geronb      J Gerontol B Psychol Sci Soc Scigeronb      The Journals of Gerontology Series B: Psychological Sciences and Social Sciences      J Gerontol B Psychol Sci Soc Sci      1079-5014      1758-5368              Oxford University Press                    9909310.1093/geronb/56.1.S36                        Journal of Gerontology: Social Sciences                            Changing Attitudes Toward Aging Policy in the United States During the 1980s and 1990s        A Cohort Analysis                                          Silverstein            Merril                                a                                                Angelelli            Joseph J.                                b                                                Parrott            Tonya M.                                c                          aAndrus Gerontology Center, University of Southern California, Los Angeles        bCenter for Gerontology and Health Care Research, Brown University, Providence, Rhode Island        cDepartment of Sociology, Quinnipiac University, Hamden, Connecticut                    Merril Silverstein, University of Southern California, Andrus Gerontology Center, 3715 McClintock Ave., Los Angeles, CA 90089-0191 E-mail: merrils@usc.edu.                    1        1        2001            56      1      S36      S43                        22          8          2000                          20          9          1999                            The Gerontological Society of America        2001                    Objectives.        This research assessed how the attitudes of Americans toward government programs that serve older people changed between the mid-1980s and late 1990s and how much of the shift was due to intracohort change and how much was due to cohort replacement.        Methods.        Data come from three nationally representative cross-sectional samples, surveyed by telephone in 1986 (N = 1,209), 1990 (N = 1,500), and 1997 (N = 1,559).        Results.        Attitudes of Americans have become less supportive of expanding entitlement programs for older people and more supportive of cutting their costs and benefits. Between 1986 and 1997, most cohorts, particularly older adults, grew more in favor of maintaining Social Security benefit levels but less in favor of expanding them. Young adults tended to be driving the societal shift in attitudes toward decreasing benefits. Intercohort change was more important than cohort replacement in this process. Analyses of change in 2 attitude domains between 1990 and 1997 revealed that the general population felt less strongly that older people are entitled to benefits and expressed greater opposition to the associated costs. However, young adults moderated their concerns about costs as they got older, although the young adults in the cohort replacing them had become more critical of the principle of entitlement.        Discussion.        These findings enhance the understanding of the roles that historical conditions and aging play in shaping the attitudes of adult cohorts toward public programs for older citizens. Discrepant findings based on the intercohort change in younger age groups are reconciled by differentiating maturation effects from period effects on impressionable youth.                              hwp-legacy-fpage          S36                          hwp-legacy-dochead          RESEARCH ARTICLE                          IN the United States, as in almost all developed nations, support of the older population has largely been collectivized through government-sponsored income maintenance and health care programs. However, over the last several decades, rising life expectancies and the impending aging of the baby-boom cohort have created anxiety concerning the economic viability of entitlement programs serving a swelling older population. Messages in the mass media through much of the early 1990s (e.g., Becker 1994; D'Antonio 1993; Duff 1995; Jackson 1994; Romero 1994; Samuelson 1994; Wolfe 1995) have reflected—as well as stirred—public fears concerning the anticipated demands placed on the retirement system by aging baby boomers (Cornman and Kingson 1996; Duff 1995; Wolfe 1995). It is not unusual for today's elderly to be blamed for these funding dilemmas, for instance, when they are portrayed as a leisure class of affluent retirees unfairly benefitting from the taxes paid by a younger and poorer working-age population (D'Antonio 1993; Jackson 1994; Marshall, Cook, and Marshall 1993; Romero 1994; Samuelson 1994).    Public anxiety over the consequences of a swelling older population, combined with greater wariness of "big government" on the part of the public, has revived the debate concerning the desirability of large-scale, universal, and increasingly expensive public programs that provide services and benefits as an entitlement of old age. Scholars have expressed concern that insecurity over the future of entitlement programs for the elderly may erode public support for these programs, particularly in younger generations (Achenbaum 1986; Bengtson 1993; Bengtson and Achenbaum 1993; Kingson and Williamson 1993; Myles 1995). If the late 20th century was truly a watershed period in age-group relations, then one would expect a change on the part of the public in their support for programs that benefit older people. We asked two basic questions in this investigation: (a) Have Americans grown less supportive of entitlement programs serving older people? (b) If so, how much of this weakening is the result of shifts in the attitudes of individuals through time (intracohort change), and how much is due to the emergence of a young adult cohort with a distinctly more negative attitude toward entitlements of older people than the cohort it replaced (cohort replacement)?    Two general explanations for why age strata may vary in their attitudes toward public benefits for older people have to do with their stage in the life cycle and the cohort into which they were born. Life-cycle effects on attitudes are those that correspond to the imperatives imposed by the needs and incentives associated with chronological age. Cohort effects are those that correspond to the disproportionate influence that historical periods have on the attitudes of certain age groups, particularly on the attitude formation of impressionable young adults.    Life-cycle factors are addressed by Ponza, Duncan, Corcoran, and Groskind 1988, who advance three mechanisms for explaining variations in attitudes by age: self-interest, insurance, and altruism. In terms of self-interested motivations, one would expect the younger individuals to be less strongly in favor of protecting old-age benefits than would the older population because they derive little direct benefit from these programs (Rhodebeck 1993; Shapiro, Patterson, Russell, and Young 1987). An insurance model proposes that individuals who anticipate their own old-age needs would be just as protective of benefits to older people as are older persons themselves. Under the assumptions of this model, one would expect that elevated support for protecting old-age benefits would begin in middle age. Altruism would be evident if age groups indirectly benefitting from the policy, such as young adults, support liberal benefits just as strongly—or more strongly—than the older population. Models that mix altruism and self-interest propose that the interests of the middle aged are served by old-age programs that accommodate the needs of older generations and that allow them to shift the potential costs of their own elder-care responsibilities to publically funded programs. We are not in a position to directly test these models, although they inform our discussion concerning the effects of age on public attitudes toward entitlement programs for older people.    The influence of cohort membership is rarely articulated as an explanation for age-strata differences in attitude formation around issues of significance to older people. We maintain that without differentiating chronological age from cohort membership it is not entirely possible to understand whether support for public programs and privileges for older people is influenced by life-stage processes or by ideologies permeating the sociohistorical period in which birth cohorts adopt their attitudes. This may be why there is little consensus in the literature concerning the attitudes of younger adults toward programs targeting the older population, with some research finding higher support in this age group (Day 1993; Logan and Spitze 1995), some finding lower support (Kronebusch and Schlesinger 1994; Silverstein and Parrott 1997), and others finding few differences from other age groups (Cook and Barrett 1992).    In this context, our principal interest focused on the ideological crystallization of a young cohort that is rooted in its exposure to particular economic, cultural, and historical experiences. If young adults are particularly impressionable to the values and ideologies of the contemporary culture, then the upsurge in "elder-bashing" during the 1980s and 1990s would have created social change through the emergence of a new cohort of young adults into the population, one with values distinctly more negative on this issue than the one it replaced (Logan and Spitze 1995; Ponza et al. 1988; Schlesinger and Kronebusch 1994). For example, it has been suggested that young adults who had trouble entering the labor market during the recession of the early 1990s may have blamed economic stagnation on older people by relying on stereotypes of older people as affluent and selfish (Binstock 1983). Young adults may have been particularly susceptible to contemporary media messages that reinforce negative stereotypes of older people and engender a culture of ageism (Bishop and Krause 1984; Vasil and Wass 1993).    Our argument for the importance of cohort replacement derives from the ideas of Ryder 1965, who outlined how population turnover—the continuous succession of new cohorts into the adult population and the exit of older cohorts—operates as an engine of social change. As young cohorts reach adulthood, their "fresh contact" with the world serves to refresh social institutions and promote innovation (Mannheim, 1922/Mannheim 1952). This approach provides an important point of departure for understanding how cohort membership shapes attitudes. In short, the cohort perspective suggests that historical conditions leave an indelible imprint on the attitudes of young adults at a time when they are most susceptible to absorbing the social values of the period, a phenomenon known as the impressionable youth hypothesis (Alwin, Cohen, and Newcomb 1991; Clausen 1993; Elder 1994; Marwell, Aiken, and Demerath 1987). Crucial to this argument is the way personal biography aligns with historical contingencies to produce sharp deviations in the attitudes of cohorts emerging into adulthood. The life-course perspective provides a useful point of departure for speculating about sources of aggregate social change (if it does exist) in the values of adults toward programs serving older people, leading to the following questions: Has the public become increasingly skeptical about the value and utility of entitlement programs serving older people? Have the attitudes of all cohorts on this issue shifted consistently? Have younger cohorts with uniquely different attitudes replaced older cohorts to, at least partially, produce this societal change?    In this investigation we examined sources of aggregate change in the attitudes of Americans toward public programs providing support for the older population. We heeded Glenn 1976 admonition against trying to disentangle the well-known confound among age, period, and cohort effects. Instead, we used an approach outlined by Firebaugh 1997, in which social or aggregate change is decomposed into two components: intracohort change and cohort replacement. Intracohort (or individual) change refers to that component of societal change resulting from the maturation of cohorts over historical time, and cohort replacement refers to that component resulting from the process of population turnover—the successive replacement of younger cohorts into the population and the successive removal of older cohorts.    We first tracked whether preferences of the population for public spending on the Social Security program changed between 1986 and 1997 and identified the components of change due to intracohort change and cohort replacement during a period of heightened generational tension. Second, we investigated whether the attitudes of members of the post-baby-boom cohort (those born between 1966 and 1972) changed as they grew older over the 7 years between 1990 and 1997 and whether the next cohort of young adults that replaced them in 1997 (those born between 1973 and 1979) had distinctly different attitudes than their predecessors. In summary, we charted the degree of social change in the orientations of the public toward support of older people over the late 20th century and identified the components of that change related to the aging of cohorts and the dynamics of cohort replacement.          Methods              Samples        We used data from three nationally representative telephone surveys of adults aged 18 years and older in the United States conducted in 1986, 1990, and 1997. Each sample was generated by random digit dialing of households within the contiguous United States. One adult aged 18 years or older was randomly selected from each eligible household as the designated respondent. The first survey (N = 1,209) was conducted in 1986 by Cook and colleagues (for details, see Cook and Barrett 1992) and lasted approximately 45 min on average. The second and third surveys were conducted in 1990 (N = 1,500) and in 1997 (N = 1,559), and each lasted 35 min on average (for details, see Bengtson and Harootyan 1994; Silverstein and Parrott 1997). Participation rates were 71% for the 1986 survey, 79% for the 1990 survey, and 64% for the 1997 survey.        Because our analysis compared responses to a question regarding Social Security spending that was asked in 1986 and replicated in 1997 only, the first part of our investigation contrasted the full samples from these two surveys. We present the unweighted frequency distributions of sample characteristics for the 1986 and 1997 surveys in Table 1 . The sample in 1997 had significantly more education, were less likely to be married, and had a different age distribution with proportionately fewer young adults.        The second part of the investigation focused on young adults aged 18–24 years, an age group with which we were able to test the hypothesis that youthful persons are particularly impressionable when exposed to historical change. Because the 1990 survey recorded respondents' ages for those older than 24 in 10-year categories, we were not able to investigate cohort change in older groups given the 7 years between surveys. Thus, we here describe the process of cohort change and replacement in these cohorts between 1990 and 1997 when they comprised, respectively, 210 and 168 individuals. The later cohort was significantly more likely to be male and showed trends toward having more education and more minority representation than its predecessor. In addition, we used the group that was aged 25–31 in 1997 (n = 241) to demonstrate the degree of intracohort change.                    Dependent Variables        In the first stage of the analysis, public attitudes toward the Social Security program were assessed with the following question, asked in both 1986 and 1997: "Do you feel Social Security benefits should be increased, decreased, or maintained at their current levels?" Distributions of the three responses are shown for 1986 and 1997 surveys in Table 1 . There was a statistically significant shift away from advocating increases in benefits and toward maintaining or reducing benefits. Indeed, the percentage of respondents who supported reducing benefits almost doubled, from 3.3% to 6.3%, whereas those advocating an increase in benefits went from a majority to less than half of all respondents.        In the second part of this investigation, we focused solely on the cohort change and replacement with respect to the youngest cohort between 1990 and 1997. In this section we took a more nuanced approach to assessing attitudes by taking into account the possibility that judgments about the worthiness of the target group of a social program are made independently of judgments about its financing—what has been termed ideological schizophrenia in public policy evaluation (Day 1993; Silverstein and Parrott 1997). Thus, we considered attitudes along two basic dimensions: the perceived legitimacy and utility of programs targeted at older people, and the perceived costliness and fiscal inequity of programs serving older people. Level of agreement with the following three statements reflected attitudes toward the legitimacy and utility of programs for older people: (a) "Social Security and Medicare are an earned right for persons aged 65 and older." (b) "If the government cut back on spending on programs for older people, we would all be hurt." (c) "The government should add coverage of nursing home care and home health care to Medicare, even if it means higher federal taxes for everyone." We used the level of agreement with the next three statements to illustrate attitudes toward the perceived costliness and fiscal inequity of programs for older people: (a) "Federal programs that provide benefits to older persons are too costly." (b) "Persons aged 65 and older get more than their fair share of government programs and tax benefits." (c) "All people over age 65 should pay a larger share of their medical costs than they do today." All statements were rated on a scale from 1 to 5, where 1 = "strongly disagree" and 5 = "strongly agree."        To determine whether these six items corresponded to the two underlying dimensions of interest, we performed factor analysis with oblique rotation on these items. The solution and factor loadings from this analysis (Table 2 ) revealed that for each survey year the items scaled in a way that was consistent with our expectations: The first three items loaded on the first factor (Entitlement to Programs), and the second three items loaded on a second factor (Costliness of Programs), with an interfactor correlation of −.242. Using structural equation modeling with latent variables, we replicated this factor structure and were able to show that constraining factor loadings to be equivalent between 1990 and 1997 did not produce a significant decline in model fit, χ2(4, n = 2907) = 5.9, p = .21. This suggested that the measurement model was invariant over the two waves of measurement and permitted a meaningful analysis of change in these constructs over time. Therefore, for each of the two dimensions in each time period, we computed an average score from its three corresponding scale items.                    Independent Variables        Our goal of decomposing the components of aggregate change in the population required only that age and historical time be used as variables. To match the age ranges with the 11-year interval between 1986 and 1997, we constructed the following age groups at both periods: 18–28, 29–39, 40–50, 51–61, 62–72, and 73 and older. We recognized that each of the samples may have had different compositions over time owing to shifts in population characteristics over time (e.g., increases in education) as well as sampling variability. Firebaugh and Davis 1988 pointed out that these compositional changes may or may not be of interest depending on whether the goal of the investigation is simply to estimate the amount of change due to cohort replacement or to discover why cohort replacement promotes social change. However, differences between samples due to sampling variability may prove problematic to the degree that these differences are also related to the attitudes that are being studied. Therefore, we present our findings both unadjusted and adjusted for the following variables: gender, marital status (married vs. unmarried), race (African American vs. other), and education (having attended college vs. high school graduate or less).        As an adjunct to our analyses with respect to the entire U. S. population, we also examined intracohort and cohort replacement in attitudes toward aging policy among members of the youngest cohort of American adults, those aged 18–24 years, between 1990 and 1997. We present these results without control variables applied.                    Procedure        We used the decomposition method outlined by Firebaugh 1997 to disaggregate net (or social) change into components of intracohort change and cohort replacement. Because net change is the sum of intracohort change and cohort replacement, we calculated the size of cohort replacement as CR = NC − IC, where CR = cohort replacement, NC = net change, and IC = intracohort change. The orthogonal intracohort change and cohort replacement components can be contrasted to determine how much of the net change is due to individual change and how much is due to replacement of older cohorts with younger cohorts. It is important to note that intracohort change is not the same as an aging effect and that cohort replacement is not the same as a cohort effect because period changes are confounded with maturation and cohort replenishment. Thus, it was important for us to rely on theory and "side" information to make educated guesses as to the mechanism underlying these two components of social change (see Firebaugh 1992; Glenn 1976).                    Results      Our first step in setting up the cohort table that describes the components of change was to form a series of six adjacent cohorts. In Table 3 , we assembled separate tables showing the proportion of each age group in each year who responded with each of the following three responses concerning desired change in Social Security benefits: "increase," "maintain," and "decrease." By taking row and diagonal differences between years, we calculated, respectively, the size of net social change between adjacent cohorts and the size of intracohort change. To more formally assess the change within cohorts, we also tested whether intracohort change was statistically significant for each cohort.      In estimating the components of social change for the sample, we took unweighted averages of the columns. Firebaugh 1997 noted that cohort replacement effects can derive both from changes in the values expressed by succeeding cohorts and from the restructuring of cohort representation in the population. Because we were concerned with understanding the extent to which social change derived from the unique imprint of historical context on young cohorts and not from fluctuations in the relative size of cohorts, we followed the lead of Alwin and Scott 1996 and equated the size of cohorts by calculating averages unweighted by group size to estimate the magnitude of social change and intracohort change (and by extension, cohort replacement).      Change between 1986 and 1997 in the percentage of respondents advocating for increase in Social Security benefits is shown in the first panel of Table 3 . These results show that unadjusted net change in the population was −11.6%. This drop was further separated into intracohort and cohort replacement components. The component of net change due to intracohort change was −13.5%. Thus, individuals within cohorts were changing at a rate of change that exceeded the aggregate amount of social change because of a slightly positive cohort replacement effect of 1.9%. Essentially all the change in the population between 1986 and 1997 could be attributed to historical and developmental influences that occurred within cohorts, and virtually none of the almost 12% reduction could be attributed to cohort replacement. Adjusting for demographic variables reduced the amount of change observed but did not alter this conclusion. When we examined patterns of change across different cohorts, we found a large amount of variation in the size of the intracohort effect. Most notable was the substantial reduction—by more than 27%—among those aging from 62–72 to 73 and older in their support for increasing benefits. By contrast, young adults aged 18–28 declined only 7% in their support by the time they were aged 29–39, a reduction that was not significant at the .05 level.      We next turned to those in the population who believed that Social Security benefits should be maintained. The second panel of Table 3  reveals a net increase of 9.1% in those advocating this position, and a somewhat smaller increase of 7.2% with demographic factors controlled. The intracohort component was slightly greater than the net change at an 11.8% increase, with the cohort replacement component at −2.7%. This implies that cohorts were changing toward greater protection of benefits at a rate that outpaced aggregate social change. This occurred because, after intracohort change was taken into account, more recent cohorts were, on average, slightly less in favor of maintaining benefits. With regard to change within specific cohorts, only the youngest group, those aged 18–28, did not significantly strengthen their preference for maintaining benefits. Especially striking was the large intracohort change among those 62–72 years old in 1986 toward greater protection of benefits. Because this change (24.6%) was similar in magnitude but opposite in sign to the change observed in this cohort toward increasing benefits (−27.1%), it appears that there was a substantial shift away from advocating for the expansion of benefits toward their preservation following the transition to late old age.      Finally, the last panel in Table 3  presents results for those who were in favor of reducing Social Security benefits. Although advocacy for benefit reduction clearly represented a minority opinion, there was a net 2.6% increase in this response choice, of which the intracohort component was 1.8%. The only significant increase occurred in the 18–28 cohort (in 1986), which registered a 4.2% rise in the proportion advocating the reduction of benefits—more than double the average percentage increase across all cohorts. The cohort replacement component was 0.8%, representing one third of the aggregate social change. Thus, population turnover accounted for a relatively greater amount of social change in the populations' increased preference for cutting benefits. As before, introducing control variables served to reduce the amount of change across the two surveys but left the components of change largely proportional to each other.              Young Adults: 1990–97        Given our interest in the extent to which impressionable young adults were susceptible to growing public concern over how support to older people will be financed in coming generations, we used a more nuanced approach to analyze several dimensions of these attitudes within two cohorts of young adults across 1990 and 1997: the strength of attitudes supporting the legitimacy of entitlement programs for older people and concerns over the costs of such programs. Because only the youngest cohorts were considered, intracohort and cohort replacement effects were more straightforward than in the previous section of this investigation. An assumption of the cohort replacement model is that the attitudes of the youngest adults are not influenced by intracohort (maturation or historical) change because they are just entering the system of adult cohorts. Thus, the aggregate difference between these two similarly aged young-adult cohorts separated by 7 years was attributed solely to cohort replacement. Intracohort change was calculated as before, by comparing members of the same cohort across the two periods as they aged to the next age category.        We present in Table 4  the mean agreement scores among young adult cohorts with statements reflecting entitlement and cost dimensions of aging policy. As a basis for comparison, means are also shown for the whole population (as net aggregate change). Results are presented without adjusting for gender, education, race, and martial status because negligible differences were found in the means when these variables were controlled.        With respect to the strength of attitudes toward entitlement of older people, we noted a significant decline of one fifth of a point in support for entitlement due to cohort replacement of young adults. That is, compared with their predecessors, the more recent cohort of those aged 18–24 years weakened in their support of programs targeting older people. This change was consistent with the trend observed in the general population of more than one quarter of a point decline in support for entitlement. The intracohort change in the 18–24-year-old group was not statistically significant, implying that the passage of members of this cohort to a later stage of life did not produce a change in their attitudes toward entitlement of older people to programs and benefits.        Attitudes on the part of young adults toward the cost of programs for older people are shown in the second part of Table 4 . Neither cohort replacement nor intracohort change of young adults was statistically significant, although both effects showed a slight trend of declining concern with the cost of programs serving older people. These declines were in contrast to a significant increase of more than one tenth of a point in the population's level of anxiety over cost issues.        To test whether the effects of intracohort change and cohort replacement among young adults were significantly different from the aggregate change for the full population, we conducted regression analyses in which intracohort change and cohort replacement factors (effect coded) were interacted with a dummy variable representing the two periods of measurement. In these models interaction terms reveal whether intracohort change and cohort replacement effects were different from the average social change observed in the population as a whole. These four regressions (not shown) produced significant interaction terms for intracohort change by year, for both attitudes toward entitlement and attitudes toward cost. These results implied that the trend in attitudes of members of the 18–24 cohort (as they aged to 25–31) ran counter to the average trend in the population of increasing wariness over the legitimacy of entitlement and the associated costs. Remarkably, these patterns revealed that attitudes in the general population were converging with those of this young adult cohort along both dimensions over the 7-year period.        Despite an overall increase in the wariness expressed by the public toward the legitimacy and the cost of programs for older people, it is still noteworthy that, on average, Americans tended to be sympathetic toward programs serving older people. On the basis of the scale means, the support for entitlement was consistently stronger than the level of concern over cost. Further, when we examined individual scale items, we found that even in 1997 nearly three quarters of the population agreed that entitlement programs for older people were an "earned right" of older people and less than a third agreed that they were "too costly." Thus, any discussion of the social change in the public's attitude on aging policy should be placed in the context of generally widespread support for the role of the state in taking care of older citizens (Cook and Barrett 1992; Day 1990, Day 1993; Marmor, Mashaw, and Harvey 1990).                    Discussion      In this investigation we used three nationally representative samples surveyed over an 11-year period to examine the degree to which cohort turnover and cohort change are related to changes in public attitudes toward programs that serve older people. Contrary to expectations, we found that cohort replacement played a small role in explaining the trend through the 1980s and 1990s of increasing opposition to the expansion of Social Security benefits. Instead, being embedded within a society that, as a whole, is moving toward being less generous to older people is primarily responsible for this social change in the population. The most dramatic shift in this regard occurred among those making the transition from early to late old-age, who most sharply moved away from a preference for increasing benefits levels. Whether or not this change is the result of growing concern about retrenchment in these benefits, increasing satisfaction with current benefit levels, or strengthening altruistic motivations cannot be known from the available data. The wholesale shift of this "young old" age group toward a preference for maintaining benefits may mean that this benefit-eligible group has become more protective of their benefits at a time when these programs are considered to be at risk but may also imply increasing satisfaction with current payment levels. However, that this elderly cohort also exhibited a relatively large (though not significant) increase in the percentage advocating for the reduction of benefits may imply that benefits are viewed as overly generous by their very recipients, especially if seen as imposing a burden on younger generations.      A modest impact of cohort replacement was found with regard to the increase of those advocating reductions in benefits, much of that due to the cohort turnover among young adults aged 18–28 years. More notable is the relatively large increase in this cohort calling for a reduction in benefits—more than twice the average intracohort increase. Earlier we noted that intracohort change potentially reflects the effects of maturation (life-cycle transitions) and the effects of changing historical period. Young adults may have been particularly apt to absorb political messages of the 1980s and 1990s that framed the debate over entitlement programs as a crisis that would put future generations at risk—an example of the impressionable youth hypothesis posed by Alwin and Krosnick 1991. Even though members of this young age group started out in 1986 with greater skepticism toward public programs benefitting older people, they further increased their opposition to expanding benefits. This cohort of young adults was disproportionately swayed by the ideological shift toward questioning the legitimacy of entitlement programs for older people.      Evidence of the importance of cohort turnover emerged in our examination focusing on the attitudes of the young adult cohort toward the entitlement programs for the aged. Our results provide strong evidence that the cohort more recently emerging into adulthood is substantially more skeptical than its predecessor concerning the deservingness of older people. We found no corresponding effect with regard to attitudes toward cost. However, there were intracohort differences observed with the passage of members of this young adult cohort into the next older age category, because their attitudes toward entitlement and costs did not become more conservative as did those of the overall population. Thus, although members of this cohort started out being more skeptical about entitlement and more critical about the costs of programs for older people, their attitudes stabilized so that over time they came to more closely resemble those of the population in general. It is likely that members of this cohort have moved into work and family roles that have caused them to "age out" of their substantially more negative opinions about entitlements and costs associated with the older population and adopt values more consistent with the mainstream opinion.      Putting the results of this analysis together presents an apparent contradiction concerning the importance of cohort membership in the changing attitudes of the public toward programs targeting older people. Why does the transition of members of the youngest cohort to the next age category result in the strengthening in their preference for reducing Social Security benefits and a relative weakening in their concerns over the costs of programs for older people? The confound mentioned earlier between aging and historical change in the intracohort effect may provide a clue for solving this apparent contradiction. In addressing the issue, we first draw attention to the fact that in each analytical section the young adult cohorts reflect different age intervals and are assessed over different, if overlapping, historical periods. In the first analysis the interval began in 1986, a time when public support for Social Security was the highest level it had been in more than a decade (see Cook and Barrett 1992). If young adults are, in fact, especially sensitive to changes in sociopolitical milieu (Alwin and Krosnick 1991), then the values of this age group would be most affected by the media messages, proliferating in the decade following 1986, that questioned the viability and equity of public programs for the aged. Thus, the increased support for cuts in Social Security benefits in this cohort can more safely be attributed to historical forces than to maturation. Liberalization in the attitudes of the youngest age group toward entitlements and costs is out of step with the trend in the overall population and therefore may be the result of maturation. This interpretation is strengthened by the fact that it is in the age range subsequent to 18–24, following the transition to full adulthood, that some of the most fundamental life-style changes occur.      This research is clearly not the last word on the subject of cohort change and attitudes toward aging policy. The availability of data from multiple birth cohorts collected over more than two time periods would go far toward determining the enduring effects of population turnover and maturation on attitudes toward public programs serving older people. Further, it is possible that several political events over the period studied (including the failure of Clinton's attempt at health care reform), as well as the changing tone of messages in the media and popular press, are responsible for observed changes in attitudes. However, the relative contribution of any one event is difficult to infer without data collected over many time points during the interval. Our inferences must, therefore, be speculative concerning the ultimate causes of social change in the population on this issue.      In general our findings help resolve the question of whether attitudes of young adults toward programs for the aged are a product of youth itself (and thus will change with age) or the imprint of a unique historical period (and thus will be different among future young adults; Silverstein and Parrott 1997). The answer seems to be that both are true. The attitudes of young adults are changing as they age, suggesting that the imprint of historical conditions on impressionable youth is not indelible. On the other hand, historical forces appear to be selectively shaping the attitudes of new cohorts of young adults. The more contemporary young-adult cohort was especially likely to absorb societal messages of anxiety over public programs for the aged compared with the cohort it was replacing. Will members of the current cohort of young adults also age out of their skeptical attitudes toward benefit programs for older people, and will a new skeptical younger generation replace them? Addressing this question with multipanel data will be necessary to ultimately determine the consistency with which historical contingencies act on young adults to mold their attitudes toward public programs serving the older population.                                                  Table 1.                                 Characteristics of U.S. Samples for 1986 and 1997 Surveys                                                                              Year of Survey                                                                                                                            1986                                1997                                                                            Characteristic                n                %                n                %                p                                             Gender                                                                                .295                                            Male                518                42.8                699                44.8                                                            Female                691                57.2                860                55.2                                                            Education                                                                                <.001                                            High school graduate or less                592                49.0                611                39.2                                                            Some college or more                617                51.0                948                60.8                                                            Race/Ethnicity                                                                                .751                                            African American                111                9.2                138                8.9                                                            Hispanic                52                4.3                76                4.9                                                            White non-Hispanic and other                1,046                86.5                1,345                86.3                                                            Marital Status                                                                                <.05                                            Not currently married                515                42.6                726                46.6                                                            Currently married                694                57.4                833                53.4                                                            Age                                                                                <.001                                            18–28                269                22.3                290                18.6                                                            29–39                343                28.4                426                27.3                                                            40–50                200                16.6                373                23.9                                                            51–61                166                13.8                214                13.7                                                            62–72                144                11.9                157                10.1                                                            73+                85                7.0                99                6.4                                                            Attitude Toward Social Security Spending                                                                                <.001                                            Increase                680                56.9                717                46.9                                                            Maintain                476                39.8                717                46.9                                                            Decrease                40                3.3                96                6.3                                                            Total                1,207                100.0                1,559                100.0                                                                                          Table 2.                                 Factor Analysis With Oblique Rotation Performed on Six Items Measuring Attitudes Toward Programs for Elderly Persons: 1990 and 1997 Pooled Data                                                                              Rotated Factor Loadings                                                            Variable                Entitlement Factor                Cost Factor                                            Social Security and Medicare are earned right                 0.644                −0.045                                            Cuts in spending for older people would hurt all                 0.635                −0.055                                            The government should add coverage of long-term care                 0.701                −0.031                                            Federal programs benefitting older persons are too costly                 0.262                                  0.802                                                            Persons aged 65 and older get more than their fair share                 −0.171                                  0.644                                                            All people aged 65 and older should pay a larger share of medical costs                 −0.230                                  0.545                                                            Eigenvalue                1.797                1.039                                            % variance explained                29.9                17.3                                                                        Notes: Factor loadings above .5 are shown in bold. Interfactor correlation = −.242.                                                            Table 3.                                 Percentage of U.S. Adults Responding That Social Security Benefits Should Be Increased, Maintained, or Decreased: By Age and Year of Measurement (1986 and 1997)                                                                              Increase Benefits                                                                Maintain Benefits                                                                Decrease Benefits                                                                                            Age                1986                1997                NC                IC                1986                1997                NC                IC                1986                1997                NC                IC                                            18–28                54.7                50.9                −3.8                —                40.4                41.5                1.1                —                4.9                7.7                2.8                —                                            29–39                60.2                47.7                −12.5                −7.0*                35.7                43.2                7.5                2.8                4.1                9.1                5.0                4.2**                                            40–50                55.8                48.8                −7.0                −11.4**                40.2                45.8                5.6                10.1**                4.0                5.5                1.5                1.4                                            51–61                57.2                43.5                −13.7                −12.3**                41.0                51.2                10.2                11.0**                1.8                5.3                3.5                1.3                                            62–72                56.9                47.4                −9.5                −9.8*                42.4                51.3                8.9                10.3*                0.7                1.3                .6                −0.5                                            73+                53.1                29.8                −23.3                −27.1**                45.7                67.0                21.3                24.6**                1.2                3.2                2.0                2.5                                            Mean changea                —                —                −11.6                −13.5                —                —                9.1                11.8                —                —                2.6                1.8                                            Mean change adjusteda,b                —                —                 −9.5                −12.0                —                —                7.2                10.3                —                —                2.3                1.6                                            Cohort replacementc                1.9/2.5                                                                −2.7/−3.1                                                                .79/.70                                                                                                                        Notes: NC = net change; IC = intracohort change. Significance tests performed only for IC. See Table 1  for sample sizes.                                      a              Means unweighted by age-group size; tests of statistical significance not conducted.                                      b              Means when adjusted (within age group) by gender, education, marital status, and race.                                      c              Cohort replacement evaluated with unadjusted means/adjusted means.                                      *              p < .10; **p < .05.                                                            Table 4.                                 Attitudes Held by Young Adults Toward Entitlement and Costs of Programs Serving Older People Between 1990 and 1997                                                              Age                1990                1997                IC                CR                NC                                            Mean support for entitlement                                                                                                                            18–24                3.74                3.53                —                −0.21*                —                                            25–31                —                3.78                0.04                —                —                                            Total sample                4.02                3.75                —                —                −0.27*                                            Mean concern over cost                                                                                                                            18–24                2.44                2.40                —                −0.04                —                                            25–31                —                2.36                −0.08                —                —                                            Total sample                2.27                2.38                —                —                0.11*                                                                        Notes: IC = intracohort change; CR = cohort replacement; NC = net change as measured by total mean difference. Statements were rated between 1 and 5, with 1 = "strongly agree" and 5 = "strongly disagree." Sample sizes (1990/1997) were as follows: For age range 18–24, n = 210/165 for attitudes toward entitlement and n = 205/168 for attitudes toward cost. For age range 25–32 in 1997, n = 239 for attitudes toward entitlement and n = 241 for attitudes toward cost.                                      *              p < .05.                                                      This research was funded by grants from the AARP Andrus Foundation and the National Institute of Aging (R29-AG13237). The lead author also thanks the Fulbright Scholar Program for its support.      We thank Gordon Streib and Leonard Cain for their critique of an earlier version of this research that was presented at the 1998 Annual Meeting of The Gerontological Society of America. We are indebted to Fay Lomax Cook for generously making her data available to us. We are also grateful to Xuan Chen, Anne Marenco, Yuhai Wang, and Frances Yang for their assistance in preparing the manuscript and to three anonymous reviewers for their constructive comments and suggestions.                      Achenbaum W. A., 1986. Social Security: Visions and revisions Cambridge University Press, New York.                     Alwin D. F., Cohen R. L., Newcomb T. M., 1991. Political attitudes over the life span: The Bennington women after fifty years University of Wisconsin Press, Madison.                     Alwin D. F., Krosnick J. A., 1991. Aging, cohorts, and the stability of sociopolitical orientations over the life span. American Journal of Sociology 97:169-195.                     Alwin D. F., Scott J., 1996. Attitude change: Its measurement and interpretation using longitudinal surveys. Taylor B., Thomson K., , ed.Understanding change in social attitudes 75-106. Dartmouth, Brookfield, VT.                     Becker, G. S. (1994, March 28). Cut the greybeards a smaller slice of the pie. Business Week, 20.                    Bengtson V. L., 1993. Is the "contract across generations" changing? Effects of population aging on obligations and expectations across age groups. Bengtson V. L., Achenbaum W. A., , ed.The changing contract across generations 3-24. Aldine de Gruyter, New York.                     Bengtson V. L., Achenbaum W. A., , ed.The changing contract across generations 1993Aldine de Gruyter, New York.                     Bengtson V. L., Harootyan R. A., 1994. Generational linkages and implications for public policy. Bengtson V. L., Harootyan R. A., , ed.Intergenerational linkages: Hidden connections in American society 210-234. Springer, New York.                     Binstock R. H., 1983. The aged as scapegoat. The Gerontologist 23:136-143.                     Bishop J. M., Krause D. R., 1984. Depictions of aging and old age on Saturday morning television. The Gerontologist 24:91-94.                     Clausen J. A., 1993. American lives: Looking back at the children of the great depression Free Press, New York.                     Cook, F. L., & Barrett, E. J. (1992). Support for the American welfare state: The views of Congress and the public. New York: Columbia University Press.                    Cornman J. M., Kingson E. R., 1996. Trends, issues, perspectives, and values for the aging of the baby boom cohorts. The Gerontologist 36:15-26.                     D'Antonio M., 1993. Golden oldies: The WWII generation got it all. What's left for the rest of us?. Los Angeles Times Magazine 16–20:46-49.                     Day C. L., 1990. What older Americans think: Interest groups and aging policy Princeton University Press, Princeton, NJ.                     Day C. L., 1993. Public opinion toward the costs and benefits of Social Security and Medicare. Research on Aging 15:279-298.                     Duff, C. (1995, September 28). Profiling the aged: Fat cats or hungry victims? Wall Street Journal, pp. B1, B16.                    Elder G. H., 1994. Time, human aging, and social change: Perspectives on the life course. Social Psychology Quarterly 57:4-15.                     Firebaugh G., 1992. Where does social change come from? Estimating the relative contributions of individual change and population turnover. Population Research and Policy Review 11:1-20.                     Firebaugh G., 1997. Analyzing repeated surveys Sage, Thousand Oaks, CA.                     Firebaugh G., Davis K. E., 1988. Trends in antiblack prejudice, 1972–1984: Region and cohort effects. American Journal of Sociology 94:251-272.                     Glenn N. D., 1976. Cohort analysts' futile quest: Statistical attempts to separate age, period, and cohort effects. American Sociological Review 41:900-904.                     Jackson, R. L. (1994, December 31). Seniors' benefits unfair to young, tax group says. Los Angeles Times, p. A4.                    Kingson E. R., Williamson J. B., 1993. The generational equity debate: A progressive framing of a conservative issue. Journal of Aging and Social Policy 5:31-53.                     Kronebusch K., Schlesinger M., 1994. Intergenerational tensions and conflict: Attitudes and perceptions about social justice and age-related needs. Bengtson V. L., Harootyan R. A., , ed.Intergenerational linkages: Hidden connections in American society 112-151. Springer, New York.                     Logan J. R., Spitze G. D., 1995. Self-interest and altruism in intergenerational relations. Demography 32:353-364.                     Mannheim K., , ed.Essays on the sociology of knowledge 1952Routledge & Paul. (Original work published 1922), London.                     Marmor T. R., Mashaw J. L., Harvey P. L., 1990. America's misunderstood welfare state: Persistent myths, enduring realities Basic Books, New York.                     Marshall V., Cook F., Marshall J., 1993. Conflict over intergenerational equity: Rhetoric and reality in a comparative context. Bengtson V. L., Achenbaum W. A., , ed.The changing contract across generations 119-140. Aldine de Gruyter, New York.                     Marwell G., Aiken M. T., Demerath N. J., III. 1987. The persistence of political attitudes among 1960's civil rights activists. Public Opinion Quarterly 51:359-375.                     Myles J., 1995. Neither rights nor contracts: The new means-testing in U. S. aging policy. Generations 19: (3) 20-24.                     Ponza M., Duncan G. J., Corcoran M., Groskind F., 1988. The guns of autumn? Age differences in support for income transfers to the young and old. Public Opinion Quarterly 52:441-466.                     Rhodebeck L. A., 1993. The politics of greed? Political preferences among the elderly. The Journal of Politics 55:342-364.                     Romero, D. (1994, December 27). Fighting for what they are entitled to. Los Angeles Times, pp. E1, E4.                    Ryder N. B., 1965. The cohort as a concept in the study of social change. American Sociological Review 30:843-861.                     Samuelson, R. J. (1994, April 18–24). Generational economics: The gap is real. The Washington Post National Weekly Edition, p. Op-Ed 28.                    Schlesinger M., Kronebusch K., 1994. Intergenerational tensions and conflict: Attitudes and perceptions about social justice and age-related needs. Harootyan R., Bengtson V. L., Schlesinger M., , ed.Hidden connections: Intergenerational linkages in American society 152-184. Springer, New York.                     Shapiro R. Y., Patterson K. D., Russell J., Young J. T., 1987. The polls: Public assistance. Public Opinion Quarterly 51:120-130.                     Silverstein M., Parrott T., 1997. Attitudes toward public support of the elderly: Does early involvement with grandparents moderate generational tensions?. Research on Aging 19:108-132.                     Silverstein, M., & Parrott, T. (in press). Attitudes toward government policies that assist informal caregivers: The link between personal troubles and public issues. Research on Aging.                    Vasil L., Wass H., 1993. Portrayal of the elderly in the media: A literature review and implications for educational gerontologists. Educa-tional Gerontology 19:71-85.                     Wolfe A., 1995. The age of anxiety: A new school claims that aging is just another season. Get over it. How to think and: (t to think)