| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
a Eastern Virginia Medical School, Norfolk, VA
b Department of Child and Family Development, University of Georgia, Athens
Correspondence: Maximiliane E. Szinovacz, PhD, Glennan Center for Geriatrics and Gerontology, Eastern Virginia Medical School, Hofheimer Hall, Suite 201, 825 Fairfax Avenue, Norfolk, VA 23518. E-mail: Maxres{at}visi.net.
Laurence G. Branch, PhD
 |
Abstract |
|---|
 TOP Abstract Literature ReviewTheoretical Framework and...MethodsResultsDiscussionReferences |
|---|
Key Words: Gender • Intergenerational relations • Kin relationships
Despite considerable interest both in the consequences of retirement and in the predictors of parent–adult child contacts, few studies have linked these two research areas. Studies addressing retirement effects typically focus either on social contacts or social networks in general (Bosse, Aldwin, Levenson, Spiro, and Mroczek 1993
; Bosse, Aldwin, Levenson, Workman-Daniels, and Ekerdt 1990; Bosse and Ekerdt 1981; Francis 1990; George, Fillenbaum, and Palmore 1984; Van Tilburg 1992), whereas investigations into parent–adult child relationships center on familial factors such as geographical distance, emotional attachment, or feelings of obligation (Lawton, Silverstein, and Bengtson 1994; Rossi and Rossi 1990). Yet, life-course theories stress that work and family spheres are interdependent, and that events in one sphere can influence events or relationships in the other sphere (Bengtson and Allen 1993; Moen 1996). Investigation of changes in kin contacts after retirement also has relevance for retirement preparation and counseling. Both parents and adult children may either feel uncertainty or have unrealistic expectations about their relationship after the parents' retirement (Szinovacz, Ekerdt, and Vinick 1992). For example, parents may expect enhanced support from children that the children are unable or unwilling to provide (Remnet 1987), and adult children may expect increased time commitments on the part of retired parents (e.g., to take care of grandchildren) that interfere with parents' planned retirement activities. Such scenarios may generate conflict between generations and undermine parents' adaptation to the retirement transition. This study addresses retirement effects on parents' contacts with their adult children, using panel data from the National Survey of Families and Households (NSFH).
|
Literature Review |
|---|
|
TOPAbstractLiterature ReviewTheoretical Framework and...MethodsResultsDiscussionReferences |
|---|
report an increase in time spent with friends among both women and men, whereas Bosse and coworkers 1993 find a decrease in quantitative but stability in qualitative social supports among men who retired. A smaller study of retiring men (Van Tilburg 1992) indicates that the overall size of social networks remained relatively stable over the retirement transition, but at the same time there was considerable fluctuation of network members. Relationships with relatives tended to be more stable than those with friends or acquaintances. These inconsistent results are hardly surprising, as these studies rely on data from divergent samples and periods and use quite different measures of social contacts or supports.
Whereas research pertaining to the retirement transition has focused on social network changes, investigations of parent–adult child relationships are typically cross-sectional and mostly ignore parents' employment status. These studies demonstrate continued close bonds between parents and adult children into parents' later years (Eggebeen and Davey 1998; Lawton et al. 1994; Rossi and Rossi 1990). The extent of interactions seems to depend on gender and geographical distance (Lawton et al. 1994; Rossi and Rossi 1990) as well as on status transitions of both parents and children such as parents' divorce and remarriage (Kaufman and Uhlenberg 1998). The few studies considering parents' employment status yield inconsistent results. Rossi and Rossi 1990 show that mothers' full-time employment is associated with reduced face-to-face contacts with sons, whereas sons report more contacts (visits and phone) with full-time employed fathers. Logan and Spitze 1996 report no significant effects of employment status on visits or telephone contacts between parents and adult children.
In summary, current research provides some evidence that both parents' and adult children's employment status can influence intergenerational relations, and that retirement impinges on social networks. However, the extent and direction of these effects remain unclear, and there is essentially no information linking parents' retirement to contacts with their adult children.
|
Theoretical Framework and Hypotheses |
|---|
|
TOPAbstractLiterature ReviewTheoretical Framework and...MethodsResultsDiscussionReferences |
|---|
). It is not clear, however, what type of reorganization is best suited to compensate for the loss of work-related gratifications. Some argue that engagement in new activities is most beneficial, while others stress the importance of continuity in retirees' lives (Atchley 1976, Atchley 1999; Friedman and Havighurst 1954).
The broadest theoretical approach to intergenerational relations is the intergenerational solidarity framework developed by Roberts, Richards, and Bengtson 1991. These researchers distinguish six interdependent dimensions of intergenerational solidarity (structural, normative, functional, associational, affectional, and consensual). They further demonstrate that associational solidarity (contact among family members, the dimension of solidarity that constitutes the dependent variable in this study) is influenced by normative and affectional solidarity, that is, norms and feelings of closeness (Roberts and Bengtson 1990). Other research shows that structural solidarity (geographical distance) has a strong impact on contacts among family members (Logan and Spitze 1996; Rossi and Rossi 1990). An alternative approach by Cicirelli 1989 emphasizes the importance of attachment, obligation, and dependency (or needs) for kin relationships. The first two dimensions are akin to affectional and normative solidarity, whereas dependency introduces the notion of needs as predictor of kin interactions and supports (Davey and Eggebeen 1998). More recently, researchers (Belliston, Davey, and Bookout 1999; Kaufman and Uhlenberg 1998) have begun to explore ways in which changing needs and resources of each generation might affect changes and generational differences in intergenerational solidarity. This approach underlines the importance of considering intergenerational relationships from a dyadic level (i.e., one that considers parents' relationships with each child separately).
Based on this earlier work, we suggest that kin contacts are affected by opportunity structures (especially time constraints and geographical distance), feelings of obligation (norms), and the salience of specific kin relationships. Salience most likely derives from affectional solidarity, as well as from family members' needs for kin interactions. It is important to consider these dimensions for each family member involved in kin interactions. Thus, parent–adult child contacts will depend on both parents' and children's opportunity structures and on parents' and children's feelings of obligation and salience.
We further contend that reorganization of kin contacts after parents' retirement will reflect these influences. However, we lack adequate data to establish the impact of normative solidarity (the NSFH only contains data on attitudes toward coresidence and financial supports, which probably have little relevance for norms pertaining to associational solidarity). Consequently, our hypotheses focus on salience and constraints. Whether retired parents strive to enhance contacts with their children will depend foremost on whether such contacts can compensate for the role loss experienced upon retirement and thus the salience of contacts with the children. Specifically, contacts with children can compensate for the loss of social contacts with work colleagues, and they can provide the basis for identity enhancement in the roles of parent and grandparent. Because we lack a direct measure of kin salience, we rely on indicators that have been linked to salience of kin relations in past research. Earlier investigations suggest that salience of kin interactions is influenced by gender. Specifically, women seem to attach more importance to and are more involved in kin interactions than men (Logan and Spitze 1996; Rossi and Rossi 1990), although men also seem to derive some gratification from the grandparent role (Spitze and Ward 1998). In addition, the salience of contacts may vary by child's gender. Earlier studies indicate the importance of same-sex ties (Kaufman and Uhlenberg 1998). This implies that mothers may derive more gratification from interactions with daughters, whereas men may gain more satisfaction from contacts with sons. Salience of child contacts will also depend on whether retired parents have alternative roles that may satisfy their social and identity needs, most importantly the marital role. Married men in particular seem to obtain fulfillment of social needs outside the work role foremost from their spouses (Antonucci and Akiyama 1995). In addition, emotional closeness to specific children will likely determine salience of contacts with these children after retirement. We thus hypothesize that retired parents will seek increased contacts with adult children if they feel close to the child, if the parents are not married (this applies particularly to male retirees), and if the child has children. We further hypothesize that women will be more inclined to enhance contacts with daughters, whereas men will be more likely to increase contacts with sons.
Previous research further suggests that adaptation to retirement may proceed in stages. Retirees initially engage in nonroutine projects and activities such as travel or geographical moves that tend to be exciting and satisfying but of relatively short duration (the so-called "honeymoon" period; see Atchley 1976; Ekerdt, Bosse, and Levkoff 1985). This period is followed by the establishment of a more enduring retirement lifestyle. We suggest that enhanced contacts with adult children are more likely to be a pattern established in this latter period. Thus, we hypothesize that enhancement in contacts will be more likely among parents who have been retired for two or more years.
Although salience of kin relationships is likely to influence whether retirees direct their search for compensating postretirement activities toward interactions with kin, the implementation of such aspirations can be limited by deficient opportunity, particularly geographical distance and time constraints. As noted earlier, geographical distance has been consistently linked to frequency of association with kin (Logan and Spitze 1996; Rossi and Rossi 1990). Support for the impact of other constraints is limited and inconsistent. There is some indication that children's employment may curtail relations with their parents (Kaufman and Uhlenberg 1998; Rossi and Rossi 1990). Another constraint may derive from the number of children. Although having numerous children seems to enhance parents' ability to see any child (Uhlenberg and Cooney 1990), it is not clear how number of children influences contacts with each individual child. Recent research on grandparenthood suggests, for example, that individual grandparent–grandchild contacts decline with the number of grandchildren (Uhlenberg and Hammill 1998). Similarly, parents with more children are obliged to divide attention among children, which may reduce interactions with each individual child. We thus hypothesize that post–retirement increases in parent–adult child contacts should be more pronounced for parent–child dyads who live close to each other, and that they should be less pronounced for full-time employed children and for parents with numerous children.
|
Methods |
|---|
|
TOPAbstractLiterature ReviewTheoretical Framework and...MethodsResultsDiscussionReferences |
|---|
). The wave 2 interviews were completed by 10,005 respondents. We use weights to adjust for nonresponses at time 1 and attrition at time 2.
Because our analyses focus on changes in contacts brought about by the retirement transition, we selected a subsample at risk of retirement. This subsample consists of respondents aged 55–75 at time 2 who were employed 10 hours or more at time 1 and had adult children outside the household at both times. The age range corresponds with typical retirement ages, and the restriction to 10 hours of employment ensures that parents pursued some regular employment preceding retirement. However, only very few respondents (2.2%) worked fewer than 10 hours at baseline. In addition, we excluded from the sample those parents who reported receipt of health care from any adult child (5.3%). This restriction is necessary to avoid confounding of associational (social contacts) and functional (supports) solidarity and to ensure that potential increases in contacts are not due to changes in parents' health and need for care. As NSFH respondents reported on contacts with each adult child, the data were accumulated for each parent–child dyad. Thus, the parent–child dyads and not the parents are the primary unit of analysis (for a similar approach see Uhlenberg and Hammill 1998). The final unweighted sample size consisted of 2,153 parent–adult child dyads (based on responses from 792 parents). Not considered in the analyses were adult children for whom reports were only available at one wave (children who "disappeared" between waves or were first mentioned at time 2), as well as children who moved into or out of their parents' household. This is necessary because analysis of change in contacts requires comparable information at both waves. The data refer exclusively to adult children who were age 19 or older at baseline.
Nonresponses in panel data can lead to biased results when cases with missing data are dropped from the analyses (Little and Rubin 1987), especially when data are not missing completely at random (MCAR) as is the case for our analyses. We therefore created five complete-case data sets using the regression method provided by SPSS (SPSS 1997). The data shown in the tables are based on multiple imputation inference using NORM (Schafer 1997) from analyses performed on these five data sets. To assess the sensitivity of our findings to the imputation methods used here, we compared our results with those obtained using the expectation–maximization method, as well as reliance on nonimputed data (listwise deletion), yielding highly similar findings in each case. This suggests that results are fairly stable and only minimally affected by the imputation method.
Respondents are close to or at retirement age (M = 62.8), and 56% are women. Eighteen percent of the sample are Black and 6% are from other minorities. Sixty-one percent of respondents were married at time 2. Education is slightly above the high school level (M = 12.3 years). Most respondents (65%) worked full time at baseline, and 45% retired between waves. The adult children are equally divided by gender; 64% of the children are themselves parents, and 76% were employed full time at time 2. More than one third of the children (36%) lived within 10 miles of their parents at time 2. Children's ages varied considerably (range = 23 to 55; M = 37).
Dependent Variables
Two dependent variables were used in the analyses: visits and telephone/letter contacts at time 2. Note that in all cases the analyses relied on the parents' reports. Frequency of visits and letter/telephone contacts was measured with the question: During the last 12 months, how often did you see (talk on the telephone or receive a letter) from [child]? Answer categories were: 1 = not at all, 2 = once a year, 3 = several times a year, 4 = 1–3 times a month, 5 = once a week, 6 = more than once a week. Wave 1 visits and telephone/letter contacts were included as predictor variables in all analyses. Thus, the dependent variables reflect contacts at time 2 net of contacts at time 1. At time 1, respondents reported seeing their children on average several times a month (M = 3.99) and to have somewhat more frequent telephone/letter contacts (M = 4.54).
Independent Variables
The major independent variable is parents' retirement. We rely on full withdrawal from the labor force to measure retirement. This definition may differ from self-definition as retiree (Ekerdt and DeViney 1990; Szinovacz and DeViney 1999). Because employment-related time constraints limit parents' ability to interact with children or to travel to visit children, it is most likely that parents' withdrawal from the labor force (rather than retirement from the main career job) precipitates changes in contacts with children. Length of retirement was operationalized by subdividing retirees into those retired less than 2 years and those retired 2 years or more. This dichotomy corresponds with our hypothesis concerning a potential honeymoon effect. Preliminary analyses further indicate that there are no differences between those retired 2–3 years and those retired longer. Date of retirement was only assessed in the questionnaires given to respondents and has a relatively high number of missing values (7%). Because imputation of missing values for a subpopulation (retirees) is problematic, we developed an extra category for missing cases and included this category in the analyses. Thus, the reference group consists exclusively of persons who did not retire between waves.
Additional independent variables are parents' and children's gender, parents' marital status, parents' perceived emotional closeness to each child (measured at baseline), the number of children each parent has outside the household, children's parental status, children's employment, and geographical distance. Gender of both parents and children was dummy-coded (female = 1, male = 0).
Parents' marital status was coded into two dummy variables, those married and those widowed at time 2. Divorcees and the very few never-married parents are the reference. This distinction reflects earlier research findings indicating that it is divorced parents, especially, who may lose contacts with their children (Kaufman and Uhlenberg 1998). In addition, change in marital status between waves was controlled in the analyses.
Parents' emotional closeness to each child was only measured at baseline. Parents were asked to describe their relationship with each child (1 = very poor, 7 = excellent).
Number of parents' children refers exclusively to adult children outside the household, as these are the children of interest in this study. Because of the distribution of this variable (fewer than 5% have only one child outside the household), we differentiate among parents who have only one or two children, those with three children, and those with four or more children. Although children in the household will also compete for parents' attention, contacts with these children are not subject to change. However, change in household composition was controlled in the analyses.
Children's parental status was treated as a dichotomous variable (1 = child has biological children in the household; reference = child has no biological children living in the household).
Measurement of children's employment status is based on the question: Is (child) currently working full time, part time, or not at all? This information was only provided at time 2. We differentiate between children employed full time and those employed part time or not at all. Preliminary analyses indicated that there is no difference in contacts between part-time and not employed children.
Measurement of geographic distance is based on parents' reports of how far away (in miles) each child lived at time 1 and time 2. In line with earlier research (Rossi and Rossi 1990), we use 10 miles as the criterion for distinguishing children who live within parents' immediate neighborhood from those living outside the parents' immediate neighborhood. In addition, change in distance between waves was controlled in all analyses.
Other controls included in the analyses are race, whether parents worked full time at baseline, parents' limitations in activities of daily living at time 2, change in parents' self-reported health, change in parents' income, change in spouse's employment status between waves, whether the child is the biological child of the respondent's spouse, changes in parents' and children's marital status between waves, and children's age. Means and standard deviations of all variables (including controls) are shown in Table 1 .
|
) with clusters for family unit (parent's identification number), using a random-effects model and additional adjustments for robust standard errors with the Huber/White/sandwich estimator of variance (Stata 1997). Coefficients obtained from GEE are akin to unstandardized linear regression coefficients, although Stata does not provide standardized estimates or R-square statistics.
Interactions were tested in line with standard regression procedures, that is, equations with higher order interactions were tested first and include all lower order interaction terms (Aiken and West 1991). We started with the highest order terms allowed by subgroup distribution, which are triple interactions. We first tested interactions by parents' gender, geographical distance to each child, and retirement. As this interaction proved to be significant, we performed all subsequent analyses separately for children living within 10 miles of the parent and children living further away. We then tested for interactions by parent's gender, parent's retirement, and each of the other hypothesized contingencies (child's gender, parent's marital status, emotional closeness to the child, number of children, child's parental status, child's employment) in separate equations (e.g., Parent's gender x Parent's retirement x Child's gender). If the higher order term was not significant, the next lower order term was tested (e.g., Child's gender x Retirement, Parent's gender x Retirement). This procedure provides some assurance that multiple interaction terms do not evolve through multicollinearity among the interaction terms. In case of significant interactions, we further estimated outcomes (frequency of contacts) for the pertinent subgroups from the regression equations (Aiken and West 1991). These results are not shown in the tables but are described in the Discussion section.
Tests for multicollinearity were performed. Variance inflation (Fox 1991) for the full model with controls (but without the interaction terms) yielded a mean VIF of 1.30. This indicates that multicollinearity is not a problem in our analyses.
|
Results |
|---|
|
TOPAbstractLiterature ReviewTheoretical Framework and...MethodsResultsDiscussionReferences |
|---|
Having established that visits are affected by retirement, we explored predicted contingencies of this effect. We first examined whether retirement effects differ for contacts with children living nearby (within 10 miles) and those living further away, and whether such effects vary by parent's gender. These analyses (see Table 2 , first data column) substantiate that the Retirement x Distance x Parent's gender interaction is significant. Inspection of the interaction indicates that among parent–child dyads living within 10 miles of each other, mothers who retired see their children less frequently than mothers who remained employed, whereas retired fathers report more frequent visits than continuously employed fathers. Among parent–child dyads who do not live close to each other, visits occur less frequently and are similar for employed and retired men and women.
|
Salience of post-retirement contacts was also predicted to vary by children's parental status. This effect is confirmed for parent–child dyads living more than 10 miles apart and varies by parent's gender (see Table 2 , second data column). Mothers seem to maintain a relatively high level of contacts with children who are themselves parents, regardless of the mother's employment status. In contrast, mothers' visits with childless children seem to decline after retirement. Fathers increase contacts with childless children after they retire but decrease visits with children who are parents. Among parent–child dyads living within 10 miles of each other, child's parenthood has no effect on post-retirement changes in visits.
Salience was also expected to be reflected in parents' feelings of closeness to the child and in parents' needs (based on their marital status). We find no support for either hypothesis. Retirement-related changes in visits are not contingent on parents' feelings of closeness to the child nor on their marital status.
Our final hypothesis pertaining to salience of post-retirement contacts referred to length of retirement. We expected a stronger influence for parents who were past the honeymoon period, that is, those retired 2 years or more. The data confirm that length of retirement matters (Table 2 , model 3). It is primarily mothers retired 2 years or more who decrease visits with children living nearby. Retired fathers, on the other hand, report more visits with nearby children regardless of length of retirement, although this trend is particularly pronounced for fathers retired under 2 years. Length of retirement has no impact on visits with children living more than 10 miles away.
In addition to geographical distance, we expected children's employment and number of children to constrain parents' aspirations for post-retirement contacts. We find no support for an influence of children's employment status. Number of children outside the household does moderate retirement effects for parent–child dyads living within 10 miles of each other (Table 2 , model 2). Mothers' retirement seems to lead to a decline in visits primarily if they have four or more children outside the household. In contrast, fathers are particularly prone to increase contacts after retirement if they have four or more children, whereas fathers of one or two children outside the household tend to have fewer post-retirement visits with these children. Among parent-child dyads living more than 10 miles apart, number of children does not moderate retirement effects.
|
Discussion |
|---|
|
TOPAbstractLiterature ReviewTheoretical Framework and...MethodsResultsDiscussionReferences |
|---|
Our results suggest that retirement has some effect on visits between parents and their adult children, whereas telephone contacts seem to be unaffected by parents' retirement. That we find no retirement effects on telephone contacts is hardly surprising. Telephone contacts are less likely to be constrained by parents' employment, and they also are unlikely to compensate for loss of the work role. This finding does suggest, however, that visits and telephone contacts have a different meaning and consequently should be treated separately. Some earlier studies combine these two types of associational solidarity (e.g., Lawton et al. 1994).
Retirement effects on visits are quite complex and are contingent on a variety of circumstances. Given the constraints associated with long-distance relationships, we expected retirement to have only a modest impact on visits with children living more than 10 miles from their parents. This hypothesis was partially supported: Retirement effects occur foremost among parent–child dyads living close to each other. However, consideration of the child's parental status somewhat modifies this conclusion. Among mothers, the grandparent role seems to override potential employment-related constraints, leading to fairly constant contacts over the retirement transition with adult children who are themselves parents. On the other hand, mothers' contacts with distant childless children decline after retirement, suggesting that such contacts have little appeal to mothers as a post-retirement endeavor or that childless children are particularly uninterested in contacts with retired mothers. The opposite holds for fathers' contacts with distant children, that is, after retirement they visit slightly more with childless children and slightly less with children who are parents. As these differences are fairly small, they must be interpreted with caution. Perhaps retired fathers obtain more gratification from adult-to-adult contacts with children than from greater involvement in a distant grandparent role.
Changes in post-retirement visits with children living nearby vary by child's gender, as well as by parents' gender, length of retirement, and number of children. To the extent that retired parents increase contacts with nearby children at all, such increases seem to occur primarily with daughters. This finding contradicts our assumption that retired parents may seek out activities with same-sex children. It might indicate that daughters are more inclined than sons to accommodate retired parents' needs for enhanced contacts. It is also conceivable, however, that contacts with daughters (who may be more socialized toward maintaining close relations with parents and more attuned to parents' needs) are more gratifying to retirees.
Parents' gender and length of retirement jointly influence contacts with children living within 10 miles of the parent. Contrary to our hypothesis, retired mothers tend to spend less and retired fathers more time with children than their employed counterparts. For mothers, this trend seems to become more pronounced with length of retirement and number of children, whereas it is fairly stable for fathers. We attribute this difference to the gendered division of labor throughout men's and women's working years. Employed mothers typically carry the main burden of family work, leaving them little time to engage in leisure activities (Ferree 1991; Lopata 1995). In contrast, men are less involved in kin work (Antonucci and Akiyama 1995; Rossi and Rossi 1990), and especially employed men devote little time to other family work such as household activities or child care (Ferree 1991). This means that men and women enter retirement under quite different circumstances. Retired men may very well feel that they neglected contacts with their children while working and use their extra time to "catch up" when opportunity allows (i.e., when the children live nearby). Retired women, on the other hand, will usually have maintained close ties with their children throughout their working years and may feel less of a need to reestablish a close relationship. In retirement they may see their first opportunity to engage in personal leisure activities. As the trend toward reduced contacts is most prominent among women with several children, it may reflect special needs for personal leisure among women who had been heavily burdened by multiple responsibilities. Furthermore, married women often take charge of the couple's retirement activities (Keating and Cole 1980), and this responsibility may detract from their increasing contacts with adult children. It is also possible that children are particularly amenable to enhance contacts with retired fathers, with whom they probably had fewer contacts in the past. Contacts with retired mothers, on the other hand, may lack novelty, and mothers' attempts to enhance contacts may be perceived as an imposition by their children (Remnet 1987).
We further hypothesized that time constraints may inhibit enhanced contacts with children after retirement. We find no support for the assumption that children's employment would moderate the impact of retirement. The second predicted constraint, number of children, does moderate retirement effects on contacts but not in the expected direction. Retired fathers are particularly likely to enhance contacts with nearby children if they have four or more children, whereas retired mothers' tendency to decrease contacts with nearby children prevails among mothers with four or more children outside the household. For mothers of multiple children, gratifications derived from the parent role may be saturated, and they may consequently seek other endeavors in retirement. The finding for fathers could reflect multiple opportunities, that is, fathers with several adult children have more opportunity for contacts and apparently take advantage of this opportunity. It is also conceivable that fathers of multiple children are more kin-oriented than men with fewer children and thus more inclined to gear retirement activities toward contacts with their children.
Taken together, our findings indicate that retired parents' contacts with their adult children depend on selected parent and child characteristics. This suggests that both parents' and children's expectations and wishes, as well as selected constraints (especially geographical distance), impinge on parents' post-retirement contacts with their children. As we lack data on either parents' or children's desired contact frequency, we cannot establish whether the observed patterns reflect either party's wishes or what type of compromises parents and children reach. For example, retired mothers may decrease contacts with children to pursue other endeavors, or their children may reduce contacts to avoid potential imposition by retired mothers. If retirees' expectations are not met, they may be forced to amend their plans for post-retirement activities. Some parents may also exert pressure on their children to meet their expectations. Either scenario is likely to kindle conflicts between retired parents and their children.
It is also important to note that, although all discussed interactions effects are significant and contribute significantly to model fit, the observed differences in contact frequency are quite small, suggesting considerable continuity in parent–child contacts over the retirement transition. This would suggest that post-retirement contacts with children have only limited importance as a source for retirees' social contact and identity enhancement needs. This finding is in line with earlier research suggesting that retirees derive more satisfaction from contacts with friends than from contacts with relatives (Dorfman, Heckert, Hill, and Kohout 1988). It is only under special conditions that parents seek fulfillment of such needs through contacts with their children. Furthermore, our analyses, based on all parent–adult child dyads, essentially provide a weighted (by family size) average of contacts for parents who have more than one adult child. Thus, it is quite possible that retired parents simultaneously increase contacts with one child and decrease contacts with another. Such selective changes in contacts with specific children are reflected in relationships pertaining to child characteristics but not in relationships pertaining to parent characteristics.
Our findings may also be subject to methodological and data limitations. We lack indicators for some key variables (e.g., a direct measure of kin salience) and have rather poor measures for other indicators (e.g., affectional solidarity). Indeed, our failure to find a moderating effect of perceived closeness on post-retirement contacts most likely can be attributed to measurement quality and the very skewed distribution of the closeness variable (most parents reported feeling extremely close to their children). In view of these caveats, further research is clearly needed. Most important would be comparisons of expectations and aspirations for post-retirement contacts on the part of parents and children, and assessment of the implementation of such expectations after retirement. Future studies should also explore retirement effects on other dimensions of intergenerational solidarity and further subgroup variations in retirement effects (e.g., by race).
Despite their limitations, our findings lend themselves to some recommendations for retirement planners and family counselors. They suggest that gender-stereotyped expectations about men's and women's post-retirement activities (women will spend time with kids, men will go to the golf course) may not fit reality. Similarly, we find little support for the bumper-sticker notion that retired parents will leave their children behind and devote their retirement leisure exclusively to nonfamily pursuits. Retired parents do keep in close touch with their adult children. Thus, retirement planners and family counselors need not be concerned that retirees will "abandon" their children. Rather, concern should be directed toward the implementation of realistic plans for post-retirement kin relationships and toward both generations' ability to compromise on a level of contact that meets parents' and children's expectations and needs.
|
Acknowledgments |
|---|
Received for publication April 18, 2000. Accepted for publication September 29, 2000.
|
References |
|---|
|
TOPAbstractLiterature ReviewTheoretical Framework and...MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. Jill Suitor, J. Sechrist, and K. Pillemer Within-Family Differences in Mothers' Support to Adult Children in Black and White Families Research on Aging, September 1, 2007; 29(5): 410 - 435. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
