changes have made the roles that older parents and their adult children play as joint informal mechanisms of support more salient in recent years. First, the size of current and future cohorts of older persons will challenge public and private organizations ability to provide formal support to older adults. Second, high divorce rates and low fertility rates ensure that fewer children will be relied upon to provide informal support to their older family members. Given the importance of intergenerational solidarity, researchers have been investigating the extent to which this solidarity exists in the United States and abroad. Recently, scholars have discovered that generations may perceive intergenerational solidarity very differently (Aquilino, 1999; Lynott & Roberts, 1997). The overarching finding in this body of research is that there is a generation gap. That is, parents tend to have a more positive outlook on their relations with their adult children than the children do. Yet, to date, we know little about why this generation gap exists. Research on intergenerational relationships has several noteworthy limitations. First, while many studies utilize data collected from parent-child dyads (cf. Parrott & Bengtson, 1999) few studies gauge intergenerational relations from the perspectives of both generations. To date, we know little about the degree to which adult children and their parents correspond on measures of intergenerational relations (Albrecht, Coward, & Shapiro, 1997; Roan, Hermalin, & Ofstedal, 1996). Second, while theoretical paradigms have been advanced to understand divergent reports of affective solidarity, there has not yet been a systematic examination of divergent reports of other dimensions of intergenerational relationships. The present study is designed to examine both parents and their adult childrens reports of multiple dimensions of intergenerational solidarity using a national sample. More specifically, I suggest that social structural position may serve as a significant predictor of parents and adult childrens differential or corresponding reports. Using paired data from the National Survey of Families and Households, I proposed two primary research questions. First, how do parents and their children correspond in self-reports of intergenerational relations? Second, do individuals sociodemographic characteristics affect the correspondence between adult childrens and their parents reports? BACKGROUND All too often in the study of intergenerational relations, gerontologists analyze the perspective of only one member of the parent-child dyad. While studies of multiple generations often report some agreement between generations, there is evidence to suggest that parents and their adult children often evaluate their

REVISITING THE GENERATION GAP /

129

relationships with each other in different ways. This rather modest literature generally suggests that parents are more likely to perceive their relations with children in a positive light (Albrecht et al., 1997; Aquilino, 1999; Bond & Harvey, 1991; Giarrusso, Stallings, & Bengtson, 1995; Rossi & Rossi, 1990). However, much of this generational difference is examined within the context of affective aspects of parent-child relationships. We still know relatively little about parent-child reporting differences in other dimensions of parent-child relationships such as intergenerational exchanges and contact. Bond and Harvey (1991) found that older parents report greater intergenerational contact. However, in a recent study of parent-child dyads in Taiwan, Roan and colleagues (1996) maintained that children tended to over-report contact and support relative to their elderly parents. Thus, there is a need to clarify the reporting of intergenerational solidarity among older parents and their adult children in the United States. Even less is known as to why parent-child discrepancies in reporting of intergenerational relations exist. As one of the most widely cited explanations for parent-child discrepancies in reports of intergenerational relations in the United States, the intergenerational stake hypothesis (Bengtson & Kuypers, 1971; Giarrusso et al., 1995) suggests that perceptions of intergenerational relations may be systematically skewed by the motivations of the particular respondent. This hypothesis purports that older parents view their effective relationships with their children much more positively than their children do. These differences arise due to the fact that motivations and level of investment of each generation vary at different developmental periods. Younger generations report being less close to older family members because of their desire to achieve independence and their lessened relative investment in their parents. Older generations seek generativity, and have both a greater desire to maintain continuity between generations and a greater investment in their children (rather than vice-versa). The intergenerational stake model also proposes that as generations age concurrently they become more similar. This would suggest that older parent-child dyads correspond in their reports of intergenerational solidarity to a greater extent than do younger parent-child dyads. Developmentally, the interests and motivations of parents and children become more similar as they age. Three recent studies support the intergenerational stake hypothesis. Giarusso and her colleagues (1995) support the notion that the intergenerational stake extends across the life course and is not confined to any particular developmental period. Similarly, both Aquilino (1999) and Lynott and Roberts (1997) find that older generations report greater closeness than do younger generations. It is still unclear, however, as to the degree to which the intergenerational stake hypothesis holds for other dimensions of intergenerational solidarity, such as instrumental and emotional support.

130 / SHAPIRO

TOWARD A THEORY OF DISCREPANCY IN REPORTED INTERGENERATIONAL SOLIDARITY Although the intergenerational stake hypothesis remains the predominant explanation for generational reporting differences, it fails to incorporate social structural factors into explanations of reporting differences. Additionally, this perspective does not account for generational discrepancies in forms of solidarity other than effective solidarity. As an alternative, a microstructural perspective (Risman & Park, 1988) suggests that individuals enactment and perception of intergenerational roles may be influenced by the social structural contexts (i.e., differential opportunities and expectations) that frame individuals lives. In other words, social structure both enables and constrains perceptions and evaluations of intergenerational relationships (see Suitor, Pillemer, Keeton, & Robison, 1995). Recent theoretical work has examined the social structural underpinnings of intergenerational relationships in later life. The intergenerational ambivalence perspective suggests that societies and the individuals within them are ambivalent about relationships between parents and children in adulthood (Luescher & Pillemer, 1998, p. 414). This ambivalence can be directly traced, according to Luescher and Pillemer, to contradictions and countervailing forces in society, most notably the battle between desires for personal autonomy and dependence. Thus, parent-child relationships in adulthood are viewed as containing irreconcilable contradictions at a social structural level, such as statuses, roles, and norms (Luescher & Pillemer, 1998). Discrepancies between parents and childrens reports of intergenerational relations, then, may be attributable to the different social structural positions of parents and children in our society. The logical transposition of this suggests that parents and children will have more congruous reports if they are given similar expectations, resources, and positions in society (Risman & Park, 1988). To elaborate on this, Burt (1982) suggests that actors who have structurally equivalent statuses have interdependent interests. It is this interdependence, then, that encourages similar perceptions and evaluations of ones relationships. One way in which structural forces may impinge upon individuals perceptions of intergenerational solidarity is the creation of inequalities within parent-child dyads. Social structural inequalities may create dependency within parent-child dyads when one generation has fewer opportunities (i.e., economic, social, etc.) than the other. Recent research has suggested that the dependence of one generation on the other may generate perceptual differences of intergenerational solidarity between generations. Aquilino (1999) suggests that children have more positive outlooks on parent-child relations when the parents have high educational attainment and the children have low educational attainment. In this case, an adult child with a relatively limited education may be more reliant upon better-educated parents. This reliance may translate into more favorable

REVISITING THE GENERATION GAP /

131

impressions of the parent-child relationship on the part of the child. Similarly, age and singlehood, two indirect sociodemographic indicators of dependency, may predict generationally skewed perceptions of intergenerational relations. Albrecht and colleagues (1997) have found that parents over the age of 70 and unmarried parents are significantly more likely than their adult child to perceive a potential disruption in coresidence with their adult child as negative. Thus, we would expect that variables associated with dependency such as age, health status, education, and singlehood will predict the overreporting of intergenerational solidarity. On the other hand, it is expected that when parents and their children occupy similar social structural positions, their perceptions of intergenerational solidarity will be more alike. Sex may influence the degree of structural similarity between generations, because sex-linked expectations and opportunities are such an important component of social structure. The mother-daughter relationship is expected to have the highest degree of correspondence in reports of intergenerational relations. Research has noted the particularly high solidarity in the motherdaughter relationship (Lynott & Roberts, 1997; Rossi & Rossi, 1990). In general, daughters have higher levels of contact with parents and provide more support to their parents than do sons (Coward & Dwyer, 1990). Additionally, living arrangements of parents and children may serve as a means of structural similarity. For example, we would expect the reports of coresident dyads to correspond more closely than those of non-resident dyads because of the greater interdependence inherent in coresiding dyads. One of the more powerful structural forces exerted on individuals are norms of filial obligation; culturally-bound and highly structured behaviors in which parents and children are expected to engage. Norms of filial obligation may affect generational reporting differences. When cultural expectations of filial obligation are high for a particular generation, these persons are likely to present a more favorable and culturally affirmative response. While kinship norms are highly structured, they may be modified by certain contextual factors and life events (Rossi & Rossi, 1990). Rossi and Rossi identified a number of variables that account for variation in kinship obligations: age, marital status, and education. With regard to age, Rossi and Rossi find that kinship obligations are stronger for older generations (similar to the assertions of the generational stake view). We would therefore expect that parents are more likely than their children to overreport intergenerational solidarity, particularly as the parents age. While obligations to the unmarried in general are heightened, marital disruption weakens kinship obligations. For example, divorced persons perceptions of solidarity would not be overreported as their obligations decline. However, having an unmarried parent or child may trigger an overreporting of solidarity. Finally, education increases kinship obligations, and we would expect that education is positively associated with overreporting. In sum, extant studies clearly point to two general conclusions. First, while the few studies of parent-child dyads find a high degree of correspondence on

132 / SHAPIRO

measures of intergenerational relations, adult children and their older parents often perceive and report their relationships with each other in different ways. Second, while developmental/intergenerational stake models exist to predict correspondence between parents and childrens perceptions of relations, alternative explanations of the variability between childrens and parents reports have not received much empirical attention. In this study, I have directly tested a microstructural model by analyzing data from a large national survey of older-parent/adult-child dyads. By investigating the perspectives of both adult children and their older parents, I hope to clarify the ways in which older parents and their adult children are bound together. METHOD Data Data for this analysis are from the second wave of the National Survey of Families and Households (NSFH; Sweet & Bumpass, 1996). The NSFH is a two-wave national probability sample of non-institutionalized persons aged 19 and older originally conducted in 1987 (N = 13,017), with a follow-up between 1992 and 1994 (N = 10,007). The NSFH oversamples minorities, single-parent families, and step-families. Data were collected through face-to-face interviews. At various segments of the interviews, respondents were asked to complete a self-administered questionnaire in order to facilitate the collection of sensitive information. The second wave of the NSFH includes computer assisted re-interviews with 10,007 of its original respondents. The NSFH-2 contained a total of 5,356 main respondents who had at least one living parent. Of these, interviews were conducted with a randomly-selected parent of 3,347 main respondents (i.e., adult children). In the cases where one of the parents was deceased, the living parent was interviewed. Of these parents, 757 (22.6%) were eliminated because the parents did not record the main respondent as one of their children. NSFH administrators have no explanation as to why this omission occurred. It may be that, during the interview, some parents were under the impression that the interviewer was interested in their other children rather than the child who had been interviewed already. The final sample thus consists of 2,590 matched adult-child/older-parent dyads, and analyses are performed using data from both members of each dyad at NSFH-2. Table 1 presents descriptive statistics for the study sample. Less than half of the eligible main respondents lacked complete parent interview data. Given this, several analyses were performed to test for sample selection bias. Analyses that compared those in the final sample with the 757 dyads who completed a parent interview but were eventually eliminated due to missing data find no evidence for a systematic attrition effect. No differences were found on measures of parent-child relationship quality or parents and childrens

REVISITING THE GENERATION GAP /

133

Table 1. Means and Standard Deviations for Independent Variables M

SD
9.55 .91 .49 5.84 Dyad characteristics Distance (miles) Coresiding Mother-Daughter Mother-Son Father-Daughter Father-Son Family size

M
299.26 .08 .38 .27 .20 .15 3.73

SD
768.60 .27 .42 .44 .40 .35 2.10

37.63 13.64 .63 .19 .16 .02

8.30 2.44 .48 .39 .37 .14

sociodemographic characteristics. There was also concern as to the exclusion of dyads that did not complete the parent interview at all. These analyses suggest a systematic selection bias between main respondents (i.e., adult children) in the final sample and those main respondents whose parent did not complete the parent interview. Main respondents whose parent completed the parental interview were healthier, better educated, and were less likely to be Caucasian than those who did not complete the parental interview. Given the nonrandom nature of sample selection discussed above, a two-stage procedure to correct for selection bias proposed by Heckman (1979) was implemented in the present analysis. The first stage involved determining the presence of selection bias and the computation of a predicted lambda value for each respondent. Lambda represents the probability that each respondent will not be included in the final subsample, while accounting for respondents individual characteristics (Miller & Wright, 1995). The second stage of this procedure involved adding the lambda value of each respondent to the regression models. This procedure removed the bias in the equation as a result of specification error, and produced unbiased parameter estimates. Measures
Dependent Variables

The dependent variables represent several dimensions of intergenerational relations. These variables include: a) frequency of contact either in person or by

134 / SHAPIRO

telephone or mail; b) global relationship quality; and c) exchanges of assistance with housework, shopping, or errands, and emotional support. All dependent variables were asked of both the adult child and the parent. Global relationship quality was assessed with a single item which asked the respondents to evaluate their relationship with their child/parent on a scale from 0 (low) to 10 (high). Contact between parents and adult children was assessed by asking both members of each parent-child dyad two questions about their contact with their child/parent. Respondents were asked, During the last 12 months, how often did you see (child/parent)? and During the last 12 months, about how often did you talk on the telephone or receive a letter from (child/parent)? Responses ranged from not at all to more than once a week. Three measures of intergenerational exchanges were assessed by asking each respondent whether they had received or given various forms of assistance from/to the other member of the dyad during the last month. Housework refers to helping with housework, yard work, and/or car repairs. Errands refers to helping with transportation, shopping, or errands. Emotional support refers to exchanging advice, encouragement, moral, or emotional support. Each of the intergenerational exchange variables is dummy coded (0 = not received/not given, 1 = received/given) and grouped into three categories. While many studies combine housework and errands into a single instrumental exchange, category, I left them separate because, while housework and errands both represent forms of instrumental assistance, the locus of activity varies between them (i.e., within/around the household and outside of the household).
Independent Variables

The independent variables reflect a number of social structural and sociodemographic variables that have been shown in previous research to influence intergenerational relationships (i.e., Aquilino, 1999; Eggebeen, 1992). Characteristics of the parentsFour characteristics of the parents were analyzed to determine the effect on parent-child disagreement: age (in years), self-reported health, completed years of education, and marital status. The selfreported health measure asked the respondent to compare his or her health to others of the same age. The variable was coded on a 5-point scale ranging from 1 (very poor) to 5 (excellent). Marital status is a dichotomous variable coded 1 for married and 0 for unmarried. There were too few cases of never-married and divorced parents to include separate categories. Characteristics of the adult childrenThree characteristics of the children were analyzed: age (in years), completed years of education, and marital status. Marital status contains four categories: married, divorced/separated, never married, and widowed.

REVISITING THE GENERATION GAP /

135

Characteristics of the dyadFour characteristics of the parent-child dyad were also examined: residential proximity, coresidence, family size, and sex composition. Residential proximity reflects the distance between parents and non-coresident childrens residences in miles (log). Family size reflects the total number of the parents biological children. Finally, sex composition of the dyad is categorized into four groups on the basis of parents and childrens sex: mother-daughter, mother-son, father-daughter, and father-son. Measuring the correspondence between parents and adult childrenBecause visits, telephone calls/letters, and relationship quality are continuous, ratio variables were calculated to determine the extent to which dyads reports of visits, telephone calls/letters, and relationship quality corresponded. The ratio consisted of dividing parents responses by childrens responses. Thus, a value of 1 equals correspondence, while a value greater than 1 indicates parental overreport, and a value less than 1 indicates child overreport. With respect to intergenerational exchanges, correspondence was determined when one generation reported giving the other generation reported receiving. Discordance occurred when one generation reported giving and the other generation did not report receiving. RESULTS Comparing Parents and Childrens Reports of Intergenerational Solidarity Table 2 presents the results of analyses that test whether parents will be more likely to overreport intergenerational solidarity than their adult children. In this analysis, paired sample t-tests were performed for each dyad on the means of each dependent variable. Contrary to the intergenerational stake hypothesis, parents reports are not greater than those of their children in all cases. As suggested by the intergenerational stake hypothesis, parents do report significantly greater affective relations such as relationship quality and giving emotional support. Parents also are more likely to report receiving assistance with housework and errands. However, adult children report a significantly greater degree of contact (visits and calls/letters), and are more likely to report giving assistance with housework and errands. Table 3 shows the general patterns of correspondence and discordance in parents and childrens reports of intergenerational relations. The results indicate that patterns of agreement vary across the outcomes studied. First, roughly 20% of the dyads correspond with regard to contact. However, adult children were much more likely than their parents to overreport contact by telephone, letter, or in person. Almost 70% of dyads consist of a child overreporting the frequency of visits, calls, and letters. Second, supporting the intergenerational stake hypothesis

136 / SHAPIRO

Table 2. Paired T-Tests and Correlations of Parents and Childrens Reports of Intergenerational Solidarity Parent Visits Calls/Letters Relationship quality Giving help with housework Giving help with errands Giving emotional support Receiving help with housework Receiving help with errands Receiving emotional support
*p < .05. **p < .01. ***p < .001.

Child 4.12 4.84 7.95 .23 .24 .47 .11 .15 .47

Correlation .79*** .52*** .30*** .05* .04* .03 .15*** .08*** .08***

t 42.37*** 35.70*** 28.30*** 17.89*** 11.05*** 11.00*** 4.20*** 3.70*** 1.94

3.34 4.01 9.13 .07 .13 .62 .14 .19 .50

(Giarrusso et al., 1995), parents were more likely to overreport the quality of the parent-child relationship. Parents overreported the quality of the parent-child relationship in 59.4% of dyads, compared to 13.7% of dyads in which children overreported relationship quality. As seen in the middle of Table 3, the vast majority of parent-child dyads correspond in their reports of exchanging instrumental forms of assistance. Most dyads agree that they did not exchange housework or errands within the last month (67.9% and 61.6%, respectively). In examining patterns of discordance in reports of instrumental assistance, the results indicate that children were more likely to overreport giving assistance than were their parents. Dyads in which children reported giving and parents did not report receiving housework and errands comprise 15.9% and 14.4% of the sample, respectively. However, dyads in which parents reported giving and children reported not receiving housework and errands comprise 3.2% and 5.2% of the sample, respectively. In contrast to exchanges of instrumental support, there is much less agreement within dyads about exchanges of emotional support. Slightly more than 15% (15.6%) of dyads agree that they do not exchange emotional support. However, when dyads reports correspond, exchanges of support are most likely to flow from parent to child (15.9%). In sum, the results from Tables 2 and 3 provide firm evidence that there are generational differences in reporting of intergenerational relations. The correlates of these intergenerational differences will be addressed in Table 4.

Table 3. Patterns of Correspondence in Parents and Childrens Reports of Intergenerational Solidarity Visits 23.9 19.9 26.8 0.0 0.5 0.6 67.9 59.4 13.7 8.6 71.5 6.4 69.7 0.0 1.1 1.8 61.6 Calls/ letters Errands Relationship quality Housework Emotional support 0.0 15.9 3.3 15.6

Correspondence (%)

Parent overreport (%)

Child overreport (%)

Correspondence Child gives and parent receives (%) Parent gives and child receives (%) Child does not give and parent does not receive (%) Parent does not give and child does not receive (%)

Parent overreport Parent gives and child does not receive (%) Parent receives and child does not give (%)

3.2 6.1

5.2 8.8

13.7 23.1

Child overreport Child receives and parent does not give (%) Child gives and parent does not receive (%) 2305

2305

2494

5.8 15.9 2591

7.1 14.4 2591

6.9 21.4 2591

REVISITING THE GENERATION GAP

/ 137

138 / SHAPIRO

Table 4. OLS Regression Predicting Correspondence on Visits, Calls/Letters, and Relationship Quality Visits Parental characteristics Age Health Education Married Child characteristics Age Education Divorceda Never marrieda Widoweda Dyad characteristics Distance Coresiding Family size Mother-sonb Father-daughterb Father-sonb Lambda N R2
aMarried

Calls/letters .000 .019 .004 .005

Relationship quality .000 .033* .001 .015

.002 .002 .001 .035*

.001 .001 .012 .031 .041

.003 .008 .027 .037 .032

.001 .013* .027 .022 .272**

.004 .011* .015 .060*** .057** .060 2304 .017

.012* .011 .031 .039 .092** .035 2294 .022

.004 .094 .000 .034 .080* .099** .015 2493 .020

is the omitted category. *p < .05. **p < .01. ***p < .001.

is the omitted category.

The Correlates of Parents and Childrens Reports To determine the predictors of discrepant and corresponding reports of intergenerational contact and relationship quality, a series of regression models is presented in Tables 4 and 5. In Table 4, OLS regression models are estimated with each of the dependent variables as a ratio of parents reports relative to childrens reports. In Table 5, multinomial regression models are estimated that predict the probability of either parent or child overreporting. Each dependent variable is regressed on parental characteristics, childrens characteristics, dyad characteristics, and a selection variable (lambda).

Table 5. Multinomial Logit Regression Predicting Correspondence on Intergenerational Exchanges

Parental characteristics Age Health Education Married .008 .197 .055 .001 .056** .026 .396 .024 .394 .001*** .955** .149 .408 .478* .205 .656 2590 506.40*** 2590 484.71*** .004*** .652*** .014 .061 .498** .230 .254 .001*** .208 .201** .514* .011 .370 .500 .001*** 1.104*** .018 .236 .722*** .530 .365 .048*** .043 .352* .948*** .739 .061*** .048 .210 .079 .192 .024* .013 .504** .935*** .255 .007 .011 .704*** .703*** .212 .000 .441* .010 .511*** .365*** .470*** .128 2590 247.35*** .043*** .101 .005 .232 .008 .123 .006 .024 .029** .007 .010 .604*** .021* .016 .000 .110

.032*** .054 .022 .176 .009 .024 .450*** .519*** .000 .000 .137 .020 .001 .338* .374* .085

Child characteristics Age Education Divorceda Never marrieda Widoweda

Dyad characteristics Distance Coresiding Family size Mother-sonb Father-daughterb Father-sonb Lambda

REVISITING THE GENERATION GAP

N Model chi-square

aMarried

is the omitted category. bMother-daughter is the omitted category. *p < .05. **p < .01. ***p < .001.

/ 139

140 / SHAPIRO

It is also plausible that discrepant reports may result from temporally asynchronous interviews of parents and children. Of particular concern are those questions that are temporally bound, such as whether intergenerational exchanges occurred within the 30 days preceding the interview. I examined this possibility in ancillary analyses (not shown) using the interview dates provided by the NSFH for both parents and children. The median time between interviews was two months, and the modal discrepancy was one month. There were no significant correlations between temporal asynchronicity and parent-child concordance on any of the outcome measures. In fact, the largest bivariate correlation coefficient was .02. Furthermore, when time between interviews was added to the regression models, the substantive results were unchanged. The results in Table 4 indicate that both parents and childrens characteristics are weakly associated with patterns of correspondence between generations. There are few significant or consistent effects of parental marital status, parental and child education, or health on parent-child correspondence. While childrens marital status has little influence on correspondence, the results suggest that when parents are unmarried, they are more likely to overreport visits. The residential proximity of parents and children is an important predictor of reporting differentials (Tables 4 and 5). Distance is positively associated with child overreporting of telephone calls and letters. Distance between parents and children also increases the probability of agreement with respect to exchanges of housework and errands, primarily because distance makes it difficult to provide these kinds of exchanges. Coresident dyads were more likely to overreport exchanges of assistance than their nonresident counterparts. This finding suggests that the disparity between parents and childrens reports of affective relations will increase when parents and children live apart. In contrast, coresidence increases conflicting reports of exchanges of assistance. The results clearly show that fathers are most likely to overreport intergenerational solidarity. Fathers were more likely than mothers to overreport the frequency of telephone calls/letters, visits, and relationship quality, regardless of the childs sex. With regard to giving assistance and support, fathers in father-daughter dyads were significantly more likely to overreport their contribution of housework than were mother-daughter dyads. Mothers in motherdaughter dyads were the least likely to overreport giving emotional support to their child. Examining children in father-child dyads, daughters were less likely to overreport giving assistance with housework, and both daughters and sons were less likely to overreport giving assistance with errands and emotional support than were daughters in mother-daughter dyads. The effect of age on reporting discrepancies is mixed. In contrast to the intergenerational stake hypothesis, there are no effects of parents or childrens age on contact or relationship quality. However, parents age is an important predictor of child overreporting. As seen in Table 5, parents age is positively associated with children reporting giving more assistance and support to their

REVISITING THE GENERATION GAP /

141

parents. On the other hand, childs age is negatively associated with child overreports of giving help with housework or errands. Few effects of parents marital status on reporting discrepancies were evidenced. Children were significantly more likely to overreport visits when their parent was married, and children with an unmarried parent were more likely to overreport giving assistance with errands than were children with a married parent. However, we can generalize from Table 5 that children tend to overreport their contributions of assistance and support when they are unmarried. Compared to married children, divorced or never-married children consistently overreport their assistance with housework, errands, and emotional support. DISCUSSION This investigation was designed to explore both the extent to which adult childrens and older parents reporting of intergenerational solidarity differed, as well as the correlates of these differential reports. Many scholars have argued that perspectives of both generations must be examined, but few studies report such data. The second wave of the NSFH (n = 2,590) provided an opportunity to study this question from the perspective of both generations using nationally representative data. The findings from the present study certainly support the call from researchers for the use of multiple perspectives in family research. How Do Parents and Their Children Correspond in Self-Reports of Intergenerational Relations? The results of this study suggest that there is generally a high level of agreement between how parents and children view their relationships with each other. Consistent with Roan and colleagues (1996) findings, roughly 60% of parentchild dyads agreed on exchanges of instrumental forms of assistance and contact. For optimists, this finding suggests that the analysis of data from a single informant may not be as skewed as one might think. However, the level of agreement is far too low to consider ruling out the necessity of multiple perspectives in family gerontology. On the other hand, adult childrens and their parents reports of their relationships are often discordant. For example, fewer than 35% of parent-child dyads agreed in their reports of exchanges of emotional support. There is also some confirmation of the first hypothesis of this study based on the intergenerational stake hypothesis (Giarrusso et al., 1995). Parents in the present study tended to overreport the quality of both the parent-child relationship and then giving and receiving of emotional support, both of which reflect the affective nature of parent-child solidarity. However, the findings herein suggest that the intergenerational stake hypothesis only holds for measures of affective solidarity. In the present study, adult children were likely to overreport both the

142 / SHAPIRO

amount of contact and the exchanges of assistance they had with their parent, a finding similar to that of Roan et al. (1996). Rossi and Rossi (1990) argue that this finding may suggest some form of frustrated altruism on the part of parents wherein parents reduce their exchanges of assistance in awareness of the childs need for independence. This, Rossi and Rossi suggest, may contribute to the underestimation of the parents share of intergenerational exchanges. What Factors Predict the Correspondence between Adult Childrens and Their Parents Reports? In general, sociodemographic characteristics of parents and children have a modest influence on parent-child correspondence. The variables in the multivariate analyses account for only a small proportion of the variance in reporting differentials. However, several patterns in the data are worth noting. Within the realm of individual characteristics, parents and childrens age were strong predictors of discordance, particularly with respect to intergenerational exchanges. Parent age was positively associated with the probability of child overreporting for all assistance variables, and child age was negatively associated with child overreporting of exchanges of housework and errands. These findings do not support the claim of the intergenerational stake hypothesis that, as children and parents age, their reports of intergenerational solidarity converge as their developmental needs become more similar. However, from a microstructural perspective, children with older parents may feel strong cultural pressures to provide assistance to their parents as advanced age increases the probability of their needing assistance. On the other hand, younger children may overreport because of their dependence on older parents, as dependency may create an overcommitment to a particular relationship. It is still unclear as to why age did not affect other outcomes such as relationship quality and contact, or why age did not influence parental overreporting. Marital status was a consistent and strong correlate of adult children overreporting their provision of assistance and emotional support to their parents. In particular, divorced and never-married adult children were significantly more likely than their married counterparts to overreport giving assistance and support. This finding is in contrast to the expectation that non-married persons, particularly the divorced, would be less likely to overreport providing assistance and support for their parent or child due to a weakening of kinship obligations (Rossi & Rossi, 1990). One possible explanation of this finding is that these divorced and never-married adult children may be more dependent upon their parents than married children. Adult children often become more reliant on their parents after crises in their lives such as divorce (Soldo, Sharma, & Campbell, 1984; Wolf, 1984). As discussed previously, this reliance may trigger overreporting.

REVISITING THE GENERATION GAP /

143

Dyad-level variables were more consistently related to parent-child disagreement in these analyses. The sex of dyad members was an important predictor of parent-child reporting. With regard to sex, parent-child agreement is most likely achieved when parents and children are of the same sex, and when the parent is female. This finding supports the microstructural argument that sex similarity, particularly in mother-daughter dyads, may breed similarity in the perception of parent-child solidarity. The finding that mother-child dyads are most likely to correspond supports the findings of Lynott and Roberts (1997), who determined that females perceive less of a generation gap than do males. Contrary to the predictions of microstructural theory, father-son dyads were not consistently more likely to report concordance than father-daughter dyads. Rather, these findings suggest that fathers have an overly optimistic view regarding their relationship with their children. This corroborates Seltzer and Brandreth (1995), who found that fathers report higher levels of involvement with children than found in mothers reports of father involvement. Why are fathers more likely to overreport their solidarity with adult children? Recent scholarship on fatherhood has pointed to a contradiction between conflicting cultural ideologies of fatherhood and social structure (Daly, 1994). While fathers are increasingly called upon to participate in their childrens lives, they may be restricted from doing so by structural barriers such as mens work roles, greater mobility, and a lower probability of coresidence. It is plausible that fathers are overreporting their solidarity as a response to changing cultural ideologies about fathering. An important finding herein was the effect of residential proximity on parentchild reporting discrepancies. The results suggest that, as the distance between parents and childrens residences increases, children become more likely to overreport contact with their parent by phone or letter. As suggested by Burt (1982), residential distance may directly relate to discordant perceptions of intergenerational contact because of the lack of structural interdependence in these dyads. Additionally, since adult children often have multiple demands placed upon them from work and family, children who live farther away may have substantially more structural barriers to maintaining contact with their older parents. As such, they may be more likely to take note of the trouble they may experience to maintain contact with their older parents. Distance is also negatively associated with parent and child overreports of intergenerational exchanges of assistance. This finding is most likely due to the fact that non-resident dyads provide very little help with errands and housework, thus most of these dyads have little to disagree about. Finally, coresident dyads were more likely to overreport their share of intergenerational exchanges than were those who live apart. It is plausible to suggest that residential proximity may be an indirect measure of intergenerational conflict that may manifest in reporting discrepancies. As Uhlenberg and Cooney (1990) have suggested, greater interaction between the generations, as is the case in

144 / SHAPIRO

coresident households, may increase the opportunities for developing intergenerational conflict. Limitations and Directions for Future Research Despite the uniqueness of the NSFH in enabling the examination of multiple respondents within a large national survey, it is not without its limitations. A substantial portion of those eligible to complete the parent interview did not do so, or did not provide sufficient information to be included in the analysis. Such a problem is inherent in any research in which the data are gathered from two independent sources (Albrecht et al., 1997). Although a selection variable was included in the multivariate analyses to account for these selection effects, the possibility that those dyads with poor relationships or in which the parent had substantial physical and psychological impairments were less likely to be included in the final subsample cannot be eliminated. Thus, because of the limits of the data, the findings of this research should be interpreted with caution. As intergenerational relations become more important with the graying of society, the need to examine multiple perspectives in family gerontology has also become more important. The present study provides definitive evidence that there is indeed a generation gapolder parents and their adult children often view their relationship in very different waysin American families. This conclusion should alert social scientists that existing methods of ascertaining intergenerational solidarity may be insufficient. If multiple family members are not corroborating each others reports, this would suggest that future research should rethink the operationalization of intergenerational solidarity. Moreover, these findings suggest that some form of intergenerational ambivalence may be occurring among parent-child dyads in later life (see Luescher & Pillemer, 1998). Future research should consider the ambivalence perspective in an attempt to understand the larger social structural context in which parent-child relationships are located. I have also argued that existing frameworks for understanding the often discordant reports of older parents and their children provide an incomplete understanding of the multiple dimensions of intergenerational solidarity and the dynamics within families. While structural explanations contribute to our understanding of parents and childrens intergenerational relations, they do not account for a great deal of variation in differences of perception between older parents and their adult children. Rather, psycho-social processes that capture the dynamic aspects of intergenerational relations may be equally or more important in understanding these relationships (Albrecht et al., 1997). Future research should address the quality, process, and dynamics of intergenerational relationships with the recognition that multiple perspectives are an essential tool toward this end.

REVISITING THE GENERATION GAP /

145

ACKNOWLEDGMENTS The author would like to express appreciation to Adam Davey, Melinda Kane, Krista Paulsen, and Rick Phillips for their insightful comments, and to Miles Taylor for her excellent research assistance. REFERENCES
Albrecht, S. L., Coward, R. T., & Shapiro, A. (1997). Effects of potential changes in coresidence on matched older parent-adult child dyads. Journal of Aging Studies, 1, 81-96. Aquilino, W. (1999). Two views of one relationship: Comparing parents and young adult childrens reports of the quality of intergenerational relations. Journal of Marriage and the Family, 61, 858-870. Bengtson, V. L., & Kuypers, J. A. (1971). Generational difference and the developmental stake. Aging and Human Development, 2, 249-260. Bond, J. B., & Harvey, C. D. (1991). Ethnicity and intergenerational perceptions of family solidarity. International Journal of Aging and Human Development, 33, 33-44. Burt, R. A. (1982). Toward a structural theory of action: Network models of social structure, perception, and action. New York: Academic Press. Coward, R. T., & Dwyer, J. W. (1990). The association of gender, sibling network composition, and patterns of parent care by adult children. Research on Aging, 12, 158-181. Daly, K. (1994). Uncertain terms: The social construction of fatherhood. In M. L. Dietz, R. Prus, & W. Shaffir (Eds.), Doing everyday life: Ethnography as human lived experience (pp. 170-185). Toronto: Copp Clark Pittman. Eggebeen, D. J. (1992). Family structure and intergenerational exchanges. Research on Aging, 14, 427-447. Giarrusso, R., Stallings, M., & Bengtson, V. L. (1995). The intergenerational stake hypothesis revisited: Parent-child differences in perceptions of relationships 20 years later. In V. L. Bengtson, K. W. Schaie, & L. M. Burton (Eds.), Adult intergenerational relations: Effects of social change (pp. 227-263). New York: Springer. Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica, 47, 153-161. Hogan, D., Eggebeen, D., & Clogg, C. (1993). The structure of intergenerational exchanges in American families. American Journal of Sociology, 98, 1428-1458. Ingersoll-Dayton, B., & Antonucci, T. (1988). Reciprocal and nonreciprocal sides of intimate relationships. Journal of Gerontology, 43, S65-S73. Luescher, K., & Pillemer, K. (1998). Intergenerational ambivalence: A new approach to the study of parent-child relations in later life. Journal of Marriage and the Family, 60, 413-425. Lynott, P. P., & Roberts, R. E. L. (1997). The developmental stake hypothesis and changing perceptions of intergenerational relations, 1971-1985. The Gerontologist, 37, 394-405. Miller, R. B., & Wright, D. W. (1995). Detecting and correcting attrition bias in longitudinal family research. Journal of Marriage and the Family, 57, 921-929. Parrott, T. M., & Bengtson, V. L. (1999). The effects of earlier intergenerational affection, normative expectations, and family conflict on contemporary exchanges of help and support. Research on Aging, 21, 73-105.

146 / SHAPIRO

Risman, B. J., & Park, K. (1988). Just the two of us: Parent-child relations in singleparent homes. Journal of Marriage and the Family, 50,1049-1062. Roan, C. L., Hermalin, A., & Ofstedal, M. B. (1996). Intergenerational contact and support in Taiwan: A comparison of elderly parents and childrens reports (Research Report Series, No. EA96-36). Ann Arbor: University of Michigan, Population Studies Center. Rossi, A. S., & Rossi, P. H. (1990). Of human bonding: Parent-child relations across the life course. New York: Aldine de Gruyter. Seltzer, J. A., & Brandreth, Y. (1995). What fathers say about involvement with children after separation. In W. Marsiglio (Ed.), Fatherhood: Contemporary theory, research, and social policy (pp. 166-192). Thousand Oaks, CA: Sage. Soldo, B. J., Sharma, J. M., & Campbell, R. T. (1984). Determinants of the community living arrangements of older unmarried women. Journal of Gerontology, 39, 492-498. Suitor, J. J., Pillemer, K., Keeton, S., & Robison, J. (1995). Aged parents and aging children: Determinants of relationship quality. In R. Blieszner & V. H. Bedford (Eds.), Handbook of aging and the family (pp. 223-242). Westport, CT: Greenwood Press. Sweet, J. A., & Bumpass, L. L. (1996). The National Survey of Families and Households waves 1 and 2: Data description and documentation. University of Wisconsin-Madison, Center for Demography and Ecology. Web site: http://www.ssc.wisc.edu/nsfh/home.htm. Uhlenberg, P., & Cooney, T. (1990). Family size and mother-child relations in later life. The Gerontologist, 30, 618-625. Umberson, D. (1992). Relationships between adult children and their parents: Psychological consequences for both generations. Journal of Marriage and the Family, 54, 664-674. Wolf, D. A. (1984). Kin availability and the living arrangements of older women. Social Science Research, 13, 72-89.

Direct reprint requests to: Adam Shapiro Department of Sociology, Anthropology and Criminal Justice University of North Florida Jacksonville, FL 32224 e-mail: [email protected]