No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
A Rolling Panel Model of Cohort, Period, and Aging Effects for the Analysis of the General Social Survey
Abstract
The linear dependence of age, period, and birth cohort is a challenge for the analysis of social change. With either repeated cross-sectional data or conventional panel data, raw change cannot be decomposed into over-time differences that are attributable to the effects of common experiences of alternative birth cohorts, features of the periods under observation, and the cumulation of lifecourse aging. This article proposes a rolling panel model for cohort, period, and aging effects, suggested by and tuned to the treble panel data collected for the General Social Survey from 2006 through 2014. While the model does not offer a general solution for the identification of the classical age-period-cohort accounting model, it yields warranted interpretations under plausible assumptions that are reasonable for many outcomes of interest. In particular, if aging effects can be assumed to be invariant over the course of an observation interval, and if separate panel samples of the full age distribution overlap within the same observation interval, then period and aging effects can be parameterized and interpreted separately, adjusted for cohort differences that pulse through the same observation interval. The estimated cohort effects during the observation interval are then interpretable as effects during the observation interval of entangled period and cumulated aging differences from before the observation interval.
... In other words, by design, the APC-I method does not estimate the kind of linear or nonlinear cohort effects in traditional APC models because the latter's assumption that cohort effects can occur independently and additively of age and period effects lacks theoretical grounding and is thus arbitrary and questionable. 2 Our questioning of the validity of the accounting framework is not new (Hobcraft et al., 1982;Holford, 1983) and has been echoed in recent methodological work (see, e.g., Morgan, 2022;Morgan & Lee, 2021;Neil & Sampson, 2021). ...
... To conclude, although we have argued the APC-I model better represents the conceptualization of cohort effects than previous studies, we do not claim that the APC-I method offers a final answer to cohort analysis. Like other scholars (Morgan, 2022;Morgan & Lee, 2021;Neil & Sampson, 2021), we urge demographers and social scientists to assess the validity of the traditional APC accounting model as the starting point of their analysis. Ultimately, we echo the recent sentiment that every researcher should be able to defend the correspondence between the estimand in any model and the conceptual quantity that they intend to investigate (Lundberg et al., 2021). ...
Unlabelled:
For a quarter of a century researchers have been documenting and trying to explain trends in Americans' vocabulary knowledge using data from the General Social Survey (GSS) and its WORDSUM test. Trends in Americans' vocabulary knowledge have important practical implications-for example, for educational policy and practice-and speak to the American workforce's competitiveness in the global knowledge economy. We contribute to this debate by analyzing 1978-2018 GSS data using an improved analytical approach that is consistent with theoretical notions of cohort effects and that permits simultaneously estimating inter-cohort average differences and intra-cohort life-course changes. We find that WORDSUM scores peak around age 35 and gradually decline in older ages; the scores were significantly lower in the 1980s and higher in the late 2000s and 2010s; and the 1940-1954 birth cohorts and the 1965 and later birth cohorts had notably higher and lower scores, respectively, than the expectation based on age and period main effects. We provide new evidence that such cohort differences tend to persist over the life course. Interestingly, the effects of increasing educational attainment and decreasing reading behaviors seemed to cancel out, leading to a relatively flat overall period trend. Trends in television viewing and word obsolescence did not appear to affect age, period, or cohort trends in WORDSUM scores.
Supplementary information:
The online version contains supplementary material available at 10.1007/s11113-023-09771-5.
... It implies that the identification challenge is inherent in any APC model that attempts to separate independent and additive effects of age, period, and cohort and thus cannot be solved by changing the model setup (e.g., using random effects for period and cohort as in Yang and Land, 2006; see Luo and Hodges, 2020b, for a critique) or by variable manipulation (e.g., using unequal interval widths for age, period, and cohort groups as in Robertson and Boyle, 1986;Sarma et al., 2012;see Luo et al., 2016, for a detailed discussion). The identification problem is well recognized, and its consequences have been discussed extensively (Fienberg and Mason, 1985;Fosse and Winship, 2019;Kupper et al., 1983Kupper et al., , 1985Luo and Hodges, 2020b;te Grotenhuis et al., 2016;O'Brien, 2020;Morgan and Lee, 2021;Luo, 2013). In essence, internal information derived from the data cannot help because the problem is circular: researchers do the analysis to learn precisely the kind of information needed to justify any such constraint. ...
Social scientists have frequently attempted to assess the relative contribution of age, period, and cohort variables to the overall trend in an outcome. We develop an R package APCI (and Stata command apci) to implement the age-period-cohort-interaction (APC-I) model for estimating and testing age, period, and cohort patterns in various types of outcomes for pooled cross-sectional data and multi-cohort panel data. Package APCI also provides a set of functions for visualizing the data and modeling results. We demonstrate the usage of package APCI with empirical data from the Current Population Survey. We show that package APCI provides useful visualization and analytical tools for understanding age, period, and cohort trends in various types of outcomes.
This article uses the example of voter turnout in US presidential elections to compare two new methods for age, period, and cohort (APC) analysis: the APC interaction model and the APC bounding analysis. While discussing the formal, conceptual, and interpretive differences between the two methods, the analysis demonstrates how both methods can be used to generate distinct but complementary findings. Because the two methods take alternative positions on the appropriate cohort-effect estimands, the comparison underscores the importance of well-grounded conceptual foundations in APC analysis.
The five decades of results produced by analysts of the General Social Survey (GSS) have enriched our understanding of social change, but some core modeling challenges remain. This article proposes that we more fully engage in the development of targeted models of period-based attitude and opinion change, using counterfactual reasoning, as we continue to model cohort replacement. This shift is also consistent with the recent literature on age, period, and cohort analysis, which argues for attention to age varying period effects. Two outcomes are modeled to provide material for the argument: support for government spending on drug addiction and rehabilitation and the valuation of obedience as a goal for child behavioral development.
The impact of a major historical event on child and youth development has been of great interest in the study of the life course. This study is focused on assessing the causal effect of the Great Recession on youth disconnection from school and work. Building on the insights offered by the age-period-cohort research, econometric methods, and developmental psychology, we innovatively develop a causal inference strategy that takes advantage of the multiple successive birth cohorts in the National Longitudinal Study of Youth 1997. The causal effect of the Great Recession is defined in terms of counterfactual developmental trajectories and can be identified under the assumption of short-term stable differences between the birth cohorts in the absence of the Great Recession. A meta-analysis aggregates the estimated effects over six between-cohort comparisons. Furthermore, we conduct a sensitivity analysis to assess the potential consequences if the identification assumption is violated. The findings contribute new evidence on how precipitous and pervasive economic hardship may disrupt youth development by gender and class of origin.
Random effects (RE) models have been widely used to study the contextual effects of structures such as neighborhoods or schools. The RE approach has recently been applied to age-period-cohort (APC) models that are unidentified because the predictors are exactly linearly dependent. However, research has not fully explained how the RE specification identifies these otherwise unidentified APC models. We address this challenge by first making explicit that RE-APC models have greater—not less—rank deficiency than the traditional fixed-effects model, followed by two empirical examples. We then provide intuition and a mathematical proof to explain that for APC models with one RE, treating one effect as an RE is equivalent to constraining the estimates of that effect’s linear component and the random intercept to be zero. For APC models with two REs, the effective constraints implied by the model depend on the true (i.e., in the data-generating mechanism) nonlinear components of the effects that are modeled as REs, so that the estimated linear components of the REs are determined by the true nonlinear components of those effects. In conclusion, RE-APC models impose arbitrary although highly obscure constraints and thus do not differ qualitatively from other constrained APC estimators.
Social scientists have frequently sought to understand the distinct effects of age, period, and cohort, but disaggregation of the three dimensions is difficult because cohort = period − age. We argue that this technical difficulty reflects a disconnection between how the cohort effect is conceptualized and how it is modeled in the traditional age-period-cohort framework. We propose a new method, called the age-period-cohort-interaction (APC-I) model, that is qualitatively different from previous methods in that it represents Ryder’s theoretical account about the conditions under which cohort differentiation may arise. This APC-I model does not require problematic statistical assumptions and the interpretation is straightforward. It quantifies inter-cohort deviations from the age and period main effects and also permits hypothesis testing about intra-cohort life-course dynamics. We demonstrate how this new model can be used to examine age, period, and cohort patterns in women’s labor force participation.
Since the late 1990s public opinion about cannabis legalization has become drastically more liberal, and some states have begun to legalize cannabis for recreational use. Why have attitudes changed so much? Prior research has considered a few of the reasons for this change, but this is the first comprehensive and empirically-based study to consider the wide range of potential causes for how and why this happened. We use data from the General Social Survey, National Study of Drug Use and Health, and word searches from the New York Times. We find that attitudes largely liberalized via intracohort changes. Most Americans developed more liberal views, regardless of their race and ethnicity, gender, education, religious or political affiliation, or religious engagement. Changes in cannabis use have had minimal effects on attitudes, and legalization of cannabis has not prompted attitude change in neighboring states. As to root causes, evidence suggests that a decrease in religious affiliation, a decline in punitiveness, and a shift in media framing all contributed to changing attitudes.
The intrinsic estimator (IE) has become a widely used tool for the analysis of age–period–cohort (APC) data in sociology, demography, and other fields. However, it has been recently recognized that the IE is a subtype of a larger class of estimators based on the Moore–Penrose generalized inverse (MP estimators) and that different estimators can lead to radically divergent estimates of the true, unknown APC effects. To clarify the differences and similarities of MP estimators, we introduce a canonical form of the linear constraints imposed on the true temporal effects. Using this canonical form, we compare the IE to related MP estimators, examining the conditions under which they recover the true temporal effects, the impact of the size and sign of nonlinearities on the estimated linear effects, and their sensitivity to the number of age, period, and cohort categories. We show that two MP estimators, which we call the difference estimator (DE) and the orthogonal estimator (OE), impose constraints that are both less sensitive and easier to interpret than those of the IE. We conclude with practical guidelines for researchers interested in using MP estimators to estimate temporal effects.
We cast fresh light on how and why Americans’ views on marijuana legalization shifted between 1973 and 2014. Results from age-period-cohort models show a strong negative effect of age and relatively high levels of support for legalization among baby boom cohorts. Despite the baby boom effect, the large increase in support for marijuana legalization is predominantly a broad, period-based change in the population. Additional analyses demonstrate that differences in support for legalization by education, region, and religion decline, that differences by political party increase, and that differences between whites and African Americans reverse direction. We conclude by discussing the implications of these findings and by identifying promising directions for future research on this topic.
Most public opinion attitudes in the United States are reasonably stable over time. Using data from the General Social Survey and the American National Election Studies, I quantify typical change rates across all attitudes. I quantify the extent to which change in same-sex marriage approval (and liberalization in attitudes toward gay rights in general) are among a small set of rapid changing outliers in surveyed public opinions. No measured public opinion attitude in the United States has changed more and more quickly than same-sex marriage. I use survey data from Newsweek to illustrate the rapid increase in the 1980s and 1990s in Americans who had friends or family who they knew to be gay or lesbian and demonstrate how contact with out-of-the-closet gays and lesbians was influential. I discuss several potential historical and social movement theory explanations for the rapid liberalization of attitudes toward gay rights in the United States, including the surprising influence of Bill Clinton’s 1992 presidential campaign.
Age-Period-Cohort Analysis: New Models, Methods, and Empirical Applications is based on a decade of the authors’ collaborative work in age-period-cohort (APC) analysis. Within a single, consistent HAPC-GLMM statistical modeling framework, the authors synthesize APC models and methods for three research designs: age-by-time period tables of population rates or proportions, repeated cross-section sample surveys, and accelerated longitudinal panel studies. The authors show how the empirical application of the models to various problems leads to many fascinating findings on how outcome variables develop along the age, period, and cohort dimensions. The book makes two essential contributions to quantitative studies of time-related change. Through the introduction of the GLMM framework, it shows how innovative estimation methods and new model specifications can be used to tackle the "model identification problem" that has hampered the development and empirical application of APC analysis. The book also addresses the major criticism against APC analysis by explaining the use of new models within the GLMM framework to uncover mechanisms underlying age patterns and temporal trends. Encompassing both methodological expositions and empirical studies, this book explores the ways in which statistical models, methods, and research designs can be used to open new possibilities for APC analysis. It compares new and existing models and methods and provides useful guidelines on how to conduct APC analysis. For empirical illustrations, the text incorporates examples from a variety of disciplines, such as sociology, demography, and epidemiology. Along with details on empirical analyses, software and programs to estimate the models are available on the book’s web page.
Marijuana and same-sex marriage are two of the fastest changing and most widely debated opinion and policy issues in the United States. Research has examined public opinion on marijuana legalization and same-sex marriage legalization individually, but has neglected to examine these two issues together. We use General Social Survey data from 1988 to 2014 to compare four groups: (1) those who support neither; (2) those who support marijuana but not same-sex marriage legalization; (3) those who support marriage but not marijuana legalization; and (4) those who support both. This study provides four key findings: (1) marijuana and same-sex marriage attitudes have changed simultaneously; (2) most people hold these attitudes in tandem, and there has been a precipitous decline in the percentage of people who support legalizing neither and a remarkable increase in the percentage who support legalizing both; (3) attitudes toward both issues are liberalizing across all social and ideological groups, suggesting a society-wide redefinition of both behaviors as publicly accepted issues of individual autonomy; and (4) the support bases for marijuana and marriage legalization vary systematically by sociodemographic characteristics. We conclude that notions of individual autonomy may be increasingly important to the American public and their beliefs about what the government should regulate. Landon Schnabel is a doctoral candidate in the
We used standard and multilevel models to assess the reliability of core items in the General Social Survey panel studies spanning 2006 to 2014. Most of the 293 core items scored well on the measure of reliability: 62 items (21 percent) had reliability measures greater than 0.85; another 71 (24 percent) had reliability measures between 0.70 and 0.85. Objective items, especially facts about demography and religion, were generally more reliable than subjective items. The economic recession of 2007–2009, the slow recovery afterward, and the election of Barack Obama in 2008 altered the social context in ways that may look like unreliability of items. For example, unemployment status, hours worked, and weeks worked have lower reliability than most work-related items, reflecting the consequences of the recession on the facts of peoples lives. Items regarding racial and gender discrimination and racial stereotypes scored as particularly unreliable, accounting for most of the 15 items with reliability coefficients less than 0.40. Our results allow scholars to more easily take measurement reliability into consideration in their own research, while also highlighting the limitations of these approaches.
It is claimed the hierarchical-age–period–cohort (HAPC) model solves the age–period–cohort (APC) identification problem. However, this is debateable; simulations show situations where the model produces incorrect results, countered by proponents of the model arguing those simulations are not relevant to real-life scenarios. This paper moves beyond questioning whether the HAPC model works, to why it produces the results it does. We argue HAPC estimates are the result not of the distinctive substantive APC processes occurring in the dataset, but are primarily an artefact of the data structure—that is, the way the data has been collected. Were the data collected differently, the results produced would be different. This is illustrated both with simulations and real data, the latter by taking a variety of samples from the National Health Interview Survey (NHIS) data used by Reither et al. (Soc Sci Med 69(10):1439–1448, 2009) in their HAPC study of obesity. When a sample based on a small range of cohorts is taken, such that the period range is much greater than the cohort range, the results produced are very different to those produced when cohort groups span a much wider range than periods, as is structurally the case with repeated cross-sectional data. The paper also addresses the latest defence of the HAPC model by its proponents (Reither et al. in Soc Sci Med 145:125–128, 2015a). The results lend further support to the view that the HAPC model is not able to accurately discern APC effects, and should be used with caution when there appear to be period or cohort near-linear trends.
Electronic supplementary material
The online version of this article (doi:10.1007/s11135-017-0488-5) contains supplementary material, which is available to authorized users.
Through an analysis of the 2004 through 2014 General Social Survey (GSS), this article demonstrates that the 2010 passage of the Affordable Care Act (ACA) decreased support for spending on health among Democrats, Independents, and Republicans, contrary to the conjecture that a rigid partisanship equilibrium has taken hold among voters in the United States. Instead, only a partisan deflection is present, with spending preferences declining more for Republicans than for Democrats, and with Independents in between. Through supplemental analysis of the GSS panel data, as well as comparative analysis of other GSS items on national spending preferences, government responsibility, and confidence in leaders, this article also undermines support for an alternative explanation that cannot be entirely eliminated from plausibility, which is that the identified period effect that emerged in 2010 and persisted through 2014 is a response to the Great Recession and resulting deficit spending by the federal government. Implications for public opinion research are discussed, lending support to current models of thermostat effects and policy-specific political mood from the political science literature, which are informed by an older literature on weather fronts in public opinion that originated in the sociology literature.
In many different fields, social scientists desire to understand temporal variation associated with age, time period, and cohort membership. Among methods proposed to address the identification problem in age-period-cohort analysis, the intrinsic estimator (IE) is reputed to impose few assumptions and to yield good estimates of the independent effects of age, period, and cohort groups. This article assesses the validity and application scope of IE theoretically and illustrates its properties with simulations. It shows that IE implicitly assumes a constraint on the linear age, period, and cohort effects. This constraint not only depends on the number of age, period, and cohort categories but also has nontrivial implications for estimation. Because this assumption is extremely difficult, if not impossible, to verify in empirical research, IE cannot and should not be used to estimate age, period, and cohort effects.
In this article we address the topic of increasing partisan polarization in the American mass public, focusing on the twin influences of individual-level development and cohort replacement and the interaction between the two. We posit a model of individual development that consists of declining openness to change beyond young adulthood, an increase in party-issue constraint as age advances, and cohort-specific responsiveness to changes in the partisan environment. Results from a long-term panel study provide initial evidence of these dynamics. We then use simulations to generate expectations about how these developmental processes play out across cohorts, issues, and time. These expectations are evaluated through a cohort analysis of National Election Studies data from 1972 to 2004. Overall, our results provide a new perspective on the dynamics of individual political development and their implications for the timing, extent, and future trajectory of partisan polarization in the U.S. electorate.
Narrowly defined, demography deals with the measurement of vital events (birth, death, and marriage) and migration, studies the factors that influence the rate at which those events occur, and, to a lesser extent, investigates the consequences of the patterns of these events. In this paper, we adopt this narrow definition of the field and consider only the first three events so that by limiting the scope of our review, we can examine these selected topics in detail.
After an explanation of the structure of the 2006-2014 rolling panel of the General Social Survey (GSS), this report details models that estimate six sets of alternative predicted probabilities of attrition for all baseline sample members. The report then explains the cross-sectional GSS weights distributed with the data, and it shows how the estimated probabilities of attrition can be used to specify panel weights that adjust for attrition. Alternative approaches are discussed in conclusion. Code and data are provided in the associated repository.
In the five decades since its inception in 1971, the General Social Survey (GSS) project has prospectively recorded the current characteristics, backgrounds, behaviors, and attitudes of representative cross sections of American adults covering more than two generations and more than a century of birth cohorts. A foundational resource for contemporary social science, data it produces and disseminates enable social scientists to develop broad and deep understandings into the changing fabric of US society, and aid legions of instructors and students in teaching and learning. It facilitates internationally comparative survey research and places the United States in the context of other societies through the International Social Survey Program, which it cofounded. This article first recounts the GSS's origins, design, and development. It then surveys contributions based on GSS data to studies of stratification and inequality, religion, sociopolitical trends, intergroup relations, social capital and social networks, health and well-being, culture, and methodology.
Expected final online publication date for the Annual Review of Sociology, Volume 46 is July 30, 2020. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.
Age-period-cohort (APC) analysis has a long, controversial history in sociology and related fields. Despite the existence of hundreds, if not thousands, of articles and dozens of books, there is little agreement on how to adequately analyze APC data. This article begins with a brief overview of APC analysis, discussing how one can interpret APC effects in a causal way. Next, we review methods that obtain point identification of APC effects, such as the equality constraints model, Moore-Penrose estimators, and multilevel models. We then outline techniques that entail point identification using measured causes, such as the proxy variables approach and mechanism-based models. Next, we discuss a general framework for APC analysis grounded in partial identification using bounds and sensitivity analyses. We conclude by outlining a general step-by-step procedure for conducting APC analyses, presenting an empirical example examining temporal shifts in verbal ability.
Expected final online publication date for the Annual Review of Sociology Volume 45 is July 30, 2019. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.
This paper examines the identification problem in age-period-cohort models that use either linear or categorically coded ages, periods, and cohorts or combinations of these parameterizations. These models are not identified using the traditional fixed effect regression model approach because of a linear dependency between the ages, periods, and cohorts. However, these models can be identified if the researcher introduces a single just identifying constraint on the model coefficients. The problem with such constraints is that the results can differ substantially depending on the constraint chosen. Somewhat surprisingly, age-period-cohort models that specify one or more of ages and/or periods and/or cohorts as random effects are identified. This is the case without introducing an additional constraint. I label this identification as statistical model identification and show how statistical model identification comes about in mixed models and why which effects are treated as fixed and which are treated as random can substantially change the estimates of the age, period, and cohort effects. Copyright © 2017 John Wiley & Sons, Ltd.
The conventional linear model of age, period, and cohort (APC) effects suffers from a well-known identification problem: If an outcome depends on the sum of an age effect, a period effect, and a cohort effect, one cannot distinguish these three effects because birth year = current year − age. Less well appreciated is that the linear model suffers from a conceptual problem: It assumes that the marginal effect of age is the same at all times, the marginal effect of current conditions is the same for people of all ages, and cohorts do not change over time. We propose a new way of modeling APC effects that improves substantively and methodologically on the conventional linear model. We define cohort effects as an accumulation of age-by-period interactions. Although a long-standing literature conceptualizes cohort effects in exactly this way, we are the first to provide a statistical model. Our model allows age profiles to change over time and period effects to have different marginal effects on people of different ages. Except in special cases, the parameters of our model are identified. We apply the model to analyze changes in age-specific mortality in Sweden over 150 years. Our model fits the Swedish data dramatically better than the additive model. The rate of increase of mortality with age became more steep from 1881 to 1941, but since then has been roughly constant. The impact of early-life conditions lasts for several years but is unlikely to reach to old age.
Objective
In the wake of a massive shift in public opinion on same‐sex marriage, scholars have examined the factors that predict support of marriage equality. This research, largely from cross‐sectional or trend data, leaves questionable evidence for causality. In this research note, we use newly available panel data to predict attitudes toward marriage equality.
Methods
We use three three‐wave panels of survey data from the General Social Survey (2006–2014) to estimate the effects of both time‐variant and nonvariant characteristics on attitudes toward marriage equality.
Results
Using time‐variant factors, we support prior research, finding evidence for the causal effect of political views and time period on attitudes. Unlike earlier research, we find no significant effect of church attendance or marital or parental status. Among invariant characteristics, we largely support prior research on the effect of demographic characteristics on attitudes.
Conclusion
This analysis underscores the importance of using longitudinal data to support causal inferences and augment cross‐sectional research on political attitudes.
Age-period-cohort (APC) models are designed to estimate the independent effects of age, time periods, and cohort membership. However, APC models suffer from an identification problem: There are no unique estimates of the independent effects that fit the data best because of the exact linear dependency among age, period, and cohort. Among methods proposed to address this problem, using unequal-interval widths for age, period, and cohort categories appears to break the exact linear dependency and thus solve the identification problem. However, this article shows that the identification problem remains in these models; in fact, they just implicitly impose multiple block constraints on the age, period, and cohort effects to achieve identifiability. These constraints depend on an arbitrary choice of widths for the age, period, and cohort intervals and can have nontrivial effects on the estimates. Because these assumptions are extremely difficult, if not impossible, to verify in empirical research, they are qualitatively no different from the assumptions of other constrained estimators. Therefore, the unequal-intervals approach should not be used without an explicit rationale justifying their constraints.
The recent change in the general social survey (GSS) to a rotating panel design is a landmark development for social scientists. Sociological methodologists have argued that fixed-effects (FE) models are generally the best starting point for analyzing panel data because they allow analysts to control for unobserved time-constant heterogeneity. We review these treatments and demonstrate the advantages of FE models in the context of the GSS. We also show, however, that FE models have two rarely tested assumptions that can seriously bias parameter estimates when violated. We provide simple tests for these assumptions. We further demonstrate that FE models are extremely sensitive to the correct specification of temporal lags. We provide a simulation and a proof to show that the use of incorrect lags in FE models can lead to coefficients that are the opposite sign of the true parameter values.
In submitting this preliminary and tentative analysis of age patterns of participation of the white population in recent presidential elections, we hope to make three kinds of methodological contributions. First, we call attention to the unique design features of a substantial body of archival data that seems to have been insufficiently analyzed, and we begin the exploration of the error structure of these data. Second, we consider how the statistical model for cohort analysis suggested in recent literature (e.g., Mason et al. 1973; Pullum 1980) needs to be modified to take account of the pseudo-panel design of census surveys of voting and illustrate some kinds of inference that may be appropriate for the modified model. The intention here, in part, is to improve upon an earlier attempt (Duncan 1981) to formulate the issues raised by substituting a panel design for the conventional analysis of a series of independent cross-section samples. Third, we raise the question of whether an explicit and consistent theoretical basis for the statistical model can be developed using ideas drawn from recent work on latent-trait models (e.g., Duncan 1985, 1986). Corresponding to
This analysis of changes in the party affiliations of American adults between 1952 and 1972 (1) assesses the stability of the relationship between party and a set of causal variables and (2) examines the extent to which the observed changes are attributable to changes in the electorate's demographic composition. We found that indicators of stratification position, race, region, religion, and political socialization have exerted a nearly constant causal influence on party throughout the twenty-year period. A model which assumed constant effects (equal regression slopes) across the six elections explained only 2% less of the variance in party than a model which allowed the slopes to vary across elections. Of the variables in the causal model, socialization--as indicated by father's party preference--has the largest effect on party affiliation. The addition to our model of the demographic variables, age and cohort, revealed that both factors influence individuals' party affiliations. Though age and cohort explain only a small portion of the variance in party, examination of the net differences in mean party affiliation between age groups and between cohorts showed that aging does produce a net shift away from the Democratic party and that the Depression has had lasting effects on the preferences of cohort members formulating their preferences at that time.
This study found that an intercohort decline in vocabulary at all or most educational levels in the United States in recent years was closely related to an intercohort decline in newspaper reading. The decline in newspaper reading, in turn, may have resulted largely from an increase in television watching, but other influences, such as those from women's increased participation in the labor force, seem to have been involved as well. Other types of reading apparently declined in tandem with newspaper reading, and thus differences in reading are the most promising explanation for differences in the verbal ability of the various cohorts.
Do birth-cohort differences in family configuration brought about by post-World War II increases in fertility explain declines in verbal test scores of young people in the 1960s and 1970s? Data from nine representative samples of the U.S. population in the General Social Survey data file confirm systematic declines in verbal scores for cohorts born in the post-World War II era, but reveal a trend beginning much earlier, at least with cohorts born prior to 1920, and one sustained through cohorts born in the 1960s. Despite the significance of these intercohort patterns, within-cohort factors are much more important in producing variation in verbal scores. Social and economic characteristics of the respondent's family of origin and amount of schooling are associated with the largest differences in vocabulary knowledge. Sibship size has a significant influence on the development of verbal skills, but is relatively less important than other family background factors. Birth order, however, is not independently linked to verbal scores. Finally, owing in part to the relatively weak role of family configuration in producing variation in verbal scores, there is no support for the hypothesis that cohort differences in family experiences account for the trends in verbal ability across cohorts in the U.S. population.
This paper examines how ageing and generational formative experiences affect vote choices in Britain. Using a combination of panel data and assumptions about party fortunes we estimate ageing effects. These are then entered into a model using cross-sectional data from 1964 to 2010 to estimate generational differences in vote choice. Ageing increases the likelihood of a Conservative vote substantially, but there is no trend towards lower rates of Conservative voting among newer generations. There are however identifiable political generations corresponding with periods of Conservative dominance: voters who came of age in the 1930s, 1950s and 1980s are ceteris paribus somewhat more Conservative. Our method therefore lends some support to theories of political generations, but also demonstrates the considerable impact of ageing on vote choice.
Despite the importance of the “war on drugs,” little research has examined Americans' attitudes toward drug-related issues. In particular, the extant literature has not considered period and cohort effects for views about drug control issues, although there are theoretical reasons to expect both to be important. In this paper the roles of period and cohort effects, net of individual-level predictors, for attitudes toward both governmental spending to address drug addiction and legalizing marijuana use were examined. Data from the General Social Surveys (1975 through 2006) were used. The logistic regression results showed variations in attitudes over time for both addiction spending and legalizing marijuana use. Cohort effects were also evident. The results suggest that a liberalization of attitudes, both over time and across cohorts, may be occurring, especially for the legalization of marijuana use.
The notion that aging beyond adolescence and young adulthood leads to conservatism is part of the conventional wisdom, and there are theoretical reasons to believe that certain dimensions of biological, social and psychological aging contribute to some kinds of conser vatism. For instance, with the assumption of family respon sibilities, a diffuse liberalism-humanitarianism is likely to be overshadowed by concern for specific others. Or, aging persons may become more conservative in the sense that their attitudes and values become more resistant to change, because each subsequent experience is a smaller proportion of the total background of experiences. Empirical evidence on the topic is not definitive; moreover, in view of intransigent methodological problems which plague the study of aging effects, the evidence may never be definitive. However, cohort analysis of United States survey sample data reveals that in recent years persons aging beyond young adulthood and beyond middle age have tended to become more liberal in many respects, in conformity with general societal trends. However, these people have tended to become more conservative in a relative sense since their liberalization has not kept pace with changes in the total adult population. Although the evidence suggests that attitudes probably become somewhat less susceptible to change as people grow older, there is scant evidence for any other contribution of aging to conservatism.
Surveys spanning more than 35 years show that older Americans are less likely than younger citizens to endorse increased spending on public schools. The conventional explanation for this generational cleavage presumes that citizens’ interests change as they approach or transition into retirement—the absence of school-age children and fixed incomes combine to lower their interest in supporting spending increases for public education. We show that the conventional wisdom is incorrect, based on a confusion of age and cohort effects. Cohort-period analysis shows that every cohort becomes more supportive of educational spending, rather than less, as they reach their 60s and 70s. The implications are important, for they suggest that the predicted “gray peril” to educational spending will not occur. Rather, our results suggest that public support for educational spending will continue its remarkable rise.
Cohort analyses in which the joint effects of aging, historical change and birth cohort membership are estimated for some dependent variable are often desirable on substantive grounds. Unless two of these three variables are viewed as indexing identical unmeasured causal factors, any analysis which makes estimates for only two of the three variables is subject to spurious results. But three-way cohort analysis is problematic because age, time period and birth cohort are linearly dependent on each other. Although this confounding makes estimation of some three-way cohort models impossible, this paper demonstrates that estimation is feasible in a number of such models. By exploring estimates derived for some of these models from hypothetical data for which the underlying effects are known, this paper also shows that meaningful three-way cohort analysis is difficult unless the researcher entertains relatively strong hypotheses about the nature of aging, period and cohort effects.
This paper explores the "tree ring" hypothesis that reaching young adulthood during certain historical periods raises or lowers attitudes above and beyond the contribution of demographic variables and long-term cohort trends. It examines the deviations from long-term, linear, cohort trends for twenty-eight National Opinion Research Center (NORC) General Social Survey (GSS) attitude items in birth cohorts reaching age 16 in the 1950s, 1960s, and post-1960s. Long-term cohort trends are estimated from regressions of attitudes on cohorts reaching age 16 from 1917 to 1950 (net of year and five demographics). Popular impressions are supported in that "rings" (residuals) are more liberal for Americans reaching age 16 in the 1960s. However, those reaching age 16 in the 1950s are more liberal than their immediate predecessors, not more conservative. Furthermore, the three periods are not strikingly distinctive as the items showing positive rings tend to be the same in each period.
Two generations (1972-1976 and 2006-2008) are compared using 43 replicated attitudes in the NORC General Social Survey. The report describes the generational changes (primarily liberal), weighs the causal impact of rising educational levels (liberal), cohort replacement (liberal) and period effects (mildly conservative). It argues that this long term causal mechanism is slowly eroding.
This paper tests Stinchcombe's theory of “Demographic Explanations” using 38 demographics and 45 attitude items in the National Opinion Research Center's General Social Survey 1972–1996. The results support the theory in that multiple regressions with four appropriate demographics can explain about 40% of the change in the typical attitude and results are consistent across a variety of topics. Nevertheless, demographics seldom completely account for any particular trend. I argue that the theory is not as banal as it might seem since (1) it can and sometimes does fail; (2) its “multiplication principle” gives a new perspective on social change; and (3) it lays a burden of proof at the feet of more sophisticated approaches.
We develop a mixed (fixed and random effects) models approach to the age-period-cohort (APC) analysis of micro data sets in the form of a series of the repeated cross-section sample surveys that are increasingly available to sociologists. This approach recognizes the multilevel structure of the individual-level responses. As a substantive illustration, we apply our proposed methodology to data on verbal test scores from 15 cross-sections of the General Social Survey, 1974–2000. These data have been the subject of recent debates in the sociological literature. We show how our approach can be used to shed new light on these debates by identifying and estimating age, period, and cohort components of change.
Analysts often use different conceptual definitions of a cohort effect, and therefore different statistical methods, which lead to differing empirical results. A definition often used in sociology assumes that cohorts have unique characteristics confounded by age and period effects, whereas epidemiologists often conceive that period and age effects interact to produce cohort effects. The present study aims to illustrate these differences by estimating age, period, and cohort (APC) effects on obesity prevalence in the U.S. from 1971 to 2006 using both conceptual approaches. Data were drawn from seven cross-sectional waves of the National Health and Nutrition Examination Survey. Obesity was defined as BMI >or=30 for adults and >or=95th percentile for children under the age of 20. APC effects were estimated using the classic constraint-based method (first-order effects estimated and interpreted), the Holford method (first-order effects estimated but second-order effects interpreted), and median polish method (second-order effects are estimated and interpreted). Results indicated that all methods report significant age and period effects, with lower obesity prevalence in early life as well as increasing prevalence in successive surveys. Positive cohort effects for more recently born cohorts emerged based on the constraint-based model; when cohort effects were considered second-order estimates, no significant effects emerged. First-order estimates of age-period-cohort effects are often criticized because of their reliance on arbitrary constraints, but may be conceptually meaningful for sociological research questions. Second-order estimates are statistically estimable and produce conceptually meaningful results for epidemiological research questions. Age-period-cohort analysts should explicitly state the definition of a cohort effect under consideration. Our analyses suggest that the prevalence of obesity in the U.S. in the latter part of the 20th century rose across all birth cohorts, in the manner expected based on estimated age and period effects. As such, the absence or presence of cohort effects depends on the conceptual definition and therefore statistical method used.
The culmination of one of the most famous long-term studies in American sociology, this examination of political attitudes among women who attended Bennington College in the 1930s and 1940s now spans five decades, from late adolescence to old age. Theodore Newcomb's 1930s interviews at Bennington, where the faculty held progressive views that contrasted with those of the conservative families of the students, showed that political orientations are still quite malleable in early adulthood. The studies in 1959ââ¬â60 and 1984 show the persistence of political attitudes over the adult life span: the Bennington women, raised in conservative homes, were liberalized in their college years and have remained politically involved and liberal in their views, even in their sixties and seventies. Here the authors analyze the earlier studies and then introduce the 1984 data. Using data from National Election Studies for comparison, they show that the Bennington group is more liberal and hold its opinions more intensely than both older and younger Americans, with the exception of the generation that achieved political maturity in the 1960s. The authors point out that the majority of the Bennington women's children are of this 1945ââ¬â54 generation and suggest that this factor played an important role in the stability of the women's political views. Within their own generation, the Bennington women also appear to hold stronger political views than other college-educated women. Innovative in its methodology and extremely rich in its data, this work will contribute to developmental and social psychology, sociology, political science, women's studies, and gerontology. Duane F. Alwin is professor of sociology and research scientist at the Institute for Social Research, University of Michigan. Ronald L. Cohen is professor of psychology in the Social Science Division, Bennington College. Theodore M. Newcomb (1903-1984) was professor emeritus of psychology and sociology at the University of Michigan and had taught at Bennington College from 1934 to 1941.
The relationship of age to voting turnout over a 20-year period is analyzed in a multivariate model with controls for causal
covariates and “period” and “cohort” effects. The observed curvilinear pattern of turnout with age remains after holding rival
factors constant, but the apparent curvilinearity of cohort membership disappears. Instead, a pattern of decreasing turnout
among successively younger birth cohorts is found, suggesting differences in the political socialization of voting obligations
between the nineteenth and twentieth centuries.
Society persists despite the mortality of its individual members, through processes of demographic metabolism and particularly the annual infusion of birth cohorts. These may pose a threat to stability but they also provide the opportunity for societal transformation. Each birth cohort acquires coherence and continuity from the distinctive development of its constituents and from its own persistent macroanalytic features. Successive cohorts are differentiated by the changing content of formal education, by peer-group socialization, and by idiosyncratic historical experience. Young adults are prominent in war, revolution, immigration, urbanization and technological change. Since cohorts are used to achieve structural transformation and since they manifest its consequences in characteristic ways, it is proposed that research be designed to capitalize on the congruence of social change and cohort identification.
The author reexamines the hypothesis developed by Norman Ryder that the birth and death of individuals constitutes a massive process of personnel replacement that holds enormous potential for social change. "In this paper I describe and illustrate six possible ways to estimate cohort (personnel) replacement effects: three based on algebra (Kitagawa's two-component method, forward partitioning, and backward partitioning), and three based on regression (regression standardization, survey metric analysis, and linear decomposition). Assuming monotonic change, regression methods typically are better, because standard algebraic methods are ill suited for analyzing change with regard to birth cohorts that enter or exit during the period studied."
A New Compendium of Trends in the General Social Survey
- Hout Michael
Changing Same-Sex Marriage Attitudes in America From 1988 Through 2010
- Dawn Baunach
- Michelle
Schoolbook Simplification and Its Relation to the Decline in Sat-Verbal Scores
- Donald P Hayes
- Loreen T Wolfer
- Michael F Wolfe