Attitude Similarity in Three-Generation Families: Socialization, Status Inheritance, or Reciprocal Influence? Author(s): Jennifer Glass, Vern L. Bengtson and Charlotte Chorn Dunham Source: American Sociological Review, Vol. 51, No. 5 (Oct., 1986), pp. 685-698 Published by: American Sociological Association Stable URL: http://www.jstor.org/stable/2095493 . Accessed: 23/08/2013 17:46

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

.

American Sociological Association is collaborating with JSTOR to digitize, preserve and extend access to American Sociological Review.

http://www.jstor.org

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions

ATTITUDE SIMILARITY IN THREE-GENERATION FAMILIES: SOCIALIZATION, STATUS INHERITANCE,

OR RECIPROCAL INFLUENCE?*

JENNIFER GLASS VERN L. BENGTSON University of Notre Dame CHARLOTTE CHORN DUNHAM

University of Southern California

This study examines hypotheses of attitude transmission across three ideological domains (gender roles, politics, religion) to access the adequacy of direct socialization, status inheritance, and reciprocal influence models in a developmental aging perspective. Data are from 2,044 individuals, members of three generation families, grouped to form parent-youth (G2-G3) and grandparent-parent (GJ-G2) dyads. Results suggest, first, that there is little convergence of parent-child attitudes with age when viewed cross-sectionally. Second, status inheritance processes do account for a substantial amount of observed parent-child similarity, but parental attitudes continue to significantly predict childrens’ orientations after childhood. Third, child influences on parental attitudes are relatively strong and stable across age groups, while parental influence decreases with age, although the exact pattern of influence varies by attitude domain.

For many years, social theorists have considered the role of the family in maintaining continuity in social ideologies over time (Engels [1884], 1967; Adorno et al., 1950; Parsons and Bales, 1955; Thomas and Znaniecki, 1958; Chodorow, 1978). The resulting view of the family as conservative (for example, slowing the pace of social change) and monolithic (influencing individual beliefs in a forceful and consistent manner) perhaps reached its ultimate expression in the attempts of some revolutionary movements (such as in Cambodia or China) to break up generational ties in order to foster rapid social change. And, in fact, contempo- rary research on the intergenerational transmission of attitudes has shown that parents’ attitudes, especially mothers’ attitudes, are significant posi- tive predictors of children’s attitudes in adulthood (Acock and Bengtson, 1978; Bengtson, 1975; Dalton, 1980; Jennings and Niemi, 1982; Smith, 1983).

However, this typification of the family as conservative and monolithic in its influence on ideological orientations has come under increasing scrutiny among family scholars, as they point out the diversity of influences on children and the complexity of family relationships. In this study, we focus on two basic empirical questions: 1) How

much actual similarity in social ideologies is found between American parents and children at different points across the life cycle? 2) What are the forces generating that similarity over the life course?

To answer these questions, we first describe the traditional view of attitude transmission derived from childhood socialization theory. Then we explore conceptual criticisms of the socialization paradigm from alternative theoretical perspectives. Finally, we empirically examine the dynamics of attitude transmission using responses to specific attitudinal scales from a sample of three-generation families.

SOCIALIZATION THEORY AND DEVELOPMENTAL AGING

Traditional conceptions of socialization have viewed the family, speficially parents, as the principal agent of socialization in childhood (Freud, 1933; Erickson, 1950; Heilbrun, 1965). One of the functions of the family is seen as the provision of stability and continuity to individual members. Families are thought to provide systematic social- ization through which children are taught the norms of the social order. Attitude similarity between generations, from this view, is the consequence of successful parental socialization of beliefs and values. Children learn their parents’ values, beliefs, and attitudes through both direct teaching and indirect observation, as part of the information and guidance that children either actively seek out (in the Piagetian sense) or passively accept (through social conditioning) in maneuvering their way through life.

While childhood socialization theories do not directly address the issue of parent-child similarity

*Address all correspondence to Jennifer Glass, Depart- ment of Sociology, University of Notre Dame, Notre Dame, IN 46565.

This study, based on data from the USC Longitudinal Study of Three-Generation Families, was supported by grants from the National Institute of Aging (#AG-04092) and the National Institute of Mental Health (#MH-38244). Special thanks go to Rich Williams, Richard Miller and Donna Polisar for their assistance in developing this paper, as well as to three anonymous reviewers for their helpful comments and criticisms.

American Sociological Review, 1986, Vol. 51 (October:685-698) 685

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions

686 AMERICAN SOCIOLOGICAL REVIEW

in adulthood,’ the implicit assumption of tradi- tional conceptualizations has been that childhood socialization is so intense, prolonged, and psychodynamically important that the attitudes and values formed in the family context persist well into adulthood (Chodorow, 1978; Campbell, 1969; Adorno et al., 1950). Thus, it might be expected that parents and children would continue to exhibit attitude similarity across the life course, and into later adulthood-though perhaps diminishing with time as the intensity of parent-child contact diminishes.

This traditional approach to socialization has been challenged by scholars in recent years for failing to address two important issues. First, from a macro-structural point of view, parent-child attitude similarity may be viewed more as the result of social forces that generate the inheritance of social status than as the product of individual psycho-social influence. One of the central issues in the interpretation of findings of parent-child attitude similarity is whether such similarity can be attributed to successful parental socialization, per se, or whether it has more to do with successful intergeneration transmission of class, race, reli- gious affiliation, marital status, and other promi- nent social statuses that structure life experience and mold social attitudes (Acock, 1984). What parents transmit may be social statuses, more than attitudes and values. In this way, similarities in social structural position may create attitudinal similarities between parents and adult children through a common-cause association (i.e., parents and children have undergone similar attitude- shaping experiences).

The second conceptual challenge to an uncriti- cally traditional perspective of socialization is the possibility that similarity in attitudes between parents and children could equally be due to the influence of children’s attitudes on those of their parents, especially as children age. The traditional perspective on socialization focuses on young children and adults and ignores the possibility of variability across the life course by the age and developmental stages of the parents and children at each point in life (Hagestad, 1981; Featherman, 1983). Proponents of an interactionist perspective (Bell and Harper, 1977; Lerner and Spanier, 1978; Bengtson and Troll, 1978; Hagestad, 1984) argue that children increasingly influence their parents with age, and that attempts to model intergenerational influence as a one-way process-flowing from parents to children-may be fundamentally errone- ous, since reciprocal effects occur.

Only a few studies, however, have emprically tested the reverse influence process with respect to

social attitudes. Hagestad (1984) noted that about two-thirds of parents, and one-third of children, reported “successful” influence by children in her three-generation sample. Angress (1975) found that mothers of radical college students changed their attitudes about cohabitation based on their children’s behavior. Chaffee et al. (1971) reported that adolescents influenced their parents’ television behavior. Curiously, studies of reciprocal influ- ence began in the literature on infant development (Lewis and Rosenblum, 1974), despite the fact that older children are presumably much more capable of altering parents’ stated beliefs or behavior.

Very little is known about intergenerational attitude similarity across the lifespan or the forces generating similarity past childhood (Bengtson, et al., 1985). Some theorists emphasize the impor- tance of parent-child bonds across all stages of the life cycle (Shanas, 1979; Troll et al., 1979), implying that substantial intergenerational similar- ity exists across the life course. Others emphasize flexibility and change in parent-child relations at different stages of the life course.

Theories of developmental aging (Bengtson and Kuypers, 1971; Hess and Waring, 1978; Moss and Abramowitz, 1982; Baltes, 1979; Rossi, 1980) suggest that parents and children have different investments in family relationships and different sources of power in family interaction as they move through the life course. For example, children in late adolescence may share few of the adult statuses that their parents hold and may be facing the developmental tasks of independence and differentiation from parents2 (Erickson, 1950). Such processes would suggest relatively larger absolute discrepancies between parents’ and adult children’s attitudes. This position implies that social status similarity3 should account for rela-

1 This closely parallels Manheim’s (1952) notion of youths’ “fresh contact” with the social order (see discussion in Bengtson et al., 1985).

2 We refer throughout this paper to “social statuses” occupied by parents and children. Our use of this term is broad and inclusive, intending to cover family and community statuses as well as general socio-economic status.

3 Although the developmental aging paradigm has proved to be a useful tool in the investigation of family relationships over time, it needs to be amended with a theory of social change. Elder (1974) has demonstrated the importance of looking at the impact of social and historical changes on family functioning. Rapid social changes (revolutions, economic recessions, technological advances) may encourage adult children to increasingly look to non-parental sources of information and support. Improvements in health status and economic well-being may decrease the dependence of elderly parents on their children in the future. These historical changes may limit the impact of internal family dynamics in attitude formation. Unfortunately for analytic purposes, the data on three-generation families utilized in this research report are cross-sectional rather than longitudinal. As such, we have confounded historical (or cohort) effects and developmental/life stage effects. Therefore, our

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions

ATTITUDE SIMILARITY IN FAMILIES 687

tively less of the relationship between parents’ and childrens’ attitudes at this point in the life cycle, since young adults have not yet attained many of the social statuses that inform their parents’ beliefs (few have married or become parents, many have not finished their education and have had minimal opportunities for occupational achievement, etc.). With respect to the reciprocal influence between young adults and their parents, there is reason to believe that parental influence may still be quite strong at this stage, relative to child influence. From a social exchange perspective, the ability of parents and children to influence each other should be determined by the relative resources and rewards that each bring to the interaction. At this stage, parents have only recently relinquished their authority over their children, and young adult children may still rely on their parents for material support and guidance. Moreover, young adults have little of the experience or resources that would enable them to influence their parents’ attitudes.

Turning now to middle-aged children and their elderly parents, a developmental aging perspective suggests that a different set of dynamics may characterize their relationship. As youth age, they are more likely to attain social structural position similar to their parents with respect to marital status, income property ownership, etc. By mid-life, children have achieved a variety of adult social statuses. The life experiences generated by these adult roles are likely to replace direct parental influence in the modification of social attitudes. However, this similarity of adult social roles between the generations may lead to smaller mean differences in attitudes between them. In other words, youth may gradually come to hold views more similar to their parents’ as they have children of their own, buy property, and obtain full-time employment, although their parents’ influence does not directly cause them to alter their beliefs. The respective developmental stages of middle- aged children and elderly parents suggest further that the pattern of influence between parents and children may change over time. Middle-aged children are in many ways at the height of their social power in industrialized western societies (Riley, et al., 1982). Aged parents in later life, on the other hand, may become more dependent on their middle-aged children for advice and informa- tion than before, reflecting both physical decline and a loss in social power. In this context, middle-aged parents may not view their elderly parents as appropriate social referents. These

factors suggest that adult children may increasingly influence elderly parents over time, while elderly parents’ influence on their adult childrens’ attitudes may have declined since mid-life.4

In summary, it can be said that traditional views on socialization have focused upon the process of influence from parent to child without adequately considering the impact of inherited social status, the possibility of mutual influence, and variability across the life course due to developmental aging. More recent literature suggests that attitude transmission may indeed be mutual and grounded in social and historical milieu (Elder, 1984). The degree of similarity and difference between parents and children will be affected by the dynamics of mutual influence and developmental change. Re- ciprocal influence will also be played out within a broader structural and historical context.

Keeping in mind the importance of both developmental change and mutual influence, we have selected three attitudinal domains for study- religious ideology, political ideology, and gender ideology. Three ideological areas, rather than one, were selected both to test the generalizability of the developmental aging perspective outlined here to a variety of social attitudes, and to avoid heavy reliance on one particular content area in address- ing broad conceptual issues in attitude transmis- sion. Although variability of results across atti- tudes scales is discussed, it is not the primary focus of this paper. Empirical research on religious, political, and gender ideology has tended to show both moderate intergenerational transmission and cohort effects of varying sizes (Bengtson and Troll, 1978). Prior research on religious ideology has demonstrated moderate to high parent-child agreement (Acock, 1984; Weiting, 1975; Hoge et al., 1982) and small but significant age cohort difference in religious ideology (Hyman, 1959). Although the literature on political socialization is too extensive to review here, studies generally indicate that transmission is moderate for political ideology (Jennings and Niemi, 1968, 1982). Finally, recent studies of gender role attitudes (Mason et al., 1976; Thornton et al., 1983) show that while attitudes across age cohorts have liberlized in the past 20 years, younger age cohorts have changed their opinions to a much greater extent than older age cohorts. Smith and Self (1980) report only minor transmission of gender

ability to directly assess the impact of rapid social change on attitude similarity over the life course is limited. However, follow-up data on these families 12 years later are currently being collected and will be used to confirm or amend the findings discussed here.

4 Few prominent theorists have made firm statements about the strength or endurance of dispositions socialized in childhood (Goslin, 1969). Most socialization theorists view socialization as a process begun in infancy and ending in death. However, the term “resocialization” is often used to refer to specific attempts to alter the content of earlier socialization. The presumption seems to be that socialized outcomes are relatively stable unless and until specific attempts at resocialization are made in adult- hood.

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions

688 AMERICAN SOCIOLOGICAL REVIEW

ideology from mothers to daughters in a college sample. Overall, these studies suggest significant variability in the impact of recent social changes on attitudes and attitude transmission in the three areas (social change most pronounced in the area of gender ideology, followed by political ideology and religious ideology).

In the analyses to follow, we focus on three major hypotheses. The first hypotheses is that elderly parents and their middle-aged children show smaller mean differences across all three domains of attitudes than middle-aged parents and their young adult children. In other words, attitudes of parents and children converge with age. Both developmental aging and status inherit- ance explanations suggest smaller attitude differ- ences in older generation dyads, although tradi- tional socialization theory suggests larger attitude differences with age. Developmental theories emphasize the rebelliousness of youth as they attempt to separate and establish independence from their parents. As children move into adult roles and establish their independence, their need to differentiate themselves from their parents decreases. Traditional socialization theories, how- ever, emphasize parent-child contact and parental control of resources as the forces generating parent-child attitude similarity. To disentangle these effects requires further regression analysis.

Our second hypotheses, therefore, is that parents’ attitudes continue to significantly predict children’s attitudes, after controlling for children’s current social status. However, parental influence (controlling for social status) should be weaker for elderly parents than for middle-aged parents, reflecting the diminishing intensity of parent-child interaction. Conversely, the developmental trajec- tory of status inheritance suggests that status effects should increase with age, as children take on adult roles similar to those their parents held in adulthood.

Finally, the possibility of reciprocal influence must be addressed, to insure that the causal direction of parental effects are correctly specified, The developmental aging perspective suggests that as children become older, they are better able to influence their parents. Therefore, our third hypothesis is that child influences on parental attitudes increase with age, while parental influ- ences on children’s attitudes decline with age.

Three basic analyses are performed: 1) compar- ison of absolute differences between generation pairs on three attitudinal scales (political ideology, gender ideology, religious ideology); 2) regression analysis of parental attitude score on adult child’s attitude score, with and without demographic indicators of child’s social status; and 3) structural equation modeling of both parent and child attitude scores across successive generations. The first analysis addresses the question of whether parent- child similarity declines in successive generations

within families. The second analysis attempts to disentangle the effects of adult children’s social structural position (which may be quite similar to their parents’) from the effects of parental attitudes per se on adult children’s attitudes. The third analysis addresses the question of reciprocal effects in attitude transmission, controlling for the effects of social structural variables on individual re- sponses. This analysis compares the magnitude and significance of parent-child versus child-parent influence, and compares patterns of influence within dyads across generational positions in the family.

METHODS

Sample Selection

The data for this analysis are based on responses for 2,044 individuals drawn from a broader study of three generations conducted in Southern Califor- nia in 1973. The sample was drawn from a population of 840,000 members of a Los Angeles area health care plan (described in Bengtson, 1975). To be eligible for inclusion in the original sampling frame, members had to have been males over 65 with at least one grandchild between 16-26 years of age. Sample construction proceeded by sending questionnaires to all eligible grandchildren between 16-26, their parents, and related grandpar- ents in the original sampling frame. 77 percent of the grandparents (Gi) had only one biological child respond; another 20 percent had two children respond. Among the parents (G2), 48 percent had only one child respond, 35 percent had two, and 13 percent had three or more. Although not a random sample, this sample does represent a wide group of individuals from various ethnic, economic and social backgrounds. The sample is generally representative of white, economically stable, middle- and working-class families. This sample has an advantage over some other studies of three-generation families which have drawn their sample from among college students and their parents. The mean age of the grandparent genera- tion (GI) is 67.1. Mean age for the parent generation (G2) is 43.8, while mean age for the adult child generation (G3) is 19.7.

Attitudes and opinions were measured in a self-administered, mailed questionnare which had a response rate of 70 percent (N = 2044) over all 3 generations. The questionnaires were mailed in two waves: a period of six months elapsed between the mailing of the first and second wave. The religious and political items were measured on the first wave. Because the response rate on the second wave was somewhat smaller, the number of cases used in the construction and analysis of the gender ideology scale is somewhat smaller (N = 1585). Comparison of respondents and non-respondents on age, education, sex, and income revealed no

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions

ATTITUDE SIMILARITY IN FAMILIES 689

significant attrition bias between the two panel waves.

Dyad Selection Attitudes concerning three substantive areas (polit- ical, religious and gender ideologies) were com- pared in order to determine the differences between parents and their children at different stages of the life-course in these areas. This was accomplished by the construction of dyad comparisons in which the summated scores of the children on these items were subtracted from those of the parent. Because we are studying three-generation lineages, two types of dyads exist in this analysis. The first set of dyads consisted of first and second generation (G1-G2) parents and children (N = 478); the second set, the second and third generation grouping (G2-G3) (N = 1004). We are comparing the opinions and attitudes of grandparents (Gl) with the opinions and attitudes of parents (G2), as well as comparing the opinions and attitudes of parents (G2) with those of grandchildren (G3). Because every child in the sample was compared to each participating parent, some respondents were entered into the analysis more than once (in the case of families with more than one child or more than one parent responding to the survey). A total of 1482 dyads were constructed for the analysis. Approximately 45 percent of the dyads contain either a parent or child record that has appeared as such elsewhere in the remaining dyads. 30 percent of the dyads contain both a parent and a child record that have already appeared in the sample of dyads. Approximately 9 percent of the dyads contain parent records that are replicated more than once in other dyads in the sample. This over- representation of some dyad members presents the potential for attenuation of the distribution that might not occur otherwise. However, a similar sampling procedure was employed by Acock and Bengtson (1980) using these same data; they tested the degree to which such sampling-with-replication resulted in any increase or decrease in the level of predictivity and found it did not.

The lack of independence among the sampling units is a serious issue. However, each parent-child dyad is an unique unit of analysis that is not duplicated; even though one member may appear more than once in the total sample of dyads. Not to include the duplicated member dyads would also have the effect of underrepresenting large families and two parent families, further jeopardizing the representativeness of the sample. The nature of the data collection process in this survey precludes easy elimination of replicated member dyads. Such an elimination process would effectively halve the sample.

MEASUREMENT From the attitude and opinion items within the mailed questionnaire, three distinct scales, measur-

ing three substantive areas, were used. The items in these scales have a forced choice, Likert-type format with four response options that range from strongly agree to strongly disagree. Factor analysis was used in order to determine how the scales would be constructed from these items. One advantage of this procedure is that scale score comparisons could be made in order to avoid the attenuation of correlations that occurs when using single-item comparisons (Bohrnstedt and Carter, 1971). Once appropriate items for inclusion were ascertained for each scale, scale scores were computed by adding to relevant items together and dividing the result by the total number of items used. Thus, mean scores are comparable across scales.

Gender Ideology Scale

Out of 12 possible questions that might have been included in the scale measuring gender ideology, five had high factor loadings. These items and their factor loadings for the entire sample are: “Wives should obey their husbands” (.62); “Men cannot respect a fiancee who has had sex” (.45); “Husbands should have the main say in marriage” (.60); “Womens’ lib makes sense” (.53); and “Women should not have authority over men” (.45). Coefficient alpha was used as a measure of reliability for each of these scales-for the overall sample as well as for each generation. The alpha coefficients for the overall sample, G1, G2, and G3, are .72, .62, .72, and .75 respectively. Because each generation represents a unique position within the lineage, factor coefficients by generation were examined as well. The factor loadings followed a similar pattern for each generation.

Religious Ideology Scale

The religious ideology scale (reflecting conserva- tive Christian opinions) contains four items with high factor loadings. For the total sample, the factor loadings for each item are: “Every child should have religious instruction” (.66); “God exists as in the Bible” (.90); “The United States would be better if religion had more influence” (.78); and “We are all decendents of Adam and Eve” (.76). Coefficient alpha for each generation for this scale is: .85 (Gl), .85 (G2) and .83 (G3). The total sample reliability is .85.

Political Ideology Scale

The factor loadings for the five items in this scale are: “The United States should be ready to answer any challenge to its power, anywhere in the world” (.45); “Student demonstrators deserve strongest punishment possible” (.58); “Society’s most important task is law and order” (.66); “It is

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions

690 AMERICAN SOCIOLOGICAL REVIEW

a man’s duty to work; it is sinful to be idle” (.53); and “Most people on welfare are lazy; they just won’t do a good day’s work and so cannot get hired” (.56). The alpha coefficients for GI and G3 are somewhat low (GI = .62, G2 = .72, and G3 =.66). However, dropping one item from the scale would have increased unreliability, so the entire five item scale was retained with total sample reliability of .67. Unfortunately, both this scale and the religious ideology scale contain only items that are worded in a conservative direction, raising the possibility that response biases exist.

Mean scale scores by generation for each attitude scale are shown in Appendix 1. Note that N’s are somewhat smaller for the gender ideology scale, due to the lower response rate on wave B. Scores on all three attitude scales increase with generational position, indicating greater conserva- tism in older generational cohorts. Differences between the generations were statistically signifi- cant across the three scales (F = 47.56 for political ideology, F = 10.03 for gender, F – 17.37 for religious ideology).

Social Status Variables

Factors that are assumed to influence one’s opinions and attitudes concerning religion, gender ideology, and political issues include certain measures of social status that can be used to describe an individual’s experiences and interests within society. It is assumed that the experiences implied by such variables as marital status and gender, for example, have an impact upon the attitudes and opinions that one holds regarding these issues. In this analysis, eight social status variables were used to predict differences in attitude and opinion between parents and children. These variables are: gender, age, marital status, number of children, occupational prestige, labor force participation, educational attainment, and income. The variables of labor force status and marital status were dichotomized into working or non-working and married or non-married. Occupa- tional prestige was measured with Duncan’s occupational prestige scores. Educational attain- ment and income were both measured by ordinal scales, containing more than seven categories.

RESULTS

Trends in Generational Similarity

Hypotheses 1 predicted larger generational differ- ences between youth and their parents than between middle-aged adults and their parents. Inspection of Table 1 indicated that this was not true across the three scales. None of the dyad contrasts were statistically significant, meaning that there is no evidence from this table to suggest any convergence in attitudes between adjacent generations with age. Contrary to our original

Table 1. Mean Absoloute differences between Genera- tional Dyads in 3-generation Families

Political ID Gender ID Religious ID

Grandfather/ .63 .59 .64 Father .06 .07 .07

(75) (57) (80) Grandfather/ .64 .76 .69 Mother .03 .07 .05

(167) (114) (161)

Grandmother/ .65 .55 .67 Father .05 .07 .07

(84) (57) (90) Grandmother/ .68 .78 .61 Mother .04 .05 .05

(152) (123) (144)

Gl-G2 Total .65 .70 .65 .02 .03 .03

(478) (351) (475)

Father/ .66 .72 .78 Son .03 .05 .04

(206) (146) (209)

Father/ .61 .75 .65 Daughter .03 .04 .04

(246) (196) (251)

Mother/ .57 .68 .77 Son .03 .04 .04

(255) (177) (241)

Mother/ .59 .75 .60 Daughter .02 .04 .03

(297) (228) (275)

G2-G3 Total .60 .73 .69 .01 .02 .02

(1004) (747) (976)

Standard errors are reported under each mean. (N in parentheses)

expectation, attitude differences appear to be of approximately the same magnitude whether one is looking at grandparent-parent (G1-G2) dyads or parent-adult child dyads (G2-G3).

It is important to note that this result may arise out of an unspecified cohort effect. Older cohorts may have entered adulthood with smaller parent- child differences than modem cohorts, but time since childhood has increased those parent-child differences until they equal the current parent-child differences of young adults. However, it is plausible that today’s young adults will also increase their parent-child differences as they age in the future. Longitudinal data, which will be available shortly from these same families, can be used to test this hypothesis.

Table 1 reports the results of the first analysis for all three scales by generation and sex as well. In general, generational differences are modest in size, with means less than 1 on a maximum five-point scale. The largest aggregate differences of opinion between parents and children were

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions

ATTITUDE SIMILARITY IN FAMILIES 691

found for the gender ideology scale (.70 for G1-G2 dyads; .73 for G2-G3 dyads). No consistent effects of sex composition on dyadic agreement across generations were found, supporting Acock and Bengtson’s (1978) earlier contention that few sex-specific influences exist in comparing attitude differences between generations.

Status Inheritance

The evidence from Table 1 demonstrating substan- tial intergenerational agreement should not neces- sarily be taken as evidence of strong parental influence across generational dyads. At this point, both methodological and substantive problems impede a straightforward interpretation of mean attitude agreement measures. Mean attitude agree- ment measures cannot conceptually address the issue of whether parents’ attitudes actually influ- ence their children’s attitudes (Acock, 1984). Obviously, mean attitude differences across gener- ations can obscure the variation among families in the extent of generational agreement. Therefore, we have refined our initial findings of intergenera- tional agreement by constructing regression models in which children’s attitudes are predicted as a function of both parent’s attitude and a set of variables describing the child’s own social sta- tuses. It should be noted that such a regression model allows for cohort effects between genera- tions on these attitude scores.

All regression models have been estimated with LISREL, to incorporate a measurement model for parents’ and children’s scale scores into the regression framework. In addition, regression models for G1-G2 dyads and G2-G3 dyads were estimated together in one LISREL model using the option for multiple group data. This framework insured that the factor structures for the attitude scales could be made invariant across all genera- tions, and that statistical tests could be performed to detect significant differences in regression coefficients across the two dyad types.

The constrained measurement models for politi- cal, gender, and religious ideology fit the data quite well.5 For the measurement model for political ideology, the overall coefficient of determination (R2) was .77 for Gl-G2 dyads and .78 for G2-G3 dyads. For gender, the coefficients were .86 for GI-G2 and .93 for G2-G3 dyads. Finally, the measurement model for religious ideology produced coefficients of .99 for G1-G2

dyads and .97 for G2-G3 dyads. These models were tested statistically by constructing null models in which factor structures were not constrained across generations (Sobel and Bohrnstedt, 1985). The resulting chi-square difference tests revealed statistically significant generational differ- ences in factor structure for political ideology (X2 = 400, 12 df), religious ideology (X2 = 53, 9 df), and gender ideology (X2 = 45, 12 df). We attach substantive importance to the findings for the political scale only, since the increments to chi-square for the other two scales are relatively small. Because of the theoretical importance of constraining factor loadings across generations (Thomson and Williams, 1982), the constrained models were used in all analysis. However, these constraints are not empirically supported by the data for the political ideology scale, in particular.

As is shown in the analyses to follow, greater generational agreement is generally synonymous with greater parental influence in these dyads, as measured by the amount of variance in children’s attitudes explained by parental attitudes. Table 2 presents the results of the regressions of adult child’s attitudes on parents’ attitudes and adult child’s social status variables for the two dyad types, using LISREL to model the measurement of social ideologies. This insures that unreliability or measurement error does not attenuate the relation- ships between child attitudes, child status vari- ables, and parental attitudes. It is clear from Table 2 that adding social status variables to the null model of parental influence alone results in a significant improvement in model fit. The decrease in chi-square was 266 for political ideology, 257 for gender ideology, and 289 for religious ideology (all with 28 df).

Hypothesis two predicts that parental attitudes should continue to significantly predict children’s attitudes in these dyads. The significant impact of parental attitudes on adult children’s attitudes does persist across all three scales, even after controls for age, marital status, labor force status, educa- tion, number of children, and family income are added to the equation. It is clear from these results that parental influence is not reducible to the transmission of social status, although the coeffi- cients for parental attitudes drop with the addition of social status variables to the equation. Status transmission can account for some of the attitude continuity displayed across generations, but there are definitely family socialization effects that exert an independent influence on children’s attitudes past young adulthood.

Hypothesis two also predicts that parental attitudes will have a stronger impact in G2-G3 dyads than in G1-G2 dyads. Constraining the parental coefficients to be equal across dyad type produced an insignificant increase in the overall chi-square statistic for each model, indicating that

5 The errors of measurement were not correlated by design in these models. LISREL modification indices showed only scattered error correlations that, if esti- mated, might improve model fit. These instances fit no pattern or a priori theory of measurement error (for instance, correlating one parent error term with a different child error term).

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions

692 AMERICAN SOCIOLOGICAL REVIEW

Table 2. Regression of Child’s Attitude on Parent’s Attitude and Child Social Status Variables

Political Gender Religious

Variables G1-G2 G2-G3 G1-G2 G2-G3 G1-G2 G2-G3

Parental .56** .50** .53** .49** .44** .27** .37** .32** .51** .44** .59** .55** Attitude (.54) (.49) (.51) (.48) (.40) (.25) (.34) (.30) (.51) (.44) (.58) (.55)

Age -.01 .02 -.001 -.02 -.01 -.01 (-.08) (.17) (-.004) (-.09) (.04) (-.04)

Labor Force (.03) .06** -.01 .07** -.01 .03* Status (.10) (.19) (-.02) (.14) (-.02) (.11)

Education -.10** -.12* -.11** -.15** -.01* -.04 (- .24) (- .31) (- .18) (- .24) (- .02) (-.10)

Income -.003 .00 -.01* .001 -.01* .00 (- .07) (.01) (-.16) (- .02) (- .13) (.02)

Number of .02 .09** .08** .16 .05** .10 Children (.04) (.20) (.11) (.23) (.12) (.22) Marital

Status -.05 -.05* 1.l3* -.05 -.09* -.02 (Married = 1) (-.11) (- .11) (-.17) (-.06) (-.20) (-.04)

Sex -.14** .01 -.17 -.11 .01 .12 (- .15) (.01) (-. .12) (- .07) (.01) (.12)

R 2 .22 .30 .30 .38 .14 .25 .12 .19 .23 .28 .34 .38 Goodness of

fit Index .927 .936 .927 .953 .912 .939 X2/df 1109/332 669/332 710/243

* p c .05.

** p ‘ .01. (standardized effects in parentheses)

the absolute size of the parental coefficient does not differ across older and younger dyads.

Turning to Table 3, we separated the unique contributions of parental attitudes and child social status variables as a group to the explained variance in the attitude equations displayed in Table 2. Substantial differences were found in the predictive power of parental attitudes across scales and across generations. Controlling for social status, grandparents predicted parents’ scores less well than parents predicted their young adult children’s scores across all three attitude scales. The R2 increments for G1 parental attitudes on G2 children’s scores were .15, .04, and .14 for the political, gender, and religious ideology scales, respectively. The corresponding figures for G3

children were higher, .24, .08, and .26, respec- tively. Conversely, social status variables indepen- dently predicted slightly more of the variance in G2 (parents’) scores than G3 (adult children’s) scores, for the political and gender ideology scales. While not definitive, these results suggest that the importance of parental attitudes as determinants of children’s attitudes decreases with age, while the importance of social structural variables as deter- minants of attitudes only slightly increases with age.

Looking at between-scale differences, one can see that parents’ scores were much more predictive of children’s scores for the religious and political ideology scales than for the gender scale. The predictive power of parents’ scale scores seems to

Table 3. Decomposition of R2 into Unique Contributions of Parental Attitudes and Child’s Social Status Variables

Political Gender Religious Ideology Ideology Ideology

GJ-G2 Parental Attitude .15 .04 .14 Social Status Variables .09 .11 .05

Total R2 .30 .25 .28

G2-G3 Parental Attitude .24 .08 .26 Social Status Variables .08 .07 .04

Total R2 .38 .19 .38

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions

ATTITUDE SIMILARITY IN FAMILIES 693

complement the degree of parent-child similarity in scale scores. Parents’ scores predicted at least 14 percent of the variance in their children’s religious and political attitudes, even after relevant social status variables were controlled. Yet, parents’ scores explained well under 10 percent of the variance in their children’s gender ideology scores. This is not overwhelming evidence of direct family influence on gender attitudes. In contrast, social structural variables explain a roughly consistent proportion of the variance across attitude scales. For example, among G2-G3 dyads, social struc- tural variables explain 8 percent, 7 percent, and 7 percent of the variance in G3 responses for political, gender, and religious ideology respec- tively. For G1-G2 dyads, structural variables account for 9 percent, 11 percent, and 5 percent of the explained variance in scale scores. Clearly, social structural variables do not replace parental influence where influence is low.

Reciprocal Influence

Our last analysis addresses the reciprocal nature of the attitude influence process-both parent and

child influence. A path model describing the model of the influence process estimated with LISREL is shown in Figure 1. It is assumed that parents’ social statuses affect children’s attitudes only indirectly through parents’ attitudes6. These mod- els were initially estimated with and without correlated disturbance terms. The simple uncorre- lated model was more parsimonious in each case,

6 Examination of the normalized residual correlations indicates that this assumption is accurate for the older dyads; less so for the younger dyads. For instance, among younger dyads, there is evidence that parents’ income and education independently affect political ideology. However, it is plausible that these parent effects are proxies for unmeasured neighborhood or peer-group influences on young adult’s political ideol- ogy. It appears that parent’s marital status may also independently influence religious ideology, although the mechanism through which this effect operates is open to speculation. Finally, parents’ employment status may independently affect young adults’ gender ideology, although the effect is weak and, we suspect, confined to mothers’ employment status.

Figure 1. Model of Reciprocal Influence Between Generations

, _

/* p C

income income

educ marital C dcmarital p status P educ status p LFstatus #kids p c LFstatus / kidsC

age se x age sex

PARENT’S ATTITUDE CHILD’S ATTITUDE ( I ) political (I) political (2) gender (2) gender (3) religious (3) religious

7K I \\ Xup Uc 7K pl p2 p3 p,4 p5 cl c2 c3 c4 c5

E1 ?2 E3 E4 5 6l1 E2 83 84 85

STRUCTURAL EQUATIONS: MEASUREMENT MODELS:

Pattz BCatt +rpXp +Up pi APtt+E

Catt BPatt + rcXc + Uc Cj a ACatt + Ej

* all social status variables were allowed to freely correlate with all other social status variables

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions

694 AMERICAN SOCIOLOGICAL REVIEW

with corresponding increases in chi-square insignif- icant.

Table 4 displays the unstandardized estimates obtained from LISREL models for each attitude scale. The decrease in chi-squares between the models in Tables 2 and 4 indicate that the reciprocal effects models in Table 4 fit the data better than the parental influence models of Table 2 (X2=7,12,21, respectively, with 2 df). The pattern of results weakly supports hypothesis 3, that parent influences decline with age while child influences increase with age, controlling for social status. Large differences appear in the pattern of influence between parents and children across the three attitude scales and across the life-span. The results for political ideology best support hypothe- sis 3. In the younger G2-G3 dyads, parent influence on children is significant, while child influence on parents is not. However, in the older GI-G2 dyads this pattern is reversed-middle-aged children significantly influence their parents’ political ideology but parental influence is insignif- icant at this stage. For gender ideology, child effects on parents’ scale scores appear significant in both G1-G2 and G2-G3 dyads, while parent effects are insignificant across dyad type. This pattern of influence “upward” through the gener- ations challenges long held notions about the impact of family socialization on gender attitudes in adulthood. The last attitude scale, religious ideology, shows reciprocal influence of parents and children on each other among younger G2-G3 dyads, while older dyads show only child-to-parent

influence. We see that, once again, parent effects seem to become less significant with age.

Within LISREL, the chi-square statistic repre- sents the ability of the estimated parameters to reproduce the original variance-covariance matrix of the input data. The degrees of freedom represent the number of free sample moments (variances and covariances) unused in the process of estimating the number of parameters included in each model. Because the samples used are relatively large, it is difficult to estimate models which reproduce the original data well, or provide a “good fit.” None of our models in Tables 2 or 4 were able to reproduce the original data with a probability greater than .05. For each model, we report the ratio of chi-square to degrees of freedom. For all scales, the chi-square ratios suggest acceptable fits to the data (ratios ranging from 1.99 to 3.33). The LISREL-generated goodness-of-fit index is also uniformly high for all models (1.00 indicating perfect fit). In addition, model fit was assessed using Hoelter’s (1983) critical N method. This method determines the sample size needed to reproduce the data with a given model at an acceptable probability level. If that critical N exceeds 200 per group (400, in this case), then a given model fits the data reasonably well. Using Hoelter’s criteria, all the models produce accept- able fits to the data (CN = 431, 530, and 601 respectively).

To test the significance of differences in the influence process between younger and older dyads, models were estimated in which parent-

Table 4. Models of Reciprocal Influence on Attitudes, Controlling for Own Social Status Variables, by Dyad Type

Political Gender Religious

G1-G2 G2-G3 G1-G2 G2-G3 GI-G2 G2-G3

Parent Child Parent Child Parent Child Parent Child Parent Child Parent Child Parent’s

Attitude – .21 – .43** – -.15 – .13 -.31 – .34** Child’s Attitude .27** – .11 – 34** – .21** – .64** – .28** –

Age .01** -.00 -.00 .02 .00 .00 .01* -.01 .01 .00 -.00 -.01 Labor Force

Status – .02 .03 – .00 .06** .02 .00 – .01 .08** .02 .01 .02 .05** Education -.07** -.11** -.11** -.13** -.08** -.17** -.12** -.17** -.02 -.06* -.04** -.05* Income -.Ol* -.00 .00 -.00 -.02** -.02** -.01* -.00 -.O1* -.01** -.01** .00 Number of

Children -.01 .02 .03 .08* .05 .07** .11** .15 .01 .07** .05** 11** Marital Status -.02 -.06 -.01 -.05* .12 -.19** -.12 -.07 .03 -.14** -.15** -.04* Sex -.03 -.15** -.20** .01 -.03 -.21* -.22** -.10 .07 -.08 .03 .14**

x2 (df) 1102(330) 657(330) 689(241) x2ratio 3.33 1.99 2.86 Goodness-of fit index .93 .94 .93 .95 .91 .94 Critical N 431.41 530.02 601.10

*p < .05.

p < .01.

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions

ATTITUDE SIMILARITY IN FAMILIES 695

child effects, child-parent effects, or both were constrained to be equal across dyad type. Surpris- ingly few significant differences were uncovered among any of the attitude scales when constrained and unconstrained models were compared. For political ideology, the difference in chi-square between the fully constrained and unconstrained models was 1.76, with two degrees of freedom gained. Thus, no statistically significant difference exists between the size of the parent effect for younger and older dyads or between the size of the child effect for younger and older dyads. In addition, both the child-parent and parent-child effects were significantly different from zero in the newly estimated constrained model. The best- fitting constrained model for political ideology is displayed in Table 5.

For gender ideology as well, the most parsimo- nious model was the fully constrained model in which both parent-child and child-parent effects were constrained to have equal effects across dyad type. However, in this constrained model, only child effects on parents were significant. Finally, the best-fitting model for religious ideology was a partially constrained model in which only child-to- parent influences were set equal in the two dyad types. When parent-child effects were also con- strained, the difference in chi-square between models rose to 6.28 with only one degree of freedom gained. So, the difference in elderly and middle aged parents’ influence on children’s religious ideology was significant. Only middle aged parents’ religious attitudes significantly shaped their children’s religious beliefs; elderly parents apparently did not exert a strong independent influence on their children’s religious beliefs.

Looking across best-fitting models in Table 5, it is hard to avoid the conclusion that significant child-to-parent influence is more prevalent than parent-to-child influence. Child-parent effects are significant and equal across dyad types for all three scales. Parent-child influences are significant in younger dyads for political and religious ideologies only; for older dyads, it is political ideology alone that shows significant parent-child transmission.

DISCUSSION

This research has examined three issues concerning the transmission of attitudes across generations. The first involves the amount of ideological similarity between parents and children across life-course positions, as this may reflect increasing status similarity or the resolution of parent- adolescent conflict. Second, causal mechanisms underlying apparent continuity across generations were explored. We wished to test the possibility that observed similarities in attitudes are due to social status similarities and not to socialization. The third issue concerns the possibility of reciprocal influence. Intergenerational agreement can not necessarily be taken as evidence of parental influence, since observed similarity may be due to influence of children on their parents. Data addressing these issues suggest some important modifications of existing socialization and devel- opmental aging theory.

The first hypothesis suggested that parent-child attitudes converge with advancing age; specifically that G1-G2 dyads would show smaller attitudinal differences than would G2-G3 dyads. This hypoth- esis was not supported by the data. Attitude differences were small throughout the generational pairs, and the differences observed were the same in the younger (G2-G3) dyads as in the older (GI-G2) dyads. Some differences were manifest across ideological domains, with gender showing the greatest contrasts. These findings suggest that an uncritical use of life-course position to predict varying levels of parent-child difference can lead to overgeneralization; substantial continuity is seen across different points in life represented in this study.

The second hypothesis attempted to disentangle developmental aging and status inheritance as sources of attitude similarity. The hypothesis that parents’ attitudes predict children’s attitudes, after controlling for children’s social status, was con- firmed. However, the level of parental prediction drops with the addition of social status variables, indicating the importance of status transmission mechanisms. It should be noted that parents’

Table 5. Final Estimates of Reciprocal Effects’

G1-G2 G2-G3

parent child parent child

Political parent’s attitude – .379** – .379** ideology child’s attitude .168* – .168*

Gender parent’s attitude – .074 – .074 ideology child’s attitude .237** – .237**

Religious parent’s attitude – .14 -.31** ideology child’s attitude .31** – .31**

1 Since social status effects were listed in Table 4, they are omitted from this table.

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions

696 AMERICAN SOCIOLOGICAL REVIEW

scores are more predictive of children’s scores on religious and political ideology scales than on the gender scale. Also confirmed was the crucial second part of the hypothesis, that the parental effect would decrease in importance with age; that is, parental attitudes would have a stronger influence in G2-G3 dyads than in G1-G2 dyads. But social status variables do not replace parental influence in scales where influence is low, nor do they increase in predictive importance with age, as measured by generational position.

The principal conclusion of our second analysis, that parental attitudes exert an influence indepen- dent of social status inheritance, and that these effects, though diminished, exist past early adult- hood, can be taken with the results of the first hypothesis to suggest the continuing importance of influence processes across generational positions. However, observed parent-child similarity need not reflect parental influence, and indeed the life-course or developmental aging perspective points to reciprocal influence as the causal mechanism. The third analysis estimated LISREL models of reciprocal influence, testing the hypoth- esis that child influences on parental attitudes increase with age, while parental influences on children’s attitudes decrease with age. The data did not confirm this hypothesis. Child influences were significant and equal in magnitude across younger and older dyads for all three attitude domains. However, when significant parental influence existed among younger generational pairs, that influence did tend to decline among older genera- tional pairs. Gender ideology, the domain affected most by rapid social change, showed a pattern of only upward transmission from children to parents across the generations.

These results, taken together, suggest that while the extent of parent-child attitude similarity appears relatively stable across successive genera- tions, the forces generating these similarities appear to change over time. Direct parental influence declines in older generation dyads, while social structural variables only slightly increase in importance as predictors of attitudes. However, child influences on parents are strong in early adulthood, and stay strong over the life course. This implies that social-structural similarities and child influence produce parent-child similarities later in the life course, while reciprocal influence may produce more parent-child similarity in younger generational dyads.

The findings of this study suggest three conclusions concerning the family as an agent of socialization over the life course. First, it is important to recognize relational change beyond primary socialization. Evidence of significant influence upward through the generations suggests that the family may act as an agent of change, not an impediment to change as is implied by many conceptualizations of family socialization.

Second, one must examine the causal mecha- nisms behind observed continuity or change in socialization outcomes, especially those structural or status similarities between parents and children that are often undifferentiated from “parental influence.” Status inheritance may be, as sug- gested by these results, an important alternative route to inter-generational similarity.

Finally, variability in the impact of parent-child relations across social ideologies should be acknowl- edged. While religious and political ideologies clearly emerge as areas of strong independent family influence, gender ideology seems less affected by internal family dynamics. Perhaps fewer competing agents of socialization exist for religious or political attitudes, or perhaps these domains are less profoundly related to daily living than gender ideology, giving children little reason to question their parents’ beliefs. It may also be true that parental influence weakens during periods of rapid changes in social behavior. Whatever the source, it is clear that the family is neither a monolithic nor necessarily conservative source of influence on attitudes or beliefs past childhood.

Appendix 1. Mean Scale Scores for each ideological scale by generation

61 62 63 F Political 3.0 2.7 2.3 47.56* ideology (484) (661) (779) Gender 2.6 2.4 2.2 10.03* ideology (387) (518) (597)

Religious 3.3 3.3 2.7 17.37* ideology (479) (645) (754)

(N in parentheses)

Appendix 2. LISREL factor loadings for each attitude scale (constrained to be invariant across generations)

Item # Political Gender Religious 1 1.00 1.00 1.00 2 1.184 .692 2.094 3 .056 .956 1.473 4 1.390 .776 2.070 5 1.310 .965 –

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions

ATTITUDE SIMILARITY IN FAMILIES 697

Appendix 3. LISREL generated Error and Disturbance Terms for each latent variable (full models displayed in Table 4)

Political Gender Religious

P’s C’s P’s C’s P’s C’s G1-G2 attitude attitude attitude attitude attitude attitude

disturbance (psi) .107 .186 .257 .409 .164 .274 El 1.126 .904 1.070 .679 .275 .407

Errors E2 .613 .696 1.202 .873 .148 .148 of E3 1.033 1.232 .836 .670 .350 .374

measurement E4 .596 .617 .942 .559 .455 .525 Es .853 .602 .784 .614 – –

G2-G3 disturbance (psi) .162 1.28 .337 .472 .132 .147 El 1.004 .828 .745 .631 .339 .569 E2 .633 .871 .989 .659 .237 .417 E3 1.032 1.040 .676 .718 .340 .605 E4 .719 .581 .542 .655 .465 .567 E5 .574 .739 .611 .550 – –

REFERENCES

Acock, Alan C. 1984. “Parents and Their Children: The Study of Inter-Generation Influence.” Sociology and Social Research 68:151-71.

Acock, Alan C. and Vern L. Bengtson. 1980. “Social- ization and Attribution Processes: Actual vs. Perceived Similarity Among Parents and Youth.” Journal of Marriage and the Family 40:519-30.

Adorno, Theodor W., et. al. 1950. The Authoritarian Personality. New York: Harper.

Angress, Selma. 1975. Intergeneration relations and value congruence between young adults and their mothers.” Unpublished Ph.D. dissertation. University of Chicago, Committee on Human Development.

Baltes, Paul B. 1979. “Life Span Developmental Psychology: Some Converging Observations on His- tory and Theory. Pp. 256-74 in Life Span Develop- ment and Behavior, Vol. 2, edited by Paul Baltes and Orville Brim. New York: Academic Press.

Bell, Robert and L. V. Harper. 1977. Child Effects on Adults. New York: Wiley.

Bengtson, Vern L. 1975. “Generation and Family Effects in Value Socialization.” American Sociologi- cal Review 40:358-71.

Bengtson, Vern L., Neal E. Cutler, David J. Mangen, Victor W. Marshall. 1985. “Generations, Cohorts, and Relations Between Age Groups. Pp. 304-38 in Handbook of Aging and the Social Sciences, edited by Robert Binstock and Ethel Shanas. New York: Van Nostrand Reinhold Company.

Bengtson, Vern L. and Joseph A. Kuypers. 1971. “Generational Difference and the Developmental Stake.” Aging and Human Development 2:249-60.

Bengtson, Vern L. and Lillian Troll. 1978. “Youth and Their Parents: Feedback and Intergenerational Influ- ence in Socialization.” Pp. 215-40 in Child Influences on Marital and Family Interaction: A Life-Span Perspective, edited by Richard M. Lerner and Graham B. Spanier. New York: Academic Press.

Bohrnstedt, George W. and T. Michael Carter. 1971. “Robustness in Regression Analysis.” Pp. 118-46 in Sociological Methodology, edited by Herbert L. Costner. San Francisco: Jossey-Bass.

Campbell, Ernest Q. 1969. “Adolescent Socialization.” Pp. 821-60 in Handbook of Socialization Theory and Research, edited by David A. Goslin. Chicago: Rand-McNally.

Chaffee, Stephen H., Jack M. McLeod and Charles K. Atkin. 1971. “Parental Influences on Adolescent Media Use.” American Behavioral Scientist 14:323-40.

Chodorow, Nancy. 1978. The Reproduction of Mother- ing. Berkeley: University of California Press.

Dalton, Russell J. 1980. “Reassessing Parental Social- ization: Indicator Unreliability Versus Generational Transfer. ” American Political Science Review 74:421-31.

Elder, Glen H., Jr. 1974. Children of the Great Depression. Chicago: University of Chicago Press.

. 1984. “Families, Kin, and the Life Course: A Sociological Perspective.” Pp. 80-136 in Advances in Child Development Research: Vol. 7: The Family, edited by Ross Parke. Chicago: University of Chicago Press.

Engels, Frederik. 1967. The Origin of the Family, Private Property, and the State. New York: Interna- tional Publishers.

Erikson, Erik. 1950. Childhood and Society. New York: Norton.

Featherman, David L. 1983. “Life-Span Perspectives in Social Science Research.” Pp. 1-59 in Life-Span Development and Behavior, vol. 5, edited by Paul B. Baltes and Orville G. Brim, Jr. New York: Academic Press.

Freud, Sigmund. 1933. New Introductory Lectures in Psychoanalysis. New York: Norton.

Goslin, David A. 1969. Handbook of Socialization Theory and Research. Chicago: Rand McNally.

Hagestad, Gunhild 0. 1981. “Problems and Promises in the Social Psychology of Intergenerational Relations.” Pp. 11-46 in Aging: Stability and Change in the Family, edited by Robert W. Fogel, Elaine Hatfield, Sara B. Kiesler, and Ethel Shanas. New York: Academic Press.

. 1984. “The Continuous Bond: A Dynamic Multigenerational Perspective on Parent-child Rela- tions between Adults.” Pp. 129-58 in Minnesota Symposium on Child Psychology, edited by Marion Perlmutter. New York: Erlbaum.

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions

698 AMERICAN SOCIOLOGICAL REVIEW

Heilbrun, Authur B., Jr. 1965. “An Empirical Test of the Modeling Theory of Sex-role Learning.” Child Development 36:789-99.

Hess, Beth B. and Joan M. Waring. 1978. “Parent and Child in Later Life: Rethinking the Relationship.” Pp. 241-74 in Child Influence on Marital and Family Interaction, edited by Richard M. Lerner and Graham B. Spanier. New York: Academic Press.

Hill, Reuben, Nelson Foote, Joan Aldous, Robert Carlson and Robert MacDonald. 1970. Family Devel- opment in Three Generations. Cambridge, MA: Schenkman.

Hoelter, Jon W. 1983. “The Analysis of Covariance Structures: Goodness of Fit Indices.” Sociological Methods and Research 11:325-44.

Hoge, Dean R., Gregory Petrillo and Ella Smith. 1982. “Transmission of Religious and Social Values from Parents to Teenage Children.” Journal of Marriage and the Family 44:569-80.

Hyman, Herbert. 1959. Political Socialization. New York: Free Press.

Jennings, M. Kent and Richard Niemi. 1982. Genera- tions and Politics: A Panel Study of Young Adults and Their Parents. Princeton, NJ: Princeton University Press.

. 1968. “The Transmission of Political Values from Parent to Child.” American Political Science Review 42:169-84.

Lerner, Richard M. and Graham B. Spanier. 1978. Child Influence on Marital and Family Interaction. New York: Academic Press.

Lewis, Michael and Leonard A. Rosenblum. 1974. The Effect of the Infant on its Caregiver. New York: Wiley.

Mannheim, Karl. 1952. “The Problem of Generations (1928).” Pp. 276-320 in Essays on the Sociology of Knowledge, edited by D. Kecskemeti. London: Routledge and Kegan Paul.

Mason, Karen O., John L. Czajka and Sara Arber. 1976. “Change in U.S. Women’s Sex-role Attitudes, 1964-1974.” American Sociological Review 41:573-96.

Moss, Nancy E. and Steven I. Abramowitz. 1982. “Beyond Deficit-filling and Developmental Stakes: Cross-disciplinary Perspectives on Parental Heritage.” Journal of Marriage and the Family 44:357-66.

Parsons, Talcott and Robert Bales. 1955. Family, Socialization, Interaction Process. Glencoe, IL: Free Press.

Riley, Matilda W., Beth B. Hess, and Kathleen Bond. 1982. Aging in Society: Selected Reviews of Recent Research. Hillsdale, NY: Erlbaum.

Rossi, Alice A. 1980. “Aging and Parenthood in the Middle Years.” Pp. 138-207 in Life-Span Develop- ment and Behavior, vol. 3, edited by Paul B. Baltes and Orville G. Brim. New York: Academic Press.

Shanas, Ethel. 1979. “Social Myth and Hypotheses: The Case of the Family Relations of Old People.” The Gerontologist 19:3-9.

Smith, M. Dwayne and George D. Self. 1980. “The Congruence Between Mother and Daughters in Sex Role Attitude: A Research Note.” Journal of Marriage and the Family 42:105-109

Smith, Thomas E. 1983. “Parental Influence: A Review of the Evidence of Influence and Theoretical Model of the Parental Influence Process. ” Pp. 13-45 in Research in Sociology of Educational Socialization, vol. 4, edited by Alan Kerkhoff. Greenwich, CT: JAI Press.

Sobel, Michael and George Bohrnstedt. 1985. “Use of Null Models in Evaluating the Fit of Covariance Structure Models.” Pp. 152-78 in Sociological Methodology, edited by Nancy Tuma. San Francisco: Jossey Bass.

Thomas, William I. and Florian Znaniecki. 1958. The Polish Peasant in Europe and America. New York: Dover.

Thomson, Elizabeth and Richard Williams. 1982. “Beyond Wives’ Family Sociology: A Method for Analyzing Couple Data.” Journal of Marriage and the Family 44:999-1008.

Thornton, Arland, Duane F. Alwin and Donald Camburn. 1983. “Sex-role Attitudes and Attitude Change.” American Sociological Review 48:211-27.

Troll, Lillian, Sheila J. Miller, and Robert Atchley. 1979. Families in Later Life. Belmont, CA: Wadsworth.

Weiting, Stephen G. 1975. “An Examination of Intergenerational Patterns of Religious Belief and Practice and Practice.” Sociological Analysis 36: 137-49.

This content downloaded from 128.110.224.115 on Fri, 23 Aug 2013 17:46:05 PM All use subject to JSTOR Terms and Conditions