Altruism or self-interest? Social spending and the life course.

The primacy of self-interested individuals is often regarded as the appropriate basis for US social spending decisions. One thread of this argument has advanced age-based self-interest and politically powerful elderly to explain why Social Security and Medicare have thrived in a policy environment that has seen retrenchment in other programs. We argue that crude self-interest and individual programs considered in isolation are insufficient to understand social spending preferences. We use General Social Survey data to contrast conventional and critical explanations for understanding the role of age in preferences for social spending. Factor analyses demonstrate that social spending preferences cluster into conceptually distinctive domains. This supports our argument that social spending orientations are more complex than conventional analyses of age-based preferences for single-issue discrete programs like education, welfare or Social Security suggest. Overemphasis on age group differences misconstrues the role of age in spending orientations and whether preferences are more plausibly labeled as self-interested or altruistic. Considering how age, period and cohorts differences impact social spending domains improves understanding of how the life course influences social spending preferences.

Keywords: aged-politics, social policy, altruism

**********

Since the mid-1970s, most domestic federal social spending in the United States has been under siege. Critics of government's role in providing public goods and services have argued that all social spending must be retrenched. Within this antistatist political climate (Quadagno and Street 2005), social welfare programs (such as the former Aid For Families with Dependent Children, now replaced by Temporary Assistance to Needy Families) targeted toward poor children and working age citizens have been particularly hard hit by retrenchment and budget cuts (Mink 1998; Marmor, Cook and Scher 1997; Estes, Linkins, and Binney 1996). During the same period and despite claims by some that the 'welfare state for the elderly' (Myles 1989) is unaffordable (Lamm 1999; Howe 1997; Peterson 1994) cuts to broad-based programs like Social Security and Medicare have been made only at the margins. Despite recent political rhetoric urging partial privatization of those programs (Quadagno and Street 2006), spending on these two age-based programs dwarfs federal spending on all other domestic social programs combined (United States House of Representatives 2004).

A relatively straightforward way to interpret recent social spending trajectories is to adopt a conventional analytic approach that asserts self-interest based on single issue politics-like support for public education or opposition to welfare--as the foundation for spending preferences (see Campbell 2002; MacManus 1996). Certainly, individuals can identify particular programs they favor or dislike, even when they are uncertain of program details and specifics (Kuklinski, Quirk, Schweider, and Rich 1998). A more holistic approach to understanding social spending preference structures, however, would account for generalized orientations towards types of social spending rather than assessing particular social programs in isolation (Kohli 1996). In contrast to assumptions that each program elicits a specific, isolated spending preference response, we argue that orientations towards social spending are more diffuse and generalized than age-based, single-issue analyses suggest (see also Street and Cossman 2006). People do not form preferences for social spending in a programmatic vacuum, but rather express support for related types of spending in a more general way. Consequently, considering spending domains representing similar social programs will be more useful than single-issue analyses for understanding citizens' orientations towards social spending.

Our research uses two analytic devices to explore the complex interrelationships among individuals' preferences for social spending, age, and other temporal influences. First, we conduct theoretically informed factor analyses of General Social Survey (GSS) social spending items to verify hypothesized spending preference domains. Three domains--universal, targeted, and infrastructure--are conceptualized based on assumptions from critical gerontology and political economy about how "average citizens" might perceive benefits derived from each spending domain. Because factor analysis demonstrates that hypothesized social spending preferences cluster predictably, we can use the resulting domains to analyze generalized social spending preference structures rather than considering programs in isolation. This approach offers a more holistic analysis of social welfare orientations than does a program-by-program analytic approach.

Second, we use multiple administrations of the GSS to construct age groups and birth cohorts to assess potential temporal influences on orientations towards social spending. The conventional view that social spending preferences are exclusively derived from individuals' self-interested behavior is contrasted with the critical life course perspective (see Williamson, McNamara and Howling 2003 for an excellent discussion of this topic). This perspective suggests that social spending often represents broad-based interests that can be interpreted, in part, as altruistic orientations toward general welfare among citizens. Thus, we explore the empirical foundation of a potential "politics of aging" by evaluating levels of support for social spending domains across age groups, periods and cohorts. These analyses also allow us to assess the plausibility of characterizing spending preferences as 'self-interested' or 'altruistic.'

Age and Spending Preferences

Age, if expressed either as a unifying or discriminating factor, matters in the policymaking arena and may be a mobilizing influence in political contexts (Silverstein, Angelelli, and Parrott 2001; MacManus 1996; Rosenbaum and Button 1993). Proponents of the politics of aging position hold that (old) age is expressed as political self-interest that inoculates programs such as Social Security and health care against spending cuts, because they benefit the elderly directly (see Wallace et al. 1991). This model casts elderly Americans as a homogeneous, political power bloc of almost mythic proportions, winning political contests that further their own interests at the expense of other age groups (Howe 1997; Peterson 1994; Preston 1984).

Others reject this attribution of senior power as overly simplified (Williamson et al. 2003; Binstock and Quadagno 2001; Binstock and Day 1996; Street and Quadagno 1993; Wallace et al. 1991). Instead, political economists and critical gerontologists focus on historically specific institutions, policy contexts, and motivations, which differentially empower or constrain various subgroups of political actors. Such critical perspectives underscore heterogeneity among elderly Americans--indeed, among individuals in all age groups--and call into question whether there is a politics of aging where social spending preferences are concerned (Street and Cossman 2006; Williamson et al. 2003; Estes 1991). Critical perspectives emphasize instead how broad-based social spending reflects middle class incorporation (Street 1997; Street and Quadagno 1993) that garners cross-class and intergenerational political support. In contrast, narrowly targeted social assistance spending often stigmatizes beneficiaries and is vulnerable to retrenchment due to sociopolitical cleavages and political unpopularity. Furthermore, political economists and critical gerontologists argue that political identities, cleavages, and meanings are historically-specific social constructions transformed at the intersection of social, political, and economic institutions and are not especially conditioned by age or aging (Street and Cossman 2006; Street 1997; Blnstock and Day 1996; Estes et al 1996; Street and Quadagno 1993).

Thus, conventional politics of aging and critical life course perspectives offer competing worldviews for the roles age, self-interest, and altruism play in the politics of social spending, and whether spending preferences plausibly apply to single issue or more generalized spending orientations. A politics of aging model regards elderly citizens as an effective political force, self-interested enough as Social Security and Medicare beneficiaries to protect those programs and advance their own welfare, while simultaneously begrudging support for policies addressing the needs of citizens of other ages (Lamm 1999; Howe 1997; Peterson 1994). In contrast, the critical life course perspective emphasizes how broad-based programs like public pensions and public education generate political support among diverse groups of citizens. It assumes that political actors are complex, so that beyond self-interest they may also express other-regarding preferences--altruism--and be willing to share the costs of social programs, even when spending has no obvious direct benefit to them (Street and Cossman 2006; Metz 2002; Street and Ginn 2001; Walker and Naegele 1999; Street and Quadagno 1993; Minkler 1991).

Individualists and Collectivists

The contrasting views of conventional and political economy models of social spending preferences embody fundamentally different moral economy assumptions--differences in worldviews and in collectively shared moral assumptions underlying norms of reciprocity in which the American economy and political processes are grounded (Minkler and Cole 1997). The individualist or "exchange value" form of moral economy views reciprocity as appropriately bounded within kin relations and grounded in market exchange for "advantage or profit in individual transactions" (Hendricks and Leedham 1991:56), consistent with the self-interest assumptions of conventional politics of aging analyses. The individualist tradition regards general welfare derived from market outcomes and within families as almost always superior to any type of state provision. Social spending beyond targeted, subsistence poverty relief to alleviate market failures is regarded as undesirable because it risks undermining individual freedom, work effort, and family values. Rather than collective provision through welfare state programs, adherents to an individualist moral economy generally favor social welfare and public goods derived through individual thrift and voluntary arrangements. The core focus of the individualist orientation is on political actors motivated by self-interest.

Critical gerontologists and political economists regard the individualist perspective as an overly simplified framework for understanding the intersection between age and politics. This more critical life course perspective attributes political support for social programs--whether for a specific one like Social Security, or for more diffuse spending on general categories of education and health care--to much broader bases than merely age-based or self-interested ones. The collectivist moral economy is grounded in meeting human needs through social arrangements created to maximize life chances for all members of society over time, given resource constraints (Minkler and Cole 1997; Hendricks and Leedham 1991). Collectivist orientations are expressed in broad-based, universalist social policies that generate what political economists call middle class incorporation (Street 1997; Street and Quadagno 1993). Because universalist programs promote the social welfare of wide swaths of citizens and not just the poor, middle class stakeholders ensure robust political support, upholding citizens' rights to social welfare. Thus, the collectivist perspective emphasizes that both individually-motivated and shared interests, but not just self-interest alone, shape spending preferences over the life course. Within this framework, self-interest may still operate as concern for one's own welfare and a preference for policies that advance it, but in tandem with altruism (as we use the term in this research), expressed as a collective concern that others' needs be met and the willingness to support spending that advances general welfare.

Single Issues versus Social Spending Domains

Many social commentators and analysts focus on popular support for single social policies, e.g., support for spending on public education, cutting welfare spending, expanding Medicare, etc. However, Williamson et al (2003) underscore how important it is to understand both fact and framing to untangle the foundations of age-based political debates. When considering social spending priorities in isolation, the logic that age may predict support for particular types of spending seems apparent. For example, a common anecdote is that elderly voters defeat school bond and school-supporting tax initiatives in local elections (see Minkler 1991; Rosenbaum and Button 1989). Younger adults (particularly college age individuals and parents of school age children) might plausibly be expected to favor more education spending than older adults, or to withhold support for Social Security because of their youth.

Clearly, such age-based explanations offer a glimpse into conceivable reasons for individuals' preference formation for social spending (but see Street and Cossman 2006). However, such program-by-program consideration treats social spending as radically-separable single issues, whereas we argue that preference structures for spending are expressed in less divisible ways. General orientations of citizens towards social welfare, we argue, falls within distinct domains of social spending preferences. We identify three such domains based on insights from the moral economy framework in critical gerontology and informed by expectations about preferences associated with structural components of social welfare programs from political economy.

The universal domain includes broad-based types of social spending--income security, health care and education. While universal domain components might potentially provide direct benefits to individuals of different ages, or differing positions in the life course, we expect them to be conceptually linked because of their important contribution to general welfare. We hypothesize that what we call the targeted domain will encompass spending mainly perceived as benefiting individuals who most citizens regard as "other"--that is, individuals perceived as different from many members of the dominant community. Because many in the general population perceive assistance programs as disproportionately targeted toward poor and/ or minority Americans (Gilens 1996) and unlikely to benefit non-poor/majority individuals directly, we expect the targeted spending domain to reflect the sense of "otherness" that poverty-based programs typically engender. The infrastructure domain encompasses spending on communities and the built environment belonging in the public domain. Because benefits from infrastructure domain components are very diffuse and less divisible than the universal or targeted factors, this domain has no obvious age-related appeal.

Data and Methods

We use data from three waves of the General Social Survey (1988, 1994 and 2000) that represent relatively distinctive political periods. The six-year intervals allow us to create theoretically relevant age groups for analysis. Conducted by the National Opinion Research Center, the GSS samples a representative population of English-speaking, non-institutionalized individuals (Davis and Smith 1998) and is the largest sociological survey in the U.S. Data from repeated surveys allows for examination of variations in support for social spending domains by age, cohort and period. Because the discrete effects of age, cohort and period are difficult to disentangle (Glenn 2003; Alwin 2003), we caution readers that our discussion of temporal influences should be interpreted as suggestive, not causal.

Age Groups and Cohorts

Table 1 shows the composition of age groups and cohorts used in subsequent descriptive analyses. The six-year age groups at each GSS administration (1988, 1994, and 2000) were combined into cohorts when social spending preferences were measured for at least two consecutive survey periods for each group. For example, the 1959 to 1964 birth cohort includes those aged 24-29 in 1988, linked with individuals aged 30-35 in 1994, linked with individuals aged 36-41 in 2000 (the cohort highlighted on the diagonal in Table 1). Consequently, the analyses include 11 cohorts, since the youngest age group (18-23) in 2000 and the oldest age group (78+) in 1988 had no counterparts in adjacent survey administrations.

Social Spending Domains

We used nine GSS social spending items to construct three theoretically derived social spending domains. We expected three GSS items--Social Security, health care, and education, broad-based spending that directly benefits individuals--to comprise the universal domain. GSS items assessing support for social assistance and programs providing benefits mostly to perceived "others"--assistance to cities (often regarded as disproportionately benefiting urban minority populations), welfare, and assistance to blacks--were expected to cluster as the targeted domain. Finally, the infrastructure domain was expected to comprise GSS items relating to spending on parks, public transportation, and roads, which contribute to general welfare but which are not divisible into individual direct benefits.

The GSS social spending item format is to ask whether the government is spending (1) too much money, (2) too little money, or (3) just about the right amount on each type of social spending. Support for each item is measured conservatively, with only respondents answering that the government was spending "too little money" scored as supporting that item. That is, only respondents indicating insufficient spending on an item were scored as supporting it, essentially removing the effects of inertia ('just about right' responses) from the analysis. Because the risk of this approach is to understate rather than overstate support, the analytic trends shown here represent particularly robust relationships.

Constructing the Social Spending Domains

To construct the spending domains, each of the nine social spending items was dichotomized and pooled across time to increase sample size. Analyses prior to pooling show similar variance explained in each of the three GSS administrations for each of the three factors, indicating internal stability over time. Each respondent reporting that the government spent "too little money" on a particular item was categorized as supporting the item and scored 1; non-supporters reported that the government spent "about the right amount" or "too much money" and scored zero on the item. Factor analysis conducted using SPSS statistical software confirmed that the nine GSS social spending items clustered into the three conceptually distinct social spending domains--universal, infrastructure, and targeted--shown in Table 2.

Universal Domain

The universal domain reflects broad-based benefits that political economists describe as fostering middle class incorporation, creating political sustenance for this kind of social welfare spending. Because the universal domain represents large, widely available programs--education, health care, and Social Security--it benefits people from all income brackets, and not just the poor. Clustering preferences for these items result in confirmatory factor loadings from 0.65 to 0.74 and explain 48 percent of the variance in these measures.

Targeted Domain

Poverty-based programs (even programs misperceived as such) simply do not derive the same broad-based support as the more politically popular universal programs (Quadagno 1994). The targeted domain--spending on cities, assistance for blacks, and welfare--reflects spending that many regard as targeted towards individuals perceived as "other kinds" of people, different or distant from the majority culture (Gilens 1996; Quadagno 1994). Many Americans have systematic misapprehensions about such programs, either because they are misinformed about the costs, benefits and structures of programs, because they perceive most recipients to be race/ ethnic minorities, or to be "work-shy." Age does not seem to be a plausible catalyst for orientations towards the targeted domain; rather, issues of class (income) and race/ethnicity seem more likely to inform targeted domain spending preferences. Clustering the three GSS items representing the targeted domain results in strong factor loadings (from 0.47 to 0.83); the domain explains 45 percent of the variance in responses to its component items.

Infrastructure Domain

Governments spend on a variety of programs besides those that deliver direct and divisible benefits to individuals. Among programs that benefit the general welfare through social investments in public infrastructure are spending on mass transportation, parks, and roads. We expected preferences for infrastructure spending to differ from the universal and targeted domains, since benefits of the infrastructure domain are diffuse and indirect. Confirmatory factor analysis on the three GSS spending items shows that they load on one factor with scores ranging from 0.64 to 0.70. The resulting infrastructure domain explains 44 percent of the variance in the responses to GSS items asking about spending for mass transportation, parks, and roads.

Calculating Domain Factor Scores

We next created a single score for each respondent for each of the three domains. Respondents who support increased spending on all three GSS items comprising a particular domain were considered 100 percent supportive and are scored 1. Respondents supporting two of three items comprising the domain were considered 66 percent supportive and are scored 0.66 and so on. We use these scores to analyze the relationships between each of the domains and the age groups and cohorts. Because component variables load on each domain similarly for all age groups and cohorts, we are confident that the universal, infrastructure, and targeted domains represent distinctive clusters of spending preferences among GSS respondents, regardless of age group or cohort. To demonstrate how age group, period, and cohort membership influenced generalized social spending orientations, we present a simple tabular presentation of age, period, and cohort trends for the three spending domains (see Alwin 2003; Alwin and Scott 1996).

Findings

Table 3 shows the percent of GSS respondents supporting each spending domain by age group, period, and cohort. The youngest age groups report the highest support for increased universal spending, but all age groups are more supportive of universal spending than not. The majority of respondents in each age group supports universal spending, ranging from 52 to 70 percent across the groups, with a 66 percent average level of support (Table Three, Panel One). Small numbers of respondents in the over 78 age group suggest that their outlier status may be noise due to small sample size (see Table 3 Panel 1), nonetheless they report levels of support substantially below other age groups. When considering all respondents under 78 years of age, there is only a 10 percent range in the level of support for the universal domain.

Panel 2 in Table 3 shows variation in support for increased universal spending across the three periods of GSS administration. Support was at 65 percent overall in 1988, dipping slightly in 1994 when the incoming Republican Congress advocated cuts in all social programs. By 2000, support recovered to about 70 percent.

Panel 3 in Table 3 shows variation in support for universal spending across the cohorts we constructed (refer to Table 1). Cohort differences in spending preferences may occur because unique age/period interactions are expressed in the cohort construct (for example, attaining adulthood during the Depression, coming of age as part of the leading edge of the Baby Boom generation, etc.) (Elder 1994, 1985). There is a linear pattern of low levels of support for universal spending among the earliest born cohorts to higher levels of support among more recently born cohorts. This generalized preference structure of higher relative levels of support for universal domain spending among the more recently born cohorts holds whether considered as discrete programs (see Street and Cossman 2006) or as a generalized social spending domain.

Support for increased targeted and infrastructure spending are also shown by age, period and cohort in Table 3. The average levels of support for increased spending in both the targeted and infrastructure domains are similar, and are substantially lower than for the universal spending. The patterns of variation among age, period, and cohort are similar for universal and targeted spending, but with levels of support substantially lower for the targeted domain. In contrast, age, period, and cohort patterns of support are distinctively different for infrastructure spending when compared to both targeted and universal spending.

Support for targeted spending peaks among young respondents (age 24-29), remains at stable levels before dropping off around middle age (age 48-52), then dips even more steeply for respondents over age 60. Support for targeted spending declines overall in 1994, a pattern similar to the suppression of support for universal spending in the same period. By 2000, support for targeted spending rebounds to its 1988 level. Panel 3 Table 3 indicates higher levels of targeted spending support for more recently born cohorts and lower levels for earlier born cohorts, a cohort preference pattern similar to the universal spending domain.

Levels of support for infrastructure spending vary only 7 percent across age groups. Unlike the universal and targeted domains, where support was highest among younger respondents and lowest among older ones, the age pattern of preference is reversed for infrastructure spending. Individuals aged 24-29 express the least willingness to spend more (33 percent) and respondents aged 72-77 express the highest levels of infrastructure spending support (40 percent). Examination for period effects shows that infrastructure spending support was stable or rose over the entire 12-year period, unlike the midpoint dips in support for additional universal and targeted spending. With only a 5 percent difference across cohort preferences in infrastructure spending, the cohort pattern of the earlier born cohorts being more supportive of infrastructure spending than more recently born ones is a small but interesting departure from the cohort patterns observed in the other two spending domains.

Table 4 disaggregates support for universal, targeted and infrastructure spending in three ways (see Alwin and Scott [1996] for a discussion of this analysis technique). Panel 1 shows the percent supporting increased spending within each domain, at each GSS period, for each age group. Panel 2 shows change in percent supporting increased spending between GSS years. Intra-cohort change is displayed in Panel 3. In Table 4, net change by period--the difference in support for increased spending between a given age group at time "one" and the age group following it in time at time "two"--can be read across the rows of the table. Change in spending preferences among cohort members are read by following the patterns in the intra-cohort change panel, highlighted in the table.

Broadly speaking, support for increased universal spending is in the 60 to 70 percent range, while support for targeted and infrastructure spending hover around 40 percent. Domain-by-domain preferences are fairly consistent over time when all age groups are examined--total social change is generally below 5 percent for any given time period (except the universal domain which had a slightly higher 7 percent increase from 1994 to 2000). Intra-cohort change plays a larger role in evolving support for spending in each domain. This indicates that cohort effects--the combination of individual experiences and their unique intersections with history--likely condition support for additional social spending, regardless of domain.

When disaggregated by age, social change associated with support for increased universal spending is large and negative between 1988 and 1994, but counterbalanced by an even larger and positive change between 1994 and 2000. Only two age groups stand out much from the others. Net social change for universal spending is relatively large for the 30-35 age group (+11 percent), while support among respondents in the over-78 age group declined 22 percent over the twelve years. However, it is important to keep in mind that the small sample size for the oldest age group may bias these observations.

Overall, support for more targeted spending changes less than 1 percent over the twelve-year period, but is the most variable of the three domains across the age groups. For all but the youngest groups, support for targeted spending dropped-between 2 and 25 percent--from 1988 to 1994. From 1994 to 2000, most age groups' levels of support for the targeted domain rebounded, with a slight decline in overall support observed over the entire 12-year period.

There is little social change over twelve years in overall support for increased infrastructure spending. Disaggregated by age group, net social change for increased infrastructure spending trends positive, about 5 percent higher over the entire period. Generally, among younger age groups there were small decreases in support for increased infrastructure spending, despite the slight upward trend in overall support for infrastructure spending.

In terms of intracohort change, the targeted factor is most volatile, having the highest levels of change in support for increased spending, followed by infrastructure and universal domains respectively. Further, the pattern of intracohort change in support for increased infrastructure spending was quite different from either the universal or targeted domains. Intracohort decreases in support for both the targeted and universal domains appear to represent the effects of earlier born cohorts moving through the age structure.

Discussion

These analyses explore some of the complexities of the interrelationships among age groups, over time, and between cohorts for the generalized spending orientations we designated as universal, targeted and infrastructure domains. While disaggregation of the sort presented here cannot answer definitively which temporal influences--age or period, or their interaction as cohort effects--condition social spending preferences, the data are suggestive. Most evidence refutes explanations dependent on simple self-interest, single-issue orientations, or conventional politics of aging explanations for social spending preferences. More persuasive is a critical life course framework that focuses on general orientations towards related types of spending represented by three conceptually consistent domains. Such a critical framework recognizes orientations that can also be interpreted as altruistic or collective, as a partial explanation for the robustness of support for social spending preferences.

Whether each component item is considered in isolation or whether social spending domains are analyzed, we find little empirical support for a politics of aging model of spending preferences (see also Street and Cossman 2006). For such an interpretation to be convincing, self-interested orientations among the elderly towards social spending--high levels of support for programs such as Social Security and health care and low levels of support for programs of little potential benefit to them, like education--would be operative. What we find instead is consistently lower levels of support for direct social spending among the oldest GSS respondents when compared to younger ones, regardless of the specific GSS item or spending domain we considered. Support for increased spending in both the universal and targeted domains tends to peak in the middle of the age distribution and decline thereafter. This is true despite the heavier 'weight' in the universal factor of programs presumed to be of most direct benefit to elderly Americans-more spending on Social Security and health care.

An argument could be made that including education in the universal domain masks high levels of self-interested health and Social Security spending support among the elderly, tempered by presumably lower levels of support for education in the universal measure score. However, that conventional assumption is precisely opposite the empirically derived rank order of spending preferences expressed by elderly GSS respondents. They, like GSS subjects of other ages, report most support for education spending and least support for Social Security (Street and Cossman 2006) among the universal domain components. In fact, the broad similarity among all age groups' patterns of support imparts confidence that the universal domain is conceptually cohesive.

'Altruism' as we loosely use it in this context, represents a willingness to spend more on general welfare of no obvious direct benefit to individuals. In this sense, altruism does not decline monotonically with age as a politics of aging model would suggest; instead, the youngest and oldest age groups, likely having fewer economic resources, are somewhat less likely than mid-life individuals in their prime earning years (ages 24 to 59) to support increased spending in any domain. This can more plausibly be interpreted as the age-conditioned life course effect of needing to shepherd limited resources at particular life stages, rather than an age effect.

All age groups' levels of support for more universal and targeted spending appear similarly influenced by broad socioeconomic conditions and political trends, albeit from different starting points. No age groups were immune to influences in the political environments surrounding social welfare--particularly for universal and targeted spending--as support dipped during the mid-1990s when the Republican Revolution advocated less government spending, when "welfare as we know it" ended, and when the Clinton Administration health reforms failed. If Callahan (2005:703) is correct, that we are "heading back to the age of rugged individualism and survival of the fittest," we would expect when the 2006 wave of GSS data becomes available that a period effect of suppressed targeted and universal spending support shaped by the relentlessly conservative policy environment of recent years will again be observed.

Age, period, and cohort effects are notoriously difficult to untangle, because cohort effects express the unique interaction between age and period (Alwin 2003; Glenn 2003). Consequently, our results are only suggestive. Still, bivariate findings indicate that earlier born cohorts have social spending orientations less supportive, on average, of increased social spending when compared to more recently born cohorts. Such distinctive cohort orientations may occur because more recently born cohorts take for granted, or perceive they depend on more, mature welfare state programs that were less developed or available to earlier born cohorts. Earlier born cohorts' orientations may reflect their unique life experiences of limited social spending early in life, survival throughout the privations of the Depression and World War II, combined with lower expectations of costs for public provision of education, health, and pensions, expectations shaped before the maturation of the U.S. welfare state. A notable exception to the trend of earlier born cohorts being less supportive of social spending is their higher levels of support for the infrastructure domain compared to more recently born cohorts.

This finding seems paradoxical, given the earliest born cohorts' propensity for substantially lower spending preference levels in other domains. But their early life experiences may offer insight into otherwise surprisingly high levels and distinctive patterns of support for infrastructure spending. During their young lives, Depression-era public investment in massive public works projects provided jobs and enriched the fabric of public life: creating public buildings, schools, libraries, dams, bridges and highway systems. Tangible benefits of investment experienced over a lifetime, the employment infrastructure spending provided in their younger years, or a perception among the earliest born cohorts that infrastructure spending might be a better investment than direct spending on individuals--all are potential cohort-related explanations for high levels of support for the infrastructure domain. It seems unlikely that age or aging per se shaped such preferences for social spending; rather life course experiences expressed as cohort differences seem a better fit.

Observing patterns of cohort spending preferences is informative for other reasons. The peaks and valleys in average levels of support among age groups are smoothed when period and age are observed simultaneously as cohort patterns. This finding undermines a politics of aging explanation for the robustness of age-based social programs, since the age "effects" necessary to support a politics of aging model are not robust enough to thrive once there is a "control" for possible period effects that reflect broad social changes. Both the universal and targeted domains have components that appear in conventional welfare state analyses (i.e., programs offering direct benefits to individuals rather than diffuse contributions to society as a whole, such as the infrastructure domain). The earliest born cohorts are simply less supportive of direct individual welfare state benefits than are more recently born cohorts, regardless of program or domain considered. Despite their reliance on public pension and health care spending, the earliest born cohorts were least supportive of universal domain spending from which they could gain the most direct benefit. Whether because earlier born cohorts cleave to values of rugged individualism and expect individuals and families to provide for themselves, or because modest expectations of social spending have been largely met at current levels, patterns of age group and cohort preferences are at odds with a convincing politics of aging explanation to account for the robustness of programs like Social Security and Medicare.

Conclusion

Conventional political analyses assume that interest formation for age-based social programs is straightforward: Social Security and Medicare benefits create material and political reasons for elderly people to act in self-interest to maintain or increase their claim on benefits, but because education spending does not benefit them directly, they would rather not. The politics of aging--outcomes of political contests strongly influenced by elderly people--is consistent with individualist moral economy precepts. Age and self-interest in program benefits determine political identity, implying homogeneous age interests consistent with a welfare state for the elderly, largely determined by the elderly.

Our findings contradict the notion of a politics of aging. A conventional politics of aging should yield more support among Americans 65 and older than for younger age groups for the universal domain with its Social Security component. It did not. One might argue that including education and health in the universal domain 'trumps' the potential self-interested Social Security age effects that could be hypothesized for that domain. But elderly Americans' program preferences, examined on a program-by-program basis, are higher for spending on education than for Social Security (Street and Cossman, 2006). The oldest age groups report lower levels of support for all direct social spending, whether they stand to benefit disproportionately or not, including for discrete programs like Social Security. Younger GSS respondents express higher levels of Social Security support than older ones. Older people seem to embrace the ideal of American self-sufficiency--more willing than younger age groups to invest in infrastructure, but less likely to support spending that benefits individuals directly.

This does not mean that self-interest lacks a role in shaping spending preferences for elders, or citizens of any other age. Targeted benefits arise, in part, from self-interest manifested in the failure of citizens to accomplish crosscutting political alliances that express altruistic preferences for social spending to meet all citizens' varying needs over the life course. Combined with the fact that GSS respondents express higher levels of universal domain spending support with its direct, divisible individual benefits, suggests that self-interest is part of the calculus for preferences--just not the entire story. High levels of support for the universal domain, with its collective risk-sharing, show that Americans of various ages can put aside narrowly defined self-interest, banding together to support programs that share risk and rewards broadly, regarding others' financial security and wellbeing as important as their own. Older people likely base their preferences on similarly complex sets of criteria as do other adults, influenced by factors such as political socialization in childhood, socioeconomic status, party identification, the state of the economy, concerns about law and order, and affordability, to name a few. An inescapable conclusion is that interdependence among age groups--the regard grandparents have for their grandchildren, children for their parents, and individuals in general for other people--is recognized in the collectivist moral economy, forming the basis for largely age-irrelevant support of universal social programs. The universal domain creates solidarity between older and younger citizens who recognize the need for collective risk sharing in a rapidly changing political economy.

A more plausible explanation than a radically individualist perspective is that Americans consider their own needs--self-interest--alongside those of their fellow citizens and broader communities--altruism--when they decide on social spending preferences. Our findings suggest that altruism warrants further exploration as a potential motivation for social spending preferences, rather than conceding explanation to strictly economic self-interest. An intractable problem of conventional analyses of citizens' spending preferences is the implicit assumption that self-interest necessarily trumps any other motivation in individual decision-making.

Of course, we would be on shaky ground were we to argue strongly that because respondents to the GSS survey often supported increased spending unlikely to benefit them directly, they made these decisions based on altruistic motives. The fact is that while altruism may be a plausible reason for particular social spending preferences, so too is self-interest. Unfortunately, we simply lack the data to discern when either or both motivations are in play. That would require evidence of a very different sort from a three part fixed-choice response to a series of survey questions. What analysts often claim is objective evidence of self-interest (or what we might want to call altruism) may be understood quite differently by the subjects of our studies. To this end, researchers should supplement future work with more data about how people deliberate and decide which social programs they support.

Older citizens certainly maintain a high political profile. They are disproportionately likely to participate in electoral politics. They participate as interest group members to influence policy decisions for outcomes they favor. Yet insofar as the structure of social spending preferences is concerned, the evidence we have presented undermines a politics of aging hypothesis as an adequate explanation for the popularity of programs for elderly citizens. This does not mean that age interests are entirely insignificant, or that they are undifferentiated. But neither do age interests appear central to expressed preferences for social spending. A critical life course perspective recognizes that age-based political interests are historically contingent, shaped by position in the life course, conditioned by a variety of other statuses and constrained by policy legacies, political culture, and changing socioeconomic conditions. Age and self-interest, as political organizational principles in isolation, appear to offer less compelling explanations than a more holistic approach to unpacking the myriad influences on social spending preferences. We offer as a corrective the recognition that individuals are likely motivated by the combined influences of their complex lives moving through time (cohort effects) when they decide what types of social spending they prefer. Altruism should be considered as an additional factor when assessing how the politics of social spending are shaped.

References

Alwin, D. (2003). Generations, cohorts, and social change. In J.T. Mortimer and M.J. Shanahan (Eds.), Handbook of the Life Course (pp. 23-50). New York: Kluwer Academic/Plenum Publishers.

Alwin, D. and J. (1996). Understanding change in social attitudes." In B. Taylor and K Thomson (eds.), Understanding Change in Social Attitudes (pp. 75-106). Aldershot, England: Dartmouth.

Binstock, R. and J. Quadagno. (2001). Aging and politics, In R.H. Binstock and L.K. George (eds.), The Handbook of Aging and the Social Sciences, 5th edition (pp.333-351). San Diego: Academic Press.

Binstock, R.H. and C.L. Day. (1996). Aging and politics. In R.H. Binstock and L.K. George (eds.), The Handbook of Aging and the Social Sciences, 5th edition (pp. 362-82). San Diego: Academic Press.

Callahan, J. (2005). What's new about the new politics of aging? The Gerontologist 45(5):700-704.

Campbell, A. (2002). Self-Interest, Social Security and the distinctive participation patterns of senior citizens. American Political Science Review 96(3):565-74.

Davis, J.A. and T.W. Smith. (1998). General social surveys, 1972-1996: Cumulative File, [Computer file]. 2nd ICPSR version. Chicago: National Opinion Research Center.

Elder, G. (1985). Life course dynamics: Trajectories and transitions, 19681980. Ithaca NY: Cornell University Press.

Elder, G. (1994). Time, human agency and social change: Perspectives on the life course." Social Psychology Quarterly. 57:4-15.

Estes, C. (1991). The new political economy of aging: Introduction and critique." In M. Minkler and C. Estes (eds.), Critical Perspectives on Aging: The Political and Moral Economy of Growing Old (pp 19-36). Amityville, NY: Baywood.

Estes, C., K. Linkins, and E. Binney. (1996). The political economy of aging." In R.H. Binstock and L.K. George (eds.), The handbook of aging and the social sciences, 5th edition (pp. 346-59). San Diego: Academic Press.

Gilens, M. (1996). Race and poverty in America: Public misperceptions and the American news media." Public Opinion Quarterly. 60:515-541.

Glenn, N.D. (2003). Distinguishing age, period, and cohort effects. In J.T. Mortimer and M.J. Shanahan (Eds.), Handbook of the life course (pp. 465-76). New York: Kluwer Academic/Plenum Publishers.

Hendricks, J. and C.A. Leedham. (1991). Dependency or empowerment? Toward a moral and political economy of aging. In M. Minkler and C. Estes (eds.), Critical perspectives on aging: The political and moral economy of growing old (pp 51-66). Amityville, NY: Baywood.

Howe, Neil. (1997). Why the graying of the welfare state threatens to flatten the American dream--or worse. In R.B. Hudson (ed.) The Future of Age-Based Social Policy (pp. 36-45). Baltimore: Johns Hopkins University Press.

Kohli, M. (1996). The problem of generations: Family, economy, politics, Collegium Budapest: Public Lecture Series.

Kuklinski, J.H., P.J. Quirk, D.W. Schwieder, and R.F. Rich. (1998). Just the facts, ma'am: Political facts and public opinion, The Annals of the American Academy of Political and Social Science. 560: 143-154.

Lamm, R. (1999). Care for the elderly: What about our children? In J. Williamson and E. Kingson (eds.) The Generational Equity Debate (pp. 87-100). New York: Columbia University Press.

Marmor, T.R., F.L. Cook and S. Scher. (1997). "Social Security politics and the conflict between generations: Are we asking the right questions?" In E.R. Kingson and J.H. Schulz (eds.) Social Security in the 21st Century (pp. 195-207). New York: Oxford University Press.

MacManus, S.A. (1996). Generational combat: Young v. old: Generational combat in the 21st century. Boulder CO: Westview Press.

Metz, D. (2002). The politics of population ageing. The political quarterly. 73: 321-327.

Minkler, M. (1991). 'Generational equity' and the new victim blaming. In M. Minkler and C. Estes (eds.), Critical perspectives in aging (pp. 67-79). Amityville, NY: Baywood Press.

Minkler, M. and T. Cole. (1997). Political and moral economy: Getting to know one another." In M. Minkler and C. Estes (eds.), Critical perspectives on aging: The political and moral economy of growing old (pp. 37-44). Amityville, NY: Baywood.

Mink, G. (1998). Welfare's end. Ithaca NY: Cornell University Press.

Myles, J. (1989). Old age in the welfare state: The political economy of public pensions. (rev. ed.). Lawrence KS: University Press of Kansas.

Peterson, P. (1994). Facing up: How to rescue the economy from crushing debt and restore the American dream. New York: Simon and Schuster.

Preston, S.H. (1984). Children and the elderly in the U.S. Scientific American. 251(6):44-49.

Quadagno, J. (1994). The color of welfare: How racism undermined the war on poverty. Oxford University Press: New York.

Quadagno, J. (1989). Generational equity and the politics of the welfare state. Politics & Society. 17:353-376.

Quadagno J. and D. Street. (2005). Anti-statism in American welfare state development. Journal of Policy History. 17(1): 52-71.

Quadagno J. and D. Street. (2006). Recent Trends in U.S. Social Welfare Policy: Minor Retrenchment or Major Transformation? Research on Aging 28(3): 1-14.

Rosenbaum, W.A. and J.W. Button. (1993). The unquiet future of intergenerational politics. The Gerontologist. 33:481-90.

Rosenbaum, W.A. and J.W. Button. (1989). Is there a gray peril? Retirement politics in Florida. The Gerontologist. 29, 300-306.

Silverstein, M., J.J. Angelelli, and T. Parrott. (2001). Changing attitudes toward aging policy in the United States during the 1980s and 1990s: A cohort analysis." Journals of Gerontology. 56B(1): S36-43.

Street, D. (1997). Special interests or citizens' rights? 'Senior power,' Social Security and Medicare. International Journal of Health Services. 17(4): 727-751.

Street D. and J. Cossman. (2006). Greedy geezers or greatest generation? Social spending preferences and the elderly. Social Problems. 53(1):75-96.

Street, D. and J. Ginn. (2001). The demographic debate: the gendered political economy of pensions." In J. Ginn, D. Street and S. Arber. (eds.), Women, Work and Pensions: International Issues and Prospects (pp. 31-43). Buckingham: Open University Press.

Street, D. and J. Quadagno. (1993). The state, the elderly, and the intergenerational contract: Toward a new political economy of aging." In K.W. Schaie and W.A. Achenbaum (eds.) Societal Impact on Aging: Historical Perspectives. New York: Springer Publishing Co.

United States House of Representatives. 2004. Green Book. Appendix I. Budget Tables. http://www.gpoaccess.gov/wmprints/green/index.html. Accessed on September 30, 2005. Committee on Ways and Means.

Wallace, S. P., J. B. Williamson, R. G. Lung, and L. A. Powell. (1991). "A lamb in wolf's clothing? The reality of senior power and social policy." In M. Minkler and C. Estes (eds.), Critical Perspectives on Aging: The Political and Moral Economy of Growing Old (pp. 95-117). Amityville, NY: Baywood.

Walker, A. and G. Naegele (eds.). (1999). The Politics of Old Age in Europe. Buckingham: Open University Press.

Williamson, J. B., T. K. McNamara and S. A. Howling. (2003). Generational equity, generational interdependence, and the framing of the debate over Social Security reform." Journal of Sociology and Social Welfare. 23(3):3-14.

DEBRA STREET

Department of Sociology

State University of New York at Buffalo

JERALYNN SITTIG COSSMAN

Department of Sociology, Anthropology and Social Work

Mississippi State University

Table 1.

Sample Size and Cohort Construction
by Age and Year of GSS Interview

        1988   1994   2000

78+             156    135
72-77     85    162    150
66-71    116    180    166
60-65    119    175    206
54-59     85    226    312
48-53     92    311    391
42-47    157    372    387
36-41    193    415    356
30-35    195    468    311
24-29    215    349    215
18-23    141    172

Table 2. Factor Analysis with Principle Components Analysis
Performed on Nine Items Measuring Attitudes Toward Social
Spending in 1988, 1994 and 2000 Pooled Data from the General
Social Survey.

                          Universal   Targeted   Infrastructure

Education                   0.65

Health and Health Care      0.74

Social Security             0.69

Race and Race Relations                 0.83

Cities                                  0.66

Welfare                                 0.47

Roads                                                 0.65

Mass Transportation                                   0.70

Parks                                                 0.64

Eigenvalue                   1.45       1.33          1.31

% of Variance Explained     48.21      44.63         43.77

From the General Social Survey (1988-2000), the questions were
worded as follows: "We are faced with many problems in this country,
none of which can be solved easily or inexpensively. I'm going to
name some of these problems, and for each one I'd like you to tell me
whether you think we're spending too much money on it, too little
money, or about the right amount. Are we spending too much money,
too little money, or the right amount on ..."

Table 3. Percent Support for Three Social Spending Domains by
Age, Period and Cohort

Panel One
Age Group                 Universal   Targeted   Infrastructure

18-23                       62.3        42.5          34.0
24-29                       68.8        43.8          33.0
30-35                       70.0        40.1          35.2
36-41                       68.9        39.2          36.3
42-47                       69.4        39.2          36.2
48-53                       66.7        40.2          38.7
54-59                       66.9        36.9          37.2
60-65                       60.4        30.8          34.5
66-71                       59.2        31.8          37.8
72-77                       59.2        32.6          40.2
78+                         52.3        31.5          39.5

Panel Two
GSS Year                  Universal   Targeted   Infrastructure

1988                        64.7        40.2          33.2
1994                        62.7        36.7          36.6
2000                        69.6        39.4          37.2

Panel Three
Cohort                    Universal   Targeted   Infrastructure

1911-1916                   52.0        32.5          38.3
1917-1922                   56.1        31.5          38.3
1923-1928                   60.5        32.8          38.7
1929-1934                   59.6        33.3          35.9
1935-1940                   64.6        35.3          35.9
1941-1946                   67.2        36.5          38.6
1947-1952                   69.8        40.2          36.1
1953-1958                   68.2        39.9          36.8
1959-1964                   69.6        40.5          35.2
1965-1970                   68.7        42.7          33.7
1971-1976                   67.5        43.4          33.9
Total                       65.8        38.4          36.1

% of Variance Explained     47.8        43.8          43.3

From the General Social Survey (1988-2000), see page 82 for wording.

Table 4. Percentage supporting increased spending in each social
spending domain by age group and year.

Age     Universal Spending Domain        Social Change

        1988.0   1994.0   2000.0    1988-   1994-   1988-
                                     1994    2000    2000

18-23    66.0     67.4     61.3       1.4    -6.2    -4.8
24-29    62.9     65.6     70.2       2.6     4.6     7.3
30-35    65.3     67.0     76.3       1.6     9.3    10.9
36-41    63.5     68.5     69.3       5.0     0.8     5.8
42-47    63.8     63.8     69.8       0.0     6.0     6.1
48-53    63.9     58.2     65.5      -5.7     7.3     1.6
54-59    68.0     63.2     70.9      -4.8     7.7     2.9
60-65    61.0     49.8     65.5     -11.2    15.7     4.5
66-71    66.7     57.3     63.5      -9.4     6.2    -3.2
72-77    65.9     55.4     66.7     -10.5    11.2     0.7
 78+     70.6     53.7     48.4     -17.0    -5.2   -22.2
Total    64.7     62.7     69.6      -2.0     6.9     4.9

        Targeted Spending Domain

18-23    41.2     45.7     36.2      4.50    -9.5    -5.0
24-29    38.5     34.5     37.2      -4.0     2.7    -1.3
30-35    37.6     36.1     41.4      -1.6     5.3     3.8
36-41    39.3     33.3     37.6      -6.0     4.3    -1.7
42-47    39.4     35.4     38.9      -4.0     3.5    -0.5
48-53    41.4     28.9     35.7     -12.5     6.8    -5.6
54-59    45.6     29.6     32.7     -16.1     3.2   -12.9
60-65    40.8     33.7     23.9      -7.1    -9.8   -16.9
66-71    35.5     31.8     32.7      -3.6     0.8    -2.8
72-77    55.3     30.4     28.6     -24.9    -1.8   -26.7
 78+     39.0     34.3     33.3      -4.7    -1.0    -5.6
Total    40.2     36.7     39.4      -3.5    2.70    -0.8

        Infrastructure Spending Domain

18-23    37.7     36.3     33.3      -1.4    -3.0    -4.3
24-29    30.8     33.5     33.3       2.7    -0.2     2.5
30-35    32.5     38.3     29.3       5.8    -9.1    -3.3
36-41    32.2     34.6     35.4       2.4     0.8     3.2
42-47    32.9     36.9     38.8       4.0     1.9     5.9
48-53    33.1     40.4     35.4       7.3    -5.0     2.3
54-59    35.6     38.3     41.7       2.7     3.4     6.1
60-65    29.0     32.6     34.0       3.6     1.4     5.0
66-71    32.7     30.8     38.2      -1.8     7.3     5.5
72-77    47.4     33.3     38.4     -14.1     5.1    -9.0
 78+     31.1     38.2     39.7       7.0     1.5     8.5
Total    33.2     36.6     37.2       3.4     0.7     4.0

Age        Intracohort Change

        1988-    1994-    1988-
         1994     2000     2000

18-23
24-29    -0.4      2.8
30-35     4.0     10.7
36-41     3.2      2.3     10.3
42-47     0.3      1.3      6.3
48-53    -5.6      1.7      4.5
54-59    -0.7     12.7      2.0
60-65   -18.2      2.3      7.1
66-71    -3.8     13.7      1.6
72-77   -11.2      9.4     -4.5
 78+    -12.3     -7.0      5.6
Total

        Targeted Spending Domain

18-23
24-29    -6.7     -8.5
30-35    -2.4      6.9
36-41    -4.3      1.5      0.2
42-47    -3.9      5.6     -0.9
48-53   -10.5      0.3      1.3
54-59   -11.8      3.8     -3.6
60-65   -11.9     -5.6     -6.7
66-71    -9.0     -1.1    -17.4
72-77    -5.1     -3.3    -13.0
 78+    -20.9     3.00    -12.2
Total

        Infrastructure Spending Domain

18-23
24-29    -4.2     -3.0
30-35     7.5     -4.3
36-41     2.1     -2.9     -8.4
42-47     4.7      4.3      4.6
48-53     7.5     -1.5      6.3
54-59     5.2      1.2      3.3
60-65    -3.0     -4.2      8.7
66-71     1.8      5.6      0.9
72-77     0.7      7.6      2.6
 78+     -9.2      6.3      9.4
Total

Social Change = net change (percent difference)
Intracohort Change = net change (percent difference)

Source: General Social Survey 1988-2000