psp75061389.tif Development of Attitude Strength Over the Life Cycle: Surge and Decline Penny S. Visser and Jon A. Krosnick Ohio State University This article explores the relation of age to manifestations and antecedents of attitude strength. Three studies demonstrate that susceptibility to attitude change is greater during early and late adulthood than during middle adulthood. Three additional studies demonstrate that attitude importance, cer- tainty, and perceived quantity of attitude-relevant knowledge are greater in middle adulthood than during early or late adulthood. These antecedents may therefore explain life cycle shifts in susceptibil- ity to change. Susceptibility to change, importance, certainty, and perceived knowledge differ from one another in terms of their correlations with education, gender, and race, challenging the notion that attitude strength is a unitary construct. Evidence that people incorrectly believe that susceptibility to change declines steadily over the life course reinforces the distinction between operative and recta- attitudinal measures of attitude strength. For decades, psychologists, sociologists, political scientists, and other social scientists have engaged in a lively debate over the relation b e t w e e n age and one o f the defining features o f attitude strength: susceptibility to attitude change ( f o r reviews, see A l w i n , Cohen, & Newcomb, 1991; Sears, 1975, 1981, 1983). Several hypotheses about this relation have b e e n ad- vanced, and the b u l k o f the empirical, evidence supports o n e o f them, the hypothesis that susceptibility is high d u r i n g early adulthood and significantly lower later in life. However, these investigations have involved methods that invite alternative ex- planations for the apparent relation b e t w e e n age a n d susceptibil- ity to change. Penny S. Visser and Jon A. Krosnick, Department of Psychology, Ohio State University. This research was supported by Grant 5T32-MH19728-03 from the National Institute of Mental Health (which provided a predoctoral fel- lowship to Penny S. Visser) and Grant SBR-9022192 from the National Science Foundation (which provided a fellowship to Jon A. Krosnick at the Center for Advanced Study in the Behavioral Sciences). Study 3 was made possible by a grant from the Electric Power Research Institute to Industrial Economics, Inc. We thank Laura Lowe for her role in collecting those data. A portion of this research was presented by Penny S. Visser to the graduate school at Ohio State University in partial fulfillment of the requirements for the master of arts degree. We thank Richard Herrmarm, Philip Tetlock, and Paul Sniderman for providing data sets and John Cacioppo and Richard Petty for very helpful suggestions. Correspondence concerning this article should be addressed to Penny S. Visser, who is now at the Department of Psychology, Princeton Univer- sity, Green Hall, Princeton, New Jersey 08544-1010, or Jon A. Krosnick, Department of Psychology, Ohio State University, 1885 Neff Avenue, Columbus, Ohio 43210. Electronic mail may be sent to pvisser@ princeton.edu or krosnick@osu.edu. I n this article, we describe studies taking a n e w approach to exploring susceptibility to attitude change across the life span, avoiding the pitfalls p l a g u i n g prior studies. We also explore the relation b e t w e e n age and various attitude features believed to be responsible for attitude strength. I n doing so, we g a i n n e w insights into the n a t u r e and structure o f attitude strength gener- ally. We b e g i n b y reviewing the hypotheses we tested a n d de- scribing previous tests o f them. Hypotheses Regarding the Relation Between Age and Susceptibility to Attitude Change Five principal views o f susceptibility to attitude change across the life span have b e e n proposed. The increasing persistence hypothesis suggests that susceptibility is high i n early adulthood a n d gradually decreases across the life span ( G l e n n , 1974, 1980; see F i g u r e 1 A ) . This hypothesis is predicated on the idea that attitudes reflect the a c c u m u l a t i o n o f relevant experiences across the life span, each o f which contributes to increasing stability. I n addition, as individuals age, they typically b e c o m e increasingly entrenched in social networks o f others with similar life experi- ences and worldviews (Berelson, Lazarsfeld, & McPhee, 1954; Newcomb, Koenig, blacks, & Warwick, 1 9 6 7 ) , which is likely to reinforce existing attitudes. A second view, the impressionable years hypothesis, is based o n the n o t i o n that core attitudes, beliefs, a n d values are crystal- lized d u r i n g a period o f great plasticity in early adulthood (Sears, 1975). This period o f plasticity is thought to reflect the transition f r o m adolescence to adulthood, marked by n e w f o u n d interest in events outside o n e ' s i m m e d i a t e surroundings and a b u d d i n g political worldview ( D a w s o n & Prewitt, 1969). For the first time, y o u n g adults are able to participate in political elec- tions, they c a n choose to serve i n the military, a n d they m a y enter the work_force full time. In these ways a n d m a n y others, Journal of Personality and Social Psychology, 1998, Vol. 75, No. 6, 1389-1410 Copyright 1998 by the American Psychological Association, Inc. 0022-3514/98/$3.00 1389 1390 V I S S E R A N D K R O S N I C K A II1 t - r" 0 o ~ W 0 i i i i f ~ ~ i i , i , , , J 18 23 2 8 33 3 8 4 3 48 53 58 63 68 73 78 83 88 Age (in years) B 1 CD e- CU 0 ,,Q .,j 0 , , , , i , i , , , , i i 1 18 23 2 8 33 38 43 4 8 53 5 8 63 68 73 78 83 88 Age (in years) C G) O1 c0 r . O q) 2 o m .4D O . u) , , , , , , , , , , , , , , 18 23 2 8 33 38 43 48 53 58 63 68 7 3 78 83 88 Age (in years) Figure 1. The increasing persistence hypothesis (A), the impression- able years hypothesis (B), and the life-stages hypothesis (C). young adults interact for the first time with social and political institutions and begin to formulate opinions on a wide range o f issues. During this time, they also learn a great deal about their social and political surroundings (Strate, Parrish, Elder; & Ford, 1989), which further contributes to the increasing crystallization of their views o f the world. According to the impressionable years hypothesis, the basic orientations formed during this pe- riod o f early adulthood remain largely unaltered throughout the remaining adult years (Carlsson & Karlsson, 1970; Mannheim, 1952; Ryder, 1965). Thus, the impressionable years hypothesis suggests that susceptibility to attitude change is high during early adulthood but drops sharply after this period and remains low throughout the remainder o f the life span (see Figure 1B). T h e life stages hypothesis also predicts high susceptibility to change during early adulthood and lower susceptibility through- out middle adulthood, for all o f the reasons just outlined. But this hypothesis predicts high susceptibility to attitude change in later life as well (Sears, 1981, 1983; see Figure 1C). This upturn in susceptibility at the end o f the life cycle may occur because o f decreases in social support for one's attitudes due to the deaths o f friends or social withdrawal (Burt, 1990; Lang & Carstensen, 1994; Marsden, 1987). In addition, both the early and late adult years are associated with a multitude of role transitions (GlaSer & Strauss, 1971; Steckeurider & Cutler, 1989) that may alter people's perceptions o f the social and political world, thereby undermining the justifications that rein- forced many o f their attitudes. Finally, cognitive skills decline toward the end o f the life cycle (see, e.g., Verhaeghen & Salt- house, 1997), so people may become less and less able to ac- tively resist attitude change through counterarguing. What we call the "perpetual susceptibility" hypothesis is a sort o f null hypothesis proposing no relation between age and susceptibility to attitude change. And cross cutting all of the preceding hypotheses are two assertions about the general level o f susceptibility to change across the life cycle. The lifelong openness hypothesis proposes that individuals are at least some- what susceptible to attitude change throughout their lives (Brim & Kagan, 1980; Gergen, 1980; Lernet; 1984), whereas the persis- tence hypothesis suggests that most o f individuals' fundamental orientations are established so firmly during preadult socialization that susceptibility to attitude change thereafter is very low (Da- vies, 1965; Easton & Dennis, 1969; Hess & Torney, 1967). P r e v i o u s I n v e s t i g a t i o n s A variety o f methods, always focused on social and political attitudes, have been used in testing these hypotheses. One set o f studies tracking aggregate changes in the attitudes o f birth cohorts has generally revealed large attitude shifts in young cohorts and much smaller shifts in middle-aged and older co- horts (Converse, 1976; Converse & Markus, 1979; Crew, Sarl- vik, & Alt, 1977; Cutler & Kaufman, 1975; Glenn, 1974; Jen- nings & Niemi, 1981; Markus, 1979; Mueller, 1973; Nunn, Crockett, & Williams, 1978). Some o f these investigations have been cited as supporting the impressionable years hypothesis (e.g., Crew et al., 1977; Mueller, 1973), whereas others have been interpreted as supporting the increasing persistence hypoth- esis (e.g., Glenn, 1974; Nunn et al., 1978). Such aggregate measures o f attitude change, however, can be misleading, be- cause a great deal o f cross-cutting attitude change may occur at the individual level that cancels out at the aggregate level. To overcome this problem, a number o f researchers have as- sessed individual-level attitude stability using test-retest corre- lations o f the same attitude measure taken at two or more points ATTITUDE STRENGTH 1391 in time ( e . g , Campbell, Dunnette, Lawler, & Weick, 1970; Jen- nings, 1996; Jennings & Markus, 1984; Jennings & Niemi, 1978; Johansson & Campbell, 1971; Markus, 1979; Newcomb et al., 1967). In general, these studies have revealed weaker t e s t - r e t e s t correlations among younger adults than among older adults, as the impressionable years and increasing persistence hypotheses predict.1 However, t e s t - r e t e s t correlations are not pure measures o f attitude stability. In addition to genuine attitude change, such correlations reflect random measurement error in attitude re- ports. Thus, weaker correlations among younger age groups may indicate lower attitude stability or may simply reflect more measurement error. To separate true attitude change from random measurement error, a number o f studies have used panel data to estimate attitude stability more purely via structural equation modeling or related techniques. Although one o f these studies yielded results consistent with the increasing persistance hypothesis ('I~ler & Schuller, 1991), most have supported the impression- able years hypothesis, showing that attitude stability was lowest among 18-25-year-olds and was higher throughout the rest o f the life cycle ( A l w i n et al., 1991; Krosnick & Alwin, 1989; Sears & Funk, 1996). 2 Eliminating measurement error in estimates o f attitude stabil- ity certainly represents a methodological improvement, but this general approach to hypothesis testing still has limitations. Atti- tude stability (no matter how purely estimated) is an imperfect measure o f susceptibility to attitude change because stability reflects both susceptibility to change and opportunity for change. The more attitude-challenging experiences a person has, the more likely attitude change is. And i f exposure to attitude-chal- lenging experiences varies across the life span, attitude stability is likely to be an inaccurate index o f susceptibility to change. I n an effort to more directly assess the relation o f age to susceptibility, "l~ler and Schuller (1991) included measures o f potentially change-inducing experiences in their investigation o f attitude stability across the life span. And, indeed, exposure to some potentially change-inducing experiences was more fre- quent among younger adults. Furthermore, "l~yler and Schuller ( 1991 ) found that the correlation between the reported positive or negative nature o f the attitude-relevant experiences and subse- quent attitudes was highest among older respondents, which they interpreted as evidence that older adults w e r e more susceptible to change than younger adults. Although this correlation may be evidence that relevant per- sonal experiences influenced attitudes, the reverse could have been true instead: Attitudes could have influenced the reported nature o f experiences via self-fulfilling prophecies; people who expected to have bad experiences may have behaved in ways that made the experiences bad. Or attitudes could have shaped the way people perceived their experiences, leading the experi- ences to appear consistent with positive or negative expectations. Alternatively, attitudes may h a v e biased recall o f experiences, such that they were remembered as more consistent with current attitudes than they actually were ( C o n w a y & Ross, 1984; Lydon, Zanna, & Ross, 1988; Ross, McFarland, & Fletcher, 1981). I f these alternative processes were more common or more power- ful among older a d u l t s t h a n among young adults, they would have misleadingly inflated apparent levels o f susceptibility to change among the former individuals. Thus, although incorporating measures o f change-inducing experiences represented a step in the fight direction, this general approach to investigating the relation between age and suscepti- bility to attitude change is inherently limited. It is difficult if not impossible to precisely assess the number and nature o f change-inducing experiences respondents have had and to gauge their impact correlationally, Furthermore, nearly all prior studies o f these issues have suf- fered from a confounding o f age with cohort composition in ways that may have distorted the results obtained. For example, during the last 45 years, the proportion o f young adults attending college has been rising consistently (U.S. Bureau o f the Census, 1953, 1993). In 1952, 18% o f 18-24-year-olds in the United States had completed 1 or more years o f college, whereas this figure was 45% in 1993 (U.S. Bureau o f the Census, 1953, 1993); thus, current 18-year-olds are more educated, on average, than current 65-year-olds. Because better educated people are less susceptible to political attitude change (Sears & Gahart, 1980; ZaUer, 1990), the relatively high levels o f education in the youngest cohorts may have caused their rates o f attitude change to appear misleadingly low. To generate a clean compari- son o f different age groups, one mast control for differences between groups in terms o f educational attainment. Patterns o f mortality can also distort comparisons o f age groups in representative cross-sectional samples o f adults. For example, women tend to live longer than men, so older age groups are likely to be disproportionately female (Centers for Disease Control and Prevention, 1995). Because women may be more susceptible to attitude change than men (Eagly & Carli, 1981), the greater number o f women among the oldest age group may produce an illusory rise in susceptibility among this group, thus perhaps masking a true decline in susceptibility at the end of the life cycle. The demographic composition o f age groups is also likely to differ in terms o f race. Non-Whites have shorter life expectan- cies than Whites, which means that older age groups may be disproportionately White (Centers for Disease Control and Pre- vention, 1995; Markides & Black, 1996). On some issues, racial minority and majority group members m a y hold attitudes that differ in extremity, personal importance, or other factors associ- ated with resistance to attitude change (see, e.g., Petty & Kros- nick, 1995). Therefore, controlling for race may also provide • a more accurate assessment o f the relation between age and susceptibility to attitude change. T h e P r e s e n t I n v e s t i g a t i o n A more direct investigation o f the relation between age and susceptibility to attitude change would require exposing adults o f various ages to an identical change-inducing experience and Sears (1981) suggested, howeve~ that careful exploration of the full adult life span yields a different pattern in at least one domain: In a large panel survey, he found that test-retest correlations for racial atti- tudes were lowest among very young adults and among very old adults, consistent with the life stages hypothesis. 2 Like Sears ( 1981 ), Krosnick and Alwin (1989) found a curvilinear pattern of attitude stability across the life cycle. However, they also found decreased reliability among the oldest respondents, which accounted for the drop in attitude stability at the end of the life cycle in their data and may also account for the pattern reported by Sears ( 1981 ). 1 3 9 2 VISSER AND KROSNICK g au ging its impact. C o n d u c t i n g a laboratory e x p e r i m e n t along these lines w o u l d b e quite challenging, b e c a u s e it w o u l d b e e x t r e m e l y difficult to recruit c o m p a r a b l y representative samples o f adults o f various ages to v i s i t a laboratory. B u t large, repre- sentative samples are relatively easy to contact in a t e l e p h o n e survey, and computer-assisted t e l e p h o n e i n t e r v i e w i n g systems readily p e r m i t the e x e c u t i o n o f e x p e r i m e n t a l manipulations ( s e e S n i d e r m a n & Grob, 1996). In the first t w o studies r e p o r t e d here, w e used precisely this approach. In the context o f t e l e p h o n e interviews, representative samples o f adults were asked to express their attitudes o n vari- ous social and political issues; they w e r e then subjected to poten- tially change-inducing experiences, and attitude change was as- sessed. This p e r m i t t e d gauging o f differences b e t w e e n age groups in rates o f attitude change. In o u r third study, age-related differences in thought-induced attitude c h a n g e w e r e explored, again within the context o f a telephone s u r v e y o f a representative s a m p l e o f adults. In o u r final three studies, w e b r o a d e n e d the f o c u s b y e x a m i n i n g s o m e o f the possible causes o f the o b s e r v e d relation b e t w e e n age and susceptibility to attitude change. S t u d y 1 O u r first study f o c u s e d o n attitudes toward foreign p o l i c y positions. Respondents w e r e presented with a hypothetical sce- n a r i o o f an international conflict and w e r e asked to express their attitudes t o w a r d U.S. military intervention in that situation. T h e n all respondents w e r e asked whether t h e y w o u l d change their minds i f the United Nations r e c o m m e n d e d the o p p o s i t e o f their initial preference. C o n t r o l l i n g f o r attitudes t o w a r d the United Nations and d e m o g r a p h i c c o m p o s i t i o n o f the age groups, w e assessed the relation o f age to attitude change. M e t h o d Sample The data analyzed in this study were collected via telephone inter- views with a representative sample of 1,511 English-speaking U.S. adults living in private households. The data were collected by the Ohio State University Survey Research Unit (then called the Polimetrics Labora- tory) between April and June 1996 (for details, see Herrmann, "Fetlock, & Visser, 1997). Measures Attitudes toward U.S. military intervention. Respondents were told about a military conflict between two hypothetical countries and were given information about the military strength of the attacking country, the history of the relationship between the two countries, and the impor- tance of the attacked country to U.S. economic and security interests. Respondents were then asked " I f the attacker cannot be talked into withdrawing, should our government use our military to push back the invaders, or should we stay out of it?" Respondents who said we should stay out were asked " I f the United Nations believes that Americans should use military force in this conflict, would you be willing to change your mind about using our military?" Respondents who initially said we should use military force were asked, instead, " I f the United Nations believes that Americans should not use military force in this conflict, would you be willing to change your mind about using our military?" A measure of attitude change was coded 0 for respondents who said they would not change their minds and 1 for respondents who said they would change their minds. Forty-six percent of respondents said they would change their minds, and 54% said they would not. Attitudes toward the United Nations. Respondents were asked to indicate their feelings toward the United Nations on a scale ranging from 0 to 100, where 50 represented neutrality; higher numbers repre- sented warmer, more favorable feelings; and lower numbers represented colder, less favorable feelings. Demographic measures. Educational attainment was coded 0 for people who did not graduate from high school, .25 for high school graduates, .50 for people who attended college but did not graduate, .75 for college graduates, and 1 for people who completed postgraduate work. Gender was coded 0 for female respondents and 1 for male respon- dents, and race was coded 0 for Whites and 1 for non-Whites. Age was measured in years and recoded to range from 0 (age 18) to 1 (age 89, the highest age in the sample). 3 Results Relations o f A g e to Education, Gender, a n d R a c e A s expected, age was marginally significantly correlate d with education, such that older age groups c o n t a i n e d f e w e r highly educated p e o p l e than did younger age groups ( r = - . 0 6 , p < .05, n = 1,276). 4 Also, older age groups contained dispropor- tionately f e w e r minorities than did y o u n g e r cohorts ( r = - . 11, p < .01, n = 1,276), although n o significant relation appeared b e t w e e n gender and age ( r = .04, ns, n = 1,276). C o n t r o l l i n g f o r the d e m o g r a p h i c c o m p o s i t i o n o f the age groups was therefore necessary to e x p l o r e the e f f e c t o f age on attitude change. 5 I m p a c t o f A g e on Susceptibility to Change We first r e g r e s s e d the attitude change m e a s u r e on age and the three other d e m o g r a p h i c variables ( s e e c o l u m n 1 o f Table 1). Consistent with the increasing persistence hypothesis, the linear e f f e c t o f a g e was statistically significant and negative ( B = - 0 . 1 2 , p < .05), m e a n i n g that the likelihood o f change d e c r e a s e d w i t h increasing age. However, the i m p r e s s i o n a b l e years and life stages hypotheses predict quadratic effects o f age, w h i c h w e tested b y adding age squared as a predictor in the regression equation ( s e e c o l u m n 2 o f Table 1). T h e linear e f f e c t o f age in this equation was negative and significant ( B = - 0 . 4 7 , p < . 0 5 ) , and the quadratic e f f e c t was positive and marginally significant ( B = 0.38, p < .10). B e c a u s e the e f f e c t o f age c u b e d ( s h o w n in c o l u m n 3 o f Table 1) was not significant ( B = - 0 . 6 9 , ns), the linear and quadratic age effects in c o l u m n 2 p r o v i d e d a c o m p l e t e descrip- tion o f the robust relation o f age to attitude change. 3 Nine respondents did not report their ages and were not included in our analyses. Small numbers of respondents were not included in our subsequent studies for the same reason. 4 Because we had strong a priori expectations of the directions of age- related differences, we present one-tailed hypothesis tests throughout. s The predicted relations between age and both gender and race are due to differential rates of mortality, and, as such, they should be evident only at the end of the life cycle. And the predicted relation between age and education reflects cohort differences in educational attainment between Americans born before and after the 1930s and should also be evident only at the end of the life cycle. To capture these relations, therefore, we have reported the correlations between each of these demo- graphic factors and age squared. ATI'rrUDE STRENGTH 1393 em e4 • --, I ¢ q O e q ¢,4 o= a ~ o% o e q o i I i I ~ ~ ~ ~ ' I i I I I I I I I e ~ ¢~¢5 ¢5 ¢5 ¢~ d I I I I I I I I I I t ~ O x ~ v v *¢ v To understand the shape o f this relation, we used the regres- sion equation that included the quadratic effect o f age to estimate rates o f attitude change at various ages across the life span. We plugged into this equation the mean sample values o f gender, race, education, and attitudes toward the United Nations for the full sample, and we calculated the predicted attitude change rate for each o f a wide array o f ages ( f o r an explanation o f the procedure, see Cohen & Cohen, 1983, pp. 2 2 8 - 2 2 9 ) . Figure 2 provides a graphic representation o f the regression equation presented in column 2 o f Table 1. As Figure 2 illustrates, attitude change decreased with increasing age up to about age 60, at which point attitude change became increasingly common with increasing age. This is consistent with the life stages hypothesis. Other Correlates o f Susceptibility to Change Attitude change was related sensibly to attitudes toward the United Nations: People with more positive attitudes changed more often (B = 0.49, p < .01 ).6 However, contrary to some prior literature, attitude change rate was not significantly corre- lated with gender, race, or education (B = - 0 . 0 4 , - 0 . 0 2 , and 0.01, respectively). Study 2 Although the shifts in respondents' attitude reports observed in Study 1 may have been due to genuine attitude change, they may also simply have been due to what Kelman (1958) called compliance. Specifically, respondents may have perceived the attitude-challenging questions to be suggesting that the inter- viewer or researcher believed that they should change their atti- tudes in response to the new information provided. Conse- quently, to convey impressions o f themselves as cooperative people to their interviewers, some respondents may have claimed that the new information would change their attitudes when this was not, in fact, the case. Because young adults may be especially concerned with presenting favorable self-images (Reifman, Klein, & Murphy, 1989; Tesch, 1983), the evidence o f greater attitude change among such individuals could have been due to a greater chronic concern with conveying favorable impressions to their interviewers rather than a greater general susceptibility to attitude change. We explored this alternative explanation in Study 2. Study 2 involved an experimental design similar to that used in Study 1. Respondents were asked to express their attitudes toward various policies involving issues o f race. Immediately after each attitude measurement, respondents were presented with a counterattitudinal argument, and attitudes toward the policy were then remeasured. Here, we were able to assess the relation between age and susceptibility to attitude change controlling for age-related differences in the motivation to make favorable impressions on others. 6 We also estimated an additional regression model that included terms representing the interactions between attitudes toward the United Nations and (a) age, (b) age squared, and (c) age cubed. None of these interac- tion terms were significant predictors of attitude change. 1 3 9 4 VISSER AND KROSNICK 0.58 0 0.56 ~ 0.54 ~ ~ 0.48 "6 ~ 0.48 0.44 0.42 4) L 0.40 i i i , i , , i i f , i i i , 1 8 2 3 2 8 3 3 38 43 48 53 58 6 3 8 8 7 3 78 83 88 A g e (in y e a r s ) Figure 2. Relation of age to attitude change in Study 1. Method Sample The data analyzed in this study were collected via telephone inter- views with a representative sample of 1,113 English-speaking adult resi- dents of the five-county San Francisco-Oakland Bay area in northern California (for details, see Sniderman & Piazza, 1993).7 The data were collected by the Survey Research Center of the University of California, Berkeley, between August and October 1986. Measures Four items asked respondents whether they supported or opposed (a) increased federal spending on programs to help Blacks, (b) antidiscrimi- nation housing laws, (c) antidiscriminatory employment laws, and (d) preferential admission to colleges and universities for Black students (for the questions' wordings, see Sniderman & Piazza, 1993, pp. 142- 146). Immediately after each of these questions, respondents were pre- sented with a brief counterattitudinal challenge from an unspecified source, and attitude change was measured. For example, respondents initially supportive of preferential admission to colleges and universities for Black students were asked "Would you still feel that way, even if it means fewer opportunities for qualified Whites, or would you change your mind?" Conversely, respondents initially opposed to the admission policy were asked "Would you still feel that way, even if it means that hardly any Blacks would be able to go to the best colleges and universities, or would you change your mind?" Responses were coded 0 for people who said they would not change their minds on any of the issues and 1 for people who said they would change their minds on at least one issue. Seventy-three percent of the respondents said they would change their opinions on at least one issue, and 27% said they would not change their opinions on any issues. 8 As a means of assessing motivation to make favorable impressions on others, respondents were asked "How important is it to you to be accepted by other people: very important, somewhat important, or not important?" Responses were coded 3, 2, and 1, respectively. Education, gender, race, and age were measured and coded as in Study 1. Results Relations of Age to Education, Gender, and Race O l d e r age cohorts contained significantly f e w e r w e l l - e d u c a t e d p e o p l e than y o u n g e r cohorts ( r = - . 2 9 , p < .01, n = 1,086), but age was unrelated to gender ( r = - . 0 2 , ns, n = 1,086) and race ( r = - . 0 3 , ns, n = 1,086). Relation of Age to Susceptibility to Change W h e n d e m o g r a p h i c characteristics w e r e controlled, the linear relation o f age to attitude change was marginally significant and negative, w h i c h is consistent with the increasing persistence hypothesis ( B = - 0 . 0 9 , p < .10; see c o l u m n 4 o f Table 1). However, the quadratic e f f e c t o f age was also positive and sig- nificant ( B = 0.69, p < .05; see c o l u m n 5 o f Table 1 ), and the negative linear e f f e c t o f age was significant in the equation that i n c l u d e d the quadratic e f f e c t o f age ( B = - 0 . 5 9 , p < .01 ). T h e c u b i c e f f e c t o f age was again not significant ( B = - 0 . 0 5 , ns; see c o l u m n 6 o f Table 1 ). F i g u r e 3 illustrates the rates o f attitude change p r e d i c t e d b y the equation in c o l u m n 5 o f Table 1, and they are again clearly in line with the life stages hypothesis: Attitude change rates were highest a m o n g the youngest and oldest adults. Other Correlates of Susceptibility to Change A s expected, p e o p l e w h o had a greater n e e d to b e a c c e p t e d b y others w e r e m o r e likely to change their attitudes ( B = 0.09, p < .05).9 B e c a u s e the relation b e t w e e n age and attitude change r e m a i n e d significant even when this variable was i n c l u d e d in the r e g r e s s i o n equation ( s e e c o l u m n 5 o f Table I ) , n e e d f o r acceptance c a n n o t account for the o b s e r v e d age-related differ- ences in susceptibility to change. Attitude change was also m o r e c o m m o n a m o n g less e d u c a t e d respondents than a m o n g m o r e educated respondents ( B = - 0 . 1 6 , p < .01 ), and m o r e change o c c u r r e d a m o n g w o m e n than a m o n g m e n ( B = - 0 . 0 6 , p < .01). N o relation appeared b e t w e e n attitude change rate and race ( B = 0.05, ns). S t u d y 3 T h e results o f Study 2 suggest that the age-related differences in attitude change o b s e r v e d in Studies 1 and 2 w e r e not the result o f a stronger motivation a m o n g particular age groups to b e responsive to the perceived expectations o r wishes o f the 7 A potentially important advantage of these data is the inclusion of people living in dormitories or serving in the military in the sampling frame. Such individuals were excluded from the National Election Study surveys analyzed in many past studies of aging and attitude change. As Krosnick and Alwin (1989) noted, people living in dormitories (and perhaps people serving in the military as well) are likely to be primarily young adults. If those individuals were more resistant to attitude change (because of their high levels of education, strong social networks, or other factors) than their peers who were not living in dormitories or serving in the military, excluding the former from past analyses may have been partly responsible for the age-related differences observed previously. Our results are not subject to this confound and can therefore make a clearer statement about age-related differences. s Other ways of ceding the attitude change dependent variable yielded comparable results to those reported subsequently. 9 We also estimated an additional regression model that included terms representing the interactions between the importance of being accepted by others and (a) age, (b) age squared, and (c) age cubed. None of these interaction terms were significant predictors of attitude change. ATTITUDE STRENGTH 1 3 9 5 a . 6 9 i i I i i i i i i i t i i i i m 18 2 3 2 8 3 3 3 8 43 4 8 5 3 58 6 3 6 8 7 3 7 8 83 8 8 9 3 A g e ( i n y e a r s ) o 8 9 77 ILl 71 Figure 3. Relation of age to attitude change in Study 2. interviewer. However, t h e a n a l y s e s i n S t u d y 2 r e l i e d o n a s i n g l e i t e m to m e a s u r e t h i s m o t i v a t i o n . T h e r e l i a b l e a s s o c i a t i o n b e - t w e e n s e l f - r e p o r t s o f t h e i m p o r t a n c e o f b e i n g a c c e p t e d b y o t h e r s a n d s u b s e q u e n t a t t i t u d e c h a n g e s u g g e s t s t h a t t h e s i n g l e i t e m d i d c a p t u r e m e a n i n g f u l v a r i a n c e in t h e m o t i v a t i o n t o m a k e f a v o r a b l e i m p r e s s i o n s o n others, b u t a m o r e p r e c i s e m e a s u r e m e n t o f t h i s m o t i v a t i o n m a y h a v e y i e l d e d d i f f e r e n t results. For a s t r o n g e r t e s t o f t h i s a l t e r n a t i v e e x p l a n a t i o n , w e t u r n e d to a d e s i g n t h a t c o m p l e t e l y r e m o v e d t h e i n t e r v i e w e r f r o m t h e a t t i t u d e c h a n g e p r o c e s s , e l i m i n a t i n g t h e p o s s i b i l i t y t h a t o b s e r v e d a t t i t u d e shifts m a y s i m p l y r e f l e c t s e l f - p r e s e n t a t i o n a l c o n c e r n s . C o n s i d e r a b l e e v i d e n c e i n d i c a t e s t h a t t h e p r e s e n t a t i o n o f a p e r s u a s i v e m e s s a g e is n o t r e q u i r e d f o r a t t i t u d e c h a n g e to occur. S i m p l y t h i n k i n g a b o u t a n a t t i t u d e o b j e c t c a n c a u s e p e o p l e to b e c o m e m o r e p o s i t i v e o r m o r e n e g a t i v e t o w a r d t h e o b j e c t ( s e e , e.g., N e i j e n s , 1987; Tesser, M a r t i n , & M e n d o l i a , 1 9 9 5 ) . I f a g e is i n d e e d r e l a t e d t o s u s c e p t i b i l i t y to a t t i t u d e c h a n g e , t h o u g h t - i n d u c e d a t t i t u d e c h a n g e s h o u l d b e p a r t i c u l a r l y p r o n o u n c e d a m o n g t h e y o u n g e s t a n d o l d e s t r e s p o n d e n t s . O n t h e o t h e r h a n d , i f a p p a r e n t a g e - r e l a t e d d i f f e r e n c e s i n s u s c e p t i b i l i t y to a t t i t u d e c h a n g e w e r e a r t i f a c t s o f t h e i n t e r v i e w c o n t e x t , t h e m a g n i t u d e o f t h o u g h t - i n d u c e d a t t i t u d e c h a n g e s h o u l d n o t v a r y s i g n i f i c a n t l y a c r o s s a g e g r o u p s . I n S t u d y 3, w e t e s t e d t h e s e h y p o t h e s e s . T h e a t t i t u d e c h a n g e m e a s u r e u s e d h e r e w a s a b i t d i f f e r e n t f r o m t h a t u s e d i n S t u d i e s 1 a n d 2. I n t h o s e studies, r e s p o n d e n t s w e r e p r e s e n t e d w i t h a c o u n t e r a t t i t u d i n a l m e s s a g e a n d r e p o r t e d w h e t h e r o r n o t t h e n e w i n f o r m a t i o n w o u l d c a u s e t h e m to c h a n g e t h e i r attitudes. S t u d y 3 m e a s u r e d a t t i t u d e s e i t h e r b e f o r e o r a f t e r p e o p l e w e r e i n d u c e d to t h i n k a b o u t t h e o b j e c t so t h a t w e c o u l d d i r e c t l y a s s e s s a t t i t u d e c h a n g e . Method Sample For this study, a representative sample of 505 English-speaking adults living in private households in Ohio were interviewed by telephone. The survey was conducted by the Ohio Stute University Survey Research Unit (then called the Polimetrics Laboratory) between December 1995 and January 1996. Measures Respondents' attitudes toward global warming were assessed via the following question: Scientists use the term "global warming" to refer to the idea that the world's average temperature may be about 5 degrees Fahrenheit higher in 75 years than it is now. Overall, would you say that global warming would be good, bad, or neither good nor bad? As a means of best capturing the attitude change that occurred, good and neither good nor bad were coded 0, and bad was coded 1. Education, gender, race, and age were measured and coded the same as they had been in Studies 1 and 2. Thought Induction The thought induction manipulation used was inspired by the struc- tured reasoning tusk used in the Choice Questionnaire (Neijens, 1987). Respondents were randomly assigned to be asked about their attitudes toward global warming either before the structured thinking exercise or after it. In the thinking exercise, participants answered a series of questions about whether or not they thought global warming would affect seven aspects of the environment. Respondents were randomly assigned to be asked about either one set of seven aspects (sea levels, water shortages, the number of different kinds of plants in the world, the overall number of plants in the world, the number of plants in the county in which the respondent lived, natural scenery in the world, and natural scenery in the county in which the respondent lived) or another (annual rainfall, hurricanes and tornadoes, the number of different types of animals in the world, the overall number of animals in the world, the number of animals in the county in which the respondent lived, animal migration, and availability of food for human consumption). Respondents also reported the certainty with which they held each belief and how good or bad each expected consequence would be. For example, respondents were asked " I n the next 100 years, do you think global warming would cause the level of the oceans in the world to go up, to go down, or would global warming have no effect on the sea level?" Some respondents volunteered that global warming would cause the sea level to go up in some places and down in others, and interviewers accepted such answers. Respondents were then asked "How sure are you of your answer to that question? Extremely sure, very sure, somewhat sure, or not sure at all?" Finally, respondents who said that global warming would affect the sea levels were asked whether the change they expected (sea levels going up or down or both) would be good or bad for most people in the w o r d and to what degree. For example, respondents who said that global warming would cause the sea levels to go up were asked "For most people in the world, would the sea level going up be good, bad, or neither good nor b a d ? " Respondents who said " g o o d " or " b a d " were asked "Would it be very [good/bad], or somewhat [good/bad] ? " Simi- lar questions were asked with respect to all of the other potential conse- quences of global warming. Results Relations of Age to Education, Gender, and Race O l d e r a g e c o h o r t s c o n t a i n e d s i g n i f i c a n t l y f e w e r w e l l - e d u c a t e d p e o p l e t h a n y o u n g e r c o h o r t s ( r = - . 1 2 , p < .01, N = 505 ) a n d f e w e r r a c i a l m i n o r i t i e s ( r = - . 1 1 , p < .05, N = 5 0 5 ) , b u t a g e w a s u n r e l a t e d to g e n d e r ( r = .05, ns, N = 5 0 5 ) . 1396 VISSER AND KROSNICK Impact of the Thought Induction on Attitudes To examine the impact o f the thought induction on attitudes toward global warming, we regressed attitudes on a dummy variable coded 0 for people who expressed their attitudes before the thought induction exercise and 1 for people who expressed their attitudes after the thought induction. The question order manipulation significantly affected attitudes: Thinking about po- tential consequences caused respondents to express significantly more negative attitudes (B = 0.16, p < .001). Relation of Age to Susceptibility to Change To determine whether the magnitude o f this thought-induced attitude change varied by age, we conducted additional regres- sions predicting attitudes with question order, age, and the prod- uct o f question order and age. A significant regression coefficient for this interaction term would indicate that thought-induced attitude change varied as a function o f age. To control for com- positional differences across age groups, we included education, gender, and race in the regression, along with interaction terms between each o f these demographics and question order. The interaction between age represented linearly and question order was not significant (B = - 0 . 1 3 , ns; see column 7 o f Table 1 ). However, when we added age squared and the interac- tion between age squared and question order to the equation, the quadratic effect o f age on attitude change was positive and significant (B = 1.64, p < .05; see column 8 o f Table 1 ), and the negative linear effect o f age on attitude change was also significant in that equation (B = - 1 . 5 0 , p < .05). The cubic effect o f age on attitude change was again not significant (B = 0.03, ns; see column 9 o f Table 1 ). Figure 4 illustrates the rates o f attitude change predicted by the equation in column 8 o f Table 1. As in Studies 1 and 2, there is clear support for the life stages hypothesis here: Attitude change rates were highest among the youngest and oldest adults. 0 0 45 ~40 • 35 2O mO)~ 15 ~ lO a . , i , i i i i , , i i i i i 18 23 28 33 38 43 48 53 58 63 68 73 78 83 86 A g e ( i n y e a r s ) 5 0 Figure 4. Relation of age to attitude change in Study 3. Other Correlates of Susceptibility to Change In this study, none o f the demographic control variables were associated with attitude change (Bs = 0.17, 0.07, and - 0 . 0 1 for education, gender, and race, respectively). M e t a - A n a l y s i s o f S t u d i e s 1, 2, a n d 3 The pattern o f results across Studies 1, 2, and 3 appears quite consistent; in one case, however, a key aging effect was on the boundary o f significance. It seemed worthwhile, therefore, to assess the robustness o f the observed effects by meta-analyzing the data from all three studies. Doing so revealed that, when assessed simultaneously, the linear and quadratic effects o f age on attitude change susceptibility were significant (z = 3.69, d = 0.15, p < .001, and z = 3.13, d = 0.12, p < .001, respec- tively). More important, the linear and quadratic effects o f age on attitude change were homogeneous, meaning that they did not vary significantly across the three studies, X : ( 2 , N = 2,803) = 0.53, ns, and X2(2, N = 2,803) = 0.86, ns, respectively. S t u d y 4 Up to this point, we have focused on variation across the life span in resistance to attitude change. To better understand the meanings o f our findings, it is useful to recognize that resistance to change is one o f the four defining qualities o f strong attitudes (Krosnick & Petty, 1995). According to Krosnick and Petty, strong attitudes are those that ( a ) resist change, ( b ) persist over time, ( c ) influence thinking, and ( d ) motivate and guide behavior. Thus, i f one wants to understand why attitudes are most resistant to change in middle adulthood, it is useful to consider what causes attitude strength. Fortunately, a sizable literature now exists regarding the causal antecedents o f attitude strength. There is evidence, for example, that attitudes based on a great deal o f knowledge tend to be stronger than attitudes based on very little knowledge (Wood, Rhodes, & Biek, 1995). Similarly, attitudes about which people are highly certain tend to be stronger than attitudes about which they are less certain (Gross, Holtz, & Miller, 1995). Attitude importance is another antecedent o f attitude strength: Attitudes that people consider personally important tend to be more resistant to change, more stable over time, more likely to influence thinking, and more likely to direct behavior (Boninger, Krosnick, Berent, & Fabrigar, 1995). In fact, roughly a dozen such antecedents o f attitude strength have been identified in past work (for a review, see Petty & Krosnick, 1995). In light o f this literature, it seemed worthwhile to broaden the scope o f our investigation by exploring the relation o f age to such ante- cedents o f attitude strength. We begin by considering some o f the social and psychological processes by which various antecedents o f attitude strength may come to rise and fall over the adult l i f e span, thereby producing the observed age-related differences in susceptibility to change. We then briefly describe two other issues in the attitude strength literature that our next studies explored: the structure of attitude strength antecedents and the distinction between meta-attitudinal and operative indexes o f attitude strength and its antecedents. Underlying Mechanisms When young people enter adulthood, begin to vote in elec- tions, and otherwise take on active social responsibilities, they ATTITUDE STRENGTH 1397 are presumably inspired to learn about their places in the social world and the relevance o f public issues to their material inter- ests, reference groups or individuals, and values. As a result, people may come to attach increasingly more personal impor- tance to public issues ( s e e Boninger, Krosnick, & Berent, 1995). And this increase in personal importance may lead people to seek out information on these issues, leading to an accumulation o f relevant knowledge in memory (see, e.g., Boninger, Krosnick, Berent, & Fabrigar, 1995). Late in the life cycle, social and cognitive disengagement may occur, as people retire from the workforce and focus on new challenges o f their daily lives (e.g., transitions in places o f residence, emerging physical health prob- lems, and spousal caregiving). This increasing disconnection from the active workings o f social institutions may cause de- clines in the personal importance people attach to public issues. Simultaneously, declines in cognitive skills (e.g., Verhaeghen & Salthouse, 1997) may cause people to become less able to store and retrieve issue-relevant knowledge from memory. Thus, the personal importance o f public issues and knowledge about them may rise in early adulthood and fall in late adulthood, just as susceptibility to attitude change apparently does. Attitude importance and knowledge both reduce the likelihood o f attitude change, because they reinforce attitudes socially and cognitively (see, e.g., Boninger, Krosnick, Berent, & Fabrigar, 1995; Wood et al., 1995). Therefore, i f importance and knowledge surge just after early adulthood and decline late in life, they may be partly responsible for the surge and decline in resistance to attitude change. Structure o f Attitude Strength Antecedents Results o f this sort would also speak to another debated issue within the attitude strength literature: whether or not the various antecedents o f attitude strength (e.g., knowledge, certainty, and personal importance, among others) are manifestations o f a sin- gle, unitary underlying construct. For the most part, these ante- cedents are all positively associated with one another, and they are all positively associated with the four defining features o f strong attitudes (resistance to change, stability over time, and impact on thinking and on behavior). Because o f this, many attitude researchers have assumed, during the last five decades, that all o f these defining features and their antecedents are mani- festations o f a single underlying latent factor appropriately called attitude strength (for a review, see Krosnick, Boninger, Chuang, Berent, & Carnot, 1993). There has been empirical support for this single-construct view from some factor-analytic studies (e.g., Verplanken, 1989). However, other investigators have suggested that two underlying constructs are present (Bassili, 1996; Pomerantz, Chaiken, & Tordesillas, 1995) or have found evidence o f three distinct con- structs (Abelson, 1988; Lastovicka & Gardner, 1979), and still others have found evidence o f four (Erber, Hodges, & Wilson, 1995). Finally, some investigators have concluded that all o f the antecedents o f attitude strength are essentially independent o f each other and that no higher order attitude strength construct or constructs exist ( K r o s n i c k et al., 1993; Krosnick & Petty, 1995). Evidence that attitude importance and knowledge rise and fall together across the life span would be consistent with the single-factor view o f attitude strength antecedents, as well as with the midrange hypothesis that importance and knowledge both reflect a single underlying factor, even if other antecedents o f strength do not (Erber et al., 1995; Pomerantz et al., 1995). Meta-Attitudinal Versus Operative Indexes o f Attitude Strength This study offered an opportunity to explore another interest- ing question about the structure o f attitude strength as well. Bassili (1996) recently proposed a distinction between meta- attitudinal and operative indexes o f attitude strength ( s e e also Greenwald & Banaji, 1995, on indirect vs. direct measures). The former involve p e o p l e ' s perceptions o f their attitudes, whereas the latter describe the operation o f the attitudes more directly, unmediated b y perceptions. In our first three studies, we focused on the relation o f age to an operative index o f strength: actual rates o f attitude change. Here, we explored the relation o f age to the corresponding rneta-attitudinal index: peo- p l e ' s perceptions o f the susceptibility o f their attitudes to change. In short, we asked: Are people aware that they are especially susceptible to attitude change at the beginning and end o f the life cycle? Perceptions o f attitudinal qualities can at times be quite divergent from reality. For example, Krosnick et al. (1993) found surprisingly weak correlations between direct measures Of the volume o f attitude-relevant knowledge and p e o p l e ' s per- ceptions o f how much such knowledge they possessed ( m o r e generally, see Nisbett & Wilson, 1977). Just as some psycholo- gists have endorsed the increasing persistence hypothesis, the impressionable years hypothesis, or the life stages hypothesis, so may laypersons as well. This would reinforce the utility o f distinguishing between meta-attitudinal and operative measures in general and with regard to susceptibility to change in particu- lar. The data we analyzed for this study provided an opportunity to explore this question, and we did so. Method Sample In 1984, the National Opinion Research Center's General Social Sur- vey involved face-to-face interviews with 1,473 noninstitutionalized En- glish-speaking people 18 years of age or older living within the continen- tal United States. Measures Attitude importance, perceived attitude-relevant knowledge, and per- ceived likelihood of attitude change were assessed for six political issues: pornography, the death penalty, gun control, crime, law enforcement, and race relations, r° Importance. Respondents were asked how important each issue was to them, and the response options were one of the most important (coded 1 ), important (coded .66), not very important (coded .33), and not important at all (coded 0). The six importance scores were averaged to yield a single importance score for each respondent) 1 ro The wordings of all questions used for these analyses are shown in the cumulative General Social Survey codebook, which is available from the National Opinion Research Center. tl When the analyses reported in this study and the ones to follow were done separately for each issue individually, the same patterns emerged, although a bit more weakly. 1398 VISSER AND KROSNICK Perceived knowledge. Respondents were asked how much informa- tion they had about each issue, and the response options were all the information you need (coded t ), most of the information (coded .66), some information (coded .33), and very little information (coded 0). Perceived knowledge scores were averaged across the issues to yield a single perceived knowledge score for each respondenC 2 Perceived likelihood of attitude change. Respondents were asked how firm their opinion on each issue was, and the response options were very likely to change (coded 1 ), somewhat likely to change (coded .66), somewhat unlikely to change (coded .33), and very unlikely to change (coded 0). Likelihood of change scores were averaged across all six issues to yield a single score for each respondent. Demographic measures. Education, gender, and race were measured and coded in the same manner as in the previous studies. Results Relations of Age to Education, Gender, and Race A g e was significantly associated with education, gender, and race as expected. O l d e r age cohorts c o n t a i n e d f e w e r w e l l - e d u - cated p e o p l e ( r = - . 2 4 , p < .01, n = 1,465), f e w e r w o m e n ( r = - . 0 5 , p < .01, n = 1,467), and f e w e r racial minorities ( r = - . 0 5 , p < .01, n = 1,467) than did younger age cohorts. Relation Between Attitude Importance and Perceived Knowledge Consistent with p r e v i o u s studies (see, e.g., K r o s n i c k & A b e l - son, 1992; K r o s n i c k et al., 1993), the t w o attitude strength antecedents w e r e only w e a k l y correlated with o n e another ( r = .17, p < .01, n = 1,410). Thus, these aspects o f attitudes were largely independent. Relation of Age to Attitude Importance W h e n assessed alone, the linear e f f e c t o f age o n attitude i m p o r t a n c e was positive and significant ( B = 0.10, p < .05), indicating that i m p o r t a n c e increased with increasing age ( s e e c o l u m n 1 o f Table 2 ) . However, the quadratic effect o f age was significant and negative ( B = - 0 . 5 4 , p < .01 ), and the linear e f f e c t o f age was positive and significant when the quadratic e f f e c t was i n c l u d e d in the equation ( B = 0.56, p < .001; see c o l u m n 2 o f Table 2 ) . T h e c u b i c e f f e c t o f age was not significant ( B = - 0 . 4 5 , ns). A s F i g u r e 5 illustrates, attitude i m p o r t a n c e r o s e in early adulthood, up to age 50, and i m p o r t a n c e b e g a n to fall after about age 65. Thus, attitude i m p o r t a n c e was l o w e s t precisely when attitude change was most c o m m o n . A s c o l u m n 2 o f Table 2 illustrates, attitude i m p o r t a n c e was greater a m o n g better educated respondents than a m o n g less edu- cated respondents ( B = 0.12, p < .01), greater a m o n g w o m e n than a m o n g m e n ( B = - 0 . 1 2 , p < .001), and greater a m o n g n o n - W h i t e s than a m o n g Whites ( B = 0.10, p < .01). T h e s e relations again attest to the value o f controlling for these d e m o - graphics in exploring the relation o f age to attitude importance. Relation of Age to Perceived Knowledge The results o f regressions predicting perceived k n o w l e d g e w e r e quite similar to those i n v o l v i n g i m p o r t a n c e ( s e e Table 3 ) . T h e linear e f f e c t o f age was positive and significant in predicting perceived k n o w l e d g e ( B = 0.11, p < .05). W h e n the quadratic effect o f age was a d d e d to the equation, the linear e f f e c t o f age was positive and significant ( B = 0.93, p < .001), and the quadratic e f f e c t was significant and negative ( B = - 0 . 9 4 , p < .001 ). T h e c u b i c e f f e c t o f age was n o t significant ( B = - 0 . 0 3 , ns). F i g u r e 6 illustrates that p e r c e i v e d k n o w l e d g e rose early in the life c y c l e and fell after about age 55. Interestingly, the d e m o g r a p h i c correlates o f perceived k n o w l - edge were s o m e w h a t different f r o m those o f importance. Whereas w o m e n attached m o r e i m p o r t a n c e to these issues, m e n r e p o r t e d h a v i n g m o r e k n o w l e d g e about t h e m ( B = 0.18, p < .001 ). Whereas W h i t e s attached less i m p o r t a n c e to the issues than d i d n o n - W h i t e s , W h i t e s r e p o r t e d h a v i n g m o r e k n o w l e d g e about t h e m than did n o n - W h i t e s ( B = - 0 . 1 4 , p < .01). A n d whereas education was m o d e s t l y positively related to i m p o r - tance, the positive relation o f education to perceived k n o w l e d g e was much stronger ( B = 0 . 5 5 , p < .001 ). Thus, although i m p o r : tance and perceived k n o w l e d g e were positively correlate d with o n e another and had equivalent relations to age, they w e r e dis- tinct in terms o f their d e m o g r a p h i c correlates. Relation of Age to Perceived Likelihood of Change Perceptions o f likelihood o f attitude change w e r e partly accu- rate and partly inaccurate ( s e e Table 4 ) . A negative and signifi- cant linear e f f e c t o f age appeared ( B = - 0 . 4 4 , p < .001), m e a n i n g that perceived likelihood o f change d e c r e a s e d with age. T h e quadratic and c u b i c effects o f age were nonsignificant ( B = 0.16 and B = 0.66, r e s p e c t i v e l y ) , m e a n i n g that perceptions o f likelihood o f change d e c r e a s e d steadily across the life cycle. This is especially clearly illustrated in F i g u r e 7, w h i c h p r o v i d e s a graphic representation o f the r e g r e s s i o n equation f r o m c o l u m n 1 o f Table 4. Thus, p e o p l e w e r e correct in r e c o g n i z i n g that they were highly susceptible to attitude change during early adulthood, but they did not r e a l i z e that susceptibility increased late in life. Perceived likelihood o f attitude change was also distinct f r o m actual susceptibility in terms o f correlations with other d e m o - graphics. A meta-analysis o f the results o f Studies 1, 2, and 3 r e v e a l e d that less educated p e o p l e manifested significantly m o r e attitude change than m o r e educated p e o p l e ( d = - 0 . 0 9 , z = 2.26, p = .01 ), w o m e n manifested significantly m o r e attitude change than m e n ( d = - 0 . 0 8 , z = 2.22, p = . 0 1 ) , and the c o m b i n e d e f f e c t o f race o n susceptibility was not significant ( d = 0.01, z = 0.29, ns). In line with this last finding, r a c e was also not associated with perceived likelihood o f change ( B = - 0 . 0 5 , ns). However, m o r e e d u c a t e d respondents perceived a greater likelihood o f change than less educated respondents ( B = 0.10, p < . 0 5 ) , and m e n r e p o r t e d marginally significantly m o r e p e r c e i v e d likelihood o f attitude change than w o m e n ( B = 0.05, p < .10). Thus, perceived likelihood o f change did not m a t c h actual likelihood o f change in terms o f correlations with these t w o d e m o g r a p h i c variables, in addition to age. 12 Shifts over the life course in answers to this question could be due to changes either in the perceived volume of knowledge one possesses or in the volume of knowledge one thinks one needs. Regardless, how- ever, any shifts observed reflect changes in the gap between these two. This is clearly a meta-attitudlnal measure of knowledge, one presumably based at least in part on actual levels of knowledge (see Krosnick et al., 1993). ATI]TUDE STRENGTH 1 3 9 9 0 ~ 0 0 0 0 d d o d d ~ " a ~ ~ . . . . I o ~ o o o o " N o o o o o " I I ~ . o o o " . . . ~ m ~ o I t d d d d N ' I I * ~ o ~ o I • . 0 ,...~ • t ' N O ,.o ,-.., 8. v V o. v O v Discussion A l t h o u g h attitude i m p o r t a n c e and p e r c e i v e d k n o w l e d g e e x - hibited similar trajectories across the life cycle, they had distinct d e m o g r a p h i c correlates, suggesting that these antecedents o f attitude strength are not manifestations o f a c o m m o n underlying construct. Similarly, the meta-attitudinal m e a s u r e s o f attitude change in this study and the operative m e a s u r e s o f change u s e d in Studies 1, 2, and 3 s e e m n o t to h a v e tapped the same c o n - struct: T h e meta-attitudinal and operative m e a s u r e s had distinct d e m o g r a p h i c correlates, and they also e x h i b i t e d different pat- terns o f change across the life span. This suggests that the dis- tinction b e t w e e n meta-attitudinal and operative indexes o f atti- tude strength is a useful one. T h e distinction b e t w e e n the recta-attitudinal m e a s u r e o f atti- tude change and the attitude change m e a s u r e s used in Studies 1 and 2 is subtle but important. B e c a u s e these latter t w o studies r e l i e d o n self-reported attitude change rather than direct obser- vations o f change, o n e m i g h t t h i n k they w e r e recta-attitudinal indexes, n o t operative ones. Yet, the r e l a t i o n b e t w e e n age and susceptibility to change o b s e r v e d in Studies 1 and 2 was repli- c a t e d in Study 3 with a p u r e l y operative m e a s u r e o f change, and the e f f e c t there was h o m o g e n e o u s w i t h the effects in Studies 1 and 2. Therefore, when a n s w e r i n g the question " W o u l d y o u change y o u r m i n d i f . . . ? " respondents w e r e apparently de- scribing actual attitude change that o c c u r r e d at that m o m e n t rather than m o r e generic guesses about their likelihood o f change: R e i n f o r c i n g this conclusion, w e o b s e r v e d a different relation b e t w e e n age and a purely recta-attitudinal m e a s u r e o f attitude change. Study 5 To e x p l o r e the replicability o f the findings o f Study 4, w e turned to another set o f national survey data to assess the relation o f age to attitude i m p o r t a n c e using a different set o f political issues. M e t h o d Sample In 1984, the Center for Political Studies at the University of Michigan conducted face-to-face interviews with a nationally representative proba- bility sample of American adults for the National Election Study. During the weeks immediately preceding the 1984 U.S. presidential elec- tion, interviews were conducted with 2,257 people living in private households. Measures Attitude importance. Respondents were asked to express their atti- tudes on four political issues: government services, U.S. involvement in Central America, improving the economic status of women, and federally guaranteed jobs. 13 After each attitude question, respondents were asked how important it was to them that the federal government adhere to their 13 The wordings of all questions used for the analyses in this study, as well as those used in Study 6, are shown in the codebooks for the 1984 and 1996 American National Election Studies, which are available from the Inter-University Consortium for Political and Social Research at the University of Michigan. 2 . 7 2 2 . 7 0 _• 2 . 6 8 2 . 6 6 2 . 6 4 2 . 6 2 2 . 6 0 2 . 5 8 2 . 5 6 2 . 5 4 i 18 23 2 8 3 3 3 8 43 48 53 58 63 68 73 78 83 88 2 . 4 4 2 . 4 2 2 . 4 0 Ca 2 . 3 8 ~ 2.36 1 2 . 3 4 ~ , 2 . 3 2 ~ 2.30 . 2 . 2 8 ~ 2.28 0 . 2 . 2 4 2 . 2 2 2 . 2 0 . . . . . . . . . . . . . . . . 2 . 1 8 18 2 3 28 33 38 43 4 8 53 5 8 6 3 68 73 78 83 88 Age (in years) 1 4 0 0 VISSER AND KROSNICK Age (in years) Figure 5. Relation of attitude importance to age in Study 4. Figure 6. Relation of perceived knowledge to age in Study 4. preferred policy on that issue (extremely important, very important, somewhat important, or not important at all, coded t, .66, .33, and 0, respectively). 14 Answers to these questions were averaged to form an overall index of attitude importance. Demographic measures. Education, race, and gender were coded as they had been in previous studies. Results Relations of Age to Education, Gender, and Race A g e was significantly associated with education and r a c e as expected: O l d e r cohorts contained f e w e r w e l l - e d u c a t e d p e o p l e ( r = - . 2 9 , p < .01, n = 1,904) and f e w e r racial minorities ( r = - . 0 5 , p < .05, n = 1,902). A g e was not associated with gender ( r = - . 0 2 , ns, n = 1,902). Relation of Age to Attitude Importance T h e relation o f age to attitude i m p o r t a n c e here was c o m p a r a - ble to that seen in Study 4 ( s e e c o l u m n s 4 - 6 o f Table 2 ) . T h e linear e f f e c t o f age was significant and positive when tested b y itself ( B = 0.06, p < .05) and when tested along with a quadratic e f f e c t o f age ( B = 0.35, p < .001 ). T h e quadratic e f f e c t o f age Table 3 Unstandardized Regression Coefficients Predicting Perceived Attitude-Relevant Knowledge in Study 4 Predictor Regression 1 Regression 2 Regression 3 Age 0.11"* 0.93**** 0.92* Age squared -0.94**** - 0 . 9 0 Age cubed - 0 . 0 3 Gender 0.19"*** 0.18"*** 0.18"*** Race -0.13"** -0.14"** -0.14"** Education 0.57**** 0.55**** 0.55**** R 2 .069 .074 .074 Note. n = 1,462. * p < .10 (nmrginally significant). < .001. **p < .05. ***p < .01. ****p was also significant and negative ( B = - 0 . 3 6 , p < .01), and the c u b i c e f f e c t o f age was not significant ( B = 0.13, n s ) . A s F i g u r e 8 illustrates, i m p o r t a n c e r o s e during early adulthood, peaked at about age 45, and b e g a n to drop after about age 65. S t u d y 6 O u r final study m a d e use o f another national survey data set to e x a m i n e the relation o f age to another antecedent o f attitude strength: certainty. Attaching i m p o r t a n c e to a public issue, as w e l l as the extensive thought and i n f o r m a t i o n gathering that f o l l o w f r o m importance, m a y m a k e a person increasingly confi- dent in his o r her attitude. Thus, the d y n a m i c s o f attitude i m p o r - tance and perceived k n o w l e d g e d o c u m e n t e d in Studies 4 and 5 m a y p r o d u c e parallel changes in attitude certainty over the life cycle. I f this is true, certainty m a y b e partly responsible for the shifts in resistance to change o v e r the life span, b e c a u s e certainty enhances resistance to attitude change (see, e.g., Gross et al., 1995). A n d i f certainty manifests the same surge and d e c l i n e o v e r the life course as the other antecedents o f strength, this w o u l d r e i n f o r c e the one-factor v i e w o f attitude strength anteced- ents. Study 6 e x p l o r e d these possibilities and assessed the repli- cability o f the relation b e t w e e n age and attitude i m p o r t a n c e with yet another set o f political issues. Method Sample In 1996, the Center for Political Studies at the University of Michigan conducted face-to-face interviews with a nationally representative proba- bility sample of American adults for the National Election Study. During 14 The phrasing of this question was not optimal in that it addressed the federal government instead of simply asking about the personal importance of the issue (see Boninger, Krosnick, Berent, & Fabrigar, 1995). However, the fact that Study 5 replicated the findings of Studies 4 and 6 regarding attitude importance (in which the simpler importance questions were asked) reinforces confidence in this measure as tapping attitude importance. ATTITUDE STRENGTH 1401 Table 4 Unstandardized Regression Coefficients Predicting Perceived Likelihood o f Attitude Change in Study 4 0.46 - 0.36 Predictor Regression 1 Regression 2 Regression 3 ~ 0.44 Age -0.44**** -0.59*** -0.25 Age squared 0.16 - 0 . 7 5 ~- 0.42 Age cubed 0.66 I= Gender 0.05* 0.05 0.05* Race -0.05 - 0 . 0 5 - 0 . 0 5 "O Education 0.10" 0.10"* 0.09** ._~ 0.40 R 2 .028 .028 .029 0.38 Note. n = 1,461. * p < .10 (marginally significant). ** p < .05. *** p < .01. **** p < .001. the weeks immediately preceding the 1996 U.S. presidential election, 1,714 people living in private households were interviewed. Figure 8. M e a s u r e s , i t i i , r i i i i i i i i I 18 23 28 33 38 43 48 53 58 63 68 73 78 83 88 93 A g e ( i n y e a r s ) Relation of attitude importance to age in Study 5. 0 . 2 8 g) O) 0 . 2 7 Attitude importance. Respondents were asked to express their atti- tudes on five political issues: government spending on social services, defense spending, government assistance to Blacks, abortion, and envi- ronmental protection. After each attitude question, respondents were asked how important each issue was to them personally (extremely important, very important, somewhat important, not too important, or not important at all, coded 1, .75, .50, .25, and 0, respectively). Answers to these questions were averaged to form an overall index of attitude importance. Attitude certainty. For each of the same five issues, respondents were asked to express the certainty with which they held their attitude (very certain, pretty certain, or not very certain, coded 1, .50, and 0, respec- tively). Answers to these questions were averaged to create an overall index of attitude certainty. Demographic measures. Education, race, and gender were coded as they had been in previous studies. m c 0.26 0.25 0.24 0.23 O 0.22 0.21 • "; 0.20 0.19 0 . 1 8 i 0.17 0 . 1 6 0 . 1 5 , i i i i i ¢ ~ i i i , , i i i 18 23 28 33 38 43 48 53 58 63 68 73 78 83 88 A g e ( i n y e a r s ) Relation of perceived likelihood of attitude change to age in Figure 7. Study 4. Results Relations o f A g e to Education, Gender, a n d Race O l d e r cohorts c o n t a i n e d significantly f e w e r w e l l - e d u c a t e d p e o p l e ( r = - . 1 8 , p < .001, n = 1,709), and f e w e r racial minorities ( r = - . 0 7 , p < .01, n = 1,702). A g e was not associ- ated with gender ( r = - . 0 3 , ns, n = 1,702). Relation B e t w e e n Attitude Importance a n d Certainty T h e c o r r e l a t i o n b e t w e e n attitude i m p o r t a n c e and attitude cer- tainty was positive ( r = .41, p < .001, n = 1,710), b u t was far e n o u g h f r o m perfect to suggest that the t w o attributes are partly independent o f o n e another. Relation o f A g e to Attitude Importance T h e relation o f age to attitude i m p o r t a n c e here essentially replicated that seen in Studies 4 and 5 ( s e e c o l u m n s 7 and 8 o f Table 2 ) . A l t h o u g h the linear e f f e c t o f a g e alone was not sig- nificant ( B = 0.00, ns), the e f f e c t was positive and significant ( B = 0.29, p < .001 ) w h e n the quadratic e f f e c t was included, and the quadratic e f f e c t was negative and significant ( B = - 0 . 3 2 , p < .00I ). A l t h o u g h the c u b i c e f f e c t o f age was signifi- cant ( B = - 0 . 5 0 , p < .05), it did n o t notably alter the relation o f i m p o r t a n c e to age, and the effects o f age here w e r e n o t significantly different f r o m those o n i m p o r t a n c e that w e ob- served in Studies 4 and 5: age, X2(2, N = 5,219) = 0.75, n s ; age squared, X2(2, N = 5 , 2 1 9 ) = 1.50, ns; and age cubed, X2(2, N = 5,219) = 2.72, ns. ~ F i g u r e 9 p r o v i d e s a graphic representation o f the regression equation presented in c o l u m n 9 o f Table 2. A s F i g u r e 9 illustrates, i m p o r t a n c e again i n c r e a s e d across early adulthood, peaked at about age 50, and b e g a n to fall after about age 65. A l t h o u g h attitude i m p o r t a n c e was not as l o w a m o n g y o u n g adults here as it was in Studies 4 and 5, the s a m e basic trajectory o f attitude i m p o r t a n c e across the life c y c l e e m e r g e d in these data. E d u c a t i o n was n o t significantly related to attitude i m p o r t a n c e ( B = 0.02, ns). A s in the p r e v i o u s studies, n o n - W h i t e s attached 0.80 0.78 0.76 ~- 0.74 ~ 0 . 7 2 E 0 . 7 0 • ~ 0 . 6 8 • ~ 0 . 8 8 ~ 0 . 6 4 0 . 6 2 0 . 6 0 A g e ( i n y e a r s ) 0 . 7 7 0.75 ¢~, 0.73 " i 0.71 0.69 0 . 6 7 0.65 0.63 0.61 0.59 , , , g i J , i , i i i i i i J 18 23 28 33 38 43 48 53 58 63 68 73 78 83 88 93 m i i i , , c i i i i i , i i 18 23 28 33 38 43 48 53 58 63 68 73 78 83 88 93 A g e ( i n y e a r s ) Figure I0. Relation of attitude certainty to age in Study 6. Figure 9. Relation of attitude importance to age in Study 6. 1402 VISSER AND KROSNICK more importance to their attitudes on these issues than did Whites (B = 0.04, p < .001 ). And, again, there were no differ- ences between men and women in the importance they attached to these issues (B = 0.01, ns). Relation of Age to Certainty Consistent with the other strength antecedents, certainty in- creased early in the life cycle and declined late in the cycle (see Table 5 and Figure 10). The linear effect o f age alone was negative and significant (B = - 0 . 0 8 , p < .001). When the linear and quadratic age effects were estimated simultaneously, the linear effect was positive and significant (B = 0.23, p < .01), and the quadratic effect was negative and significant (B = - 0 . 3 6 , p < .001 ). The cubic effect o f age was not significant (B = 0.01, ns), The associations between certainty and the other demograph- ics were consistent with the associations between importance and those demographics. Certainty was unrelated to education (B = 0.00, ns), significantly greater among men than among women (B = 0.03, p < .01), and significantly greater among non-Whites than among Whites (B = 0.04, p < .01). Table 5 Unstandardized Regression Coefficients Predicting Attitude Certainty in Study 6 Predictor Regression 1 Regression 2 Regression 3 Age -0.08**** 0.23**** 0.24 Age squared -0.36**** -0.37 Age cubed 0.01 Gender 0.03**** 0.03**** 0.03*** Race 0.04*** 0.04*** 0.04*** Education 0.01 0.00 0.01 R 2 .021 .030 .030 Note. n = 1,696 * * * p < . 0 1 . ****p <.001. G e n e r a l D i s c u s s i o n Our findings have general implications regarding the litera- tures on attitude change and on the structure o f attitude strength antecedents. We discuss these implications in turn before con- sidering the limitations o f our evidence and directions for future studies. Attitude Change Aging and Susceptibility to Change Using methods that overcome various limitations o f past stud- ies, Studies 1, 2, and 3 support the life stages hypothesis and challenge the perpetual susceptibility hypothesis, the impres- sionable years hypothesis, and the increasing persistence hypoth- esis. Because considerable attitude change was evident even among the least susceptible age groups, our findings are more in line with the lifelong openness view than with the persistence hypothesis. It is interesting to note that our findings in this regard offer an extension o f some very old laboratory studies o f susceptibility to persuasion among preadults. For example, Whipple (1909) described a study by Stern ( 1 9 0 7 - 1 9 0 8 ) in which susceptibility to attitude change was found to be three times greater among 7 - 1 4 - y e a r - o l d s than among 16-19-year-olds. Similarly, Stukat (1958) found almost three times as much attitude change among 8-year-olds as among 14-year-olds, with a smooth decline in susceptibility between these ages. And Marple (1933) found that attitude change was greater among high school seniors than among college seniors, who in turn manifested more change than did older adults. In none o f these studies was representative sampling o f age groups done, so it is impossible to know whether the various age groups can legitimately be compared with one another. Furthermore, these studies did not offer oppor- tunities to differentiate among the increasing persistence, im- pressionable years, and life stages hypotheses, because they did not compare older adults o f different ages. But Stern ( 1 9 0 7 - 1908), Stukat (1958), and Marple (1933) nonetheless provided ATTITUDE STRENGTH 1403 evidence consistent with the claim that susceptibility to attitude change declines during preadulthood, a finding that resonates with our own evidence regarding the adult portion o f the life cycle. Reconciling the Present Findings With Other Evidence Although our finding of greater susceptibility to attitude change during early adulthood than during middle adulthood is consistent with most previous correlational research, our evi- dence o f greater susceptibility during late adulthood is inconsis- tent with most o f the existing literature. One possible explana- tion for this discrepancy is the methods o f analysis used. Virtu- ally without exception, previous investigations have divided samples into several discrete age groups and examined the aver- age level o f attitude stability for each group. This strategy re- quires researchers to lump together spans o f ages to achieve subsamples that are sufficiently large to yield reliable parameter estimates for each group. This is particularly necessary at the upper end o f the life cycle; survey samples include relatively small numbers o f very old adults, because o f mortality and other factors. As a result, the oldest age group in previous investigations has often spanned a wide range (e.g., ages 65 to 90 years) but has contained many more people at the younger end o f the range than at the older end. Averaging across these individuals may therefore have masked changes that occur at the very end o f the life cycle, because the oldest members o f the group were far outnumbered by the younger members o f the group. In the studies reported here, we took an analytic approach not yet used in any studies o f aging and attitude change. Rather than arbitrarily breaking the samples into subgroups, we used linear and nonlinear regression models to more precisely assess the relation between age and susceptibility to change across the entire adult life cycle. And, in fact, when we reanalyzed our data using the traditional approach, our results were notably different. For example, we broke the sample from Study 1 into five age categories ( 1 8 - 2 5 , 2 6 - 3 5 , 3 6 - 5 0 , 5 1 - 6 5 , and 6 6 - 9 7 years), and we gauged the average proportion o f respondents in each group who exhibited attitude change. Consistent with most pre- vious research, we found striking evidence o f increased suscep- tibility to change among young adults. Also consistent with this literature, we found no evidence o f an upturn in susceptibility at the end o f the life cycle: The mean rates o f attitude change for these age groups (adjusted for differences between groups in education, gender, and race) were .57, .49, .44, .45, and .43, respectively. Comparable results were obtained when we conducted similar analyses with the data from Studies 2 and 3. Thus, the methods o f analyses used seem likely to account for the discrepancy between our results and the results o f prior investigations regarding susceptibility to attitude change at the end o f the life cycle. One other difference between the current research and previ- ous investigations may also have contributed to the discrepancy. As we have discussed, earlier investigations used attitude stabil- ity as an index o f susceptibility to attitude change, which is likely to underestimate true susceptibility to change among peo- ple who experience few attitude-challenging experiences. If older adults were particularly unlikely to encounter these sorts o f experiences, attitude stability levels would paint a distorted portrait o f susceptibility to change at the end o f the life cycle. Thus, when the dependent variable is attitude stability rather than attitude change, even linear and nonlinear regression analy- ses may yield misleading evidence about the late adult years. Understanding the Relation Between Age and Susceptibility to Attitude Change Virtually without exception, previous investigations o f aging and attitude change have been entirely descriptive. The causal mechanism or mechanisms underlying the relation between age and susceptibility to attitude change have gone largely ignored. The present investigation provides an initial step toward under- standing causal underpinnings by documenting relations o f age to antecedents o f attitude strength. Specifically, we found that attitude importance, perceived knowledge, and certainty surged and declined roughly in parallel with changes in susceptibility to attitude change. Changes in these antecedents o f attitude strength may therefore provide clues about the causal mecha- nisms responsible for the dynamics o f susceptibility to attitude change. In particulaz; resistance to change may increase because peo- ple attach more importance to their attitudes, think more fre- quently and deeply about them as a result (see Boninger, Kros- nick, Berent, & Fabrigar, 1995), and consequently develop a confidence and sense o f knowledgeability that motivates resis- tance to change. And resistance to change may decline later because importance, perceived knowledge, and certainty drop. Future investigations directly testing these mediation hypotheses (by measuring susceptibility to change as well as importance, perceived knowledge, and certainty in the same representative sample survey) clearly seem to be warranted. In addition to antecedents o f attitude strength, a number o f other mechanisms may also be partly responsible for the relation o f age to susceptibility to attitude change. Our initial discussion o f the life stages hypothesis alluded to some o f these mecha- nisms. For example, early and late adulthood are both periods o f relatively frequent role transitions during which individuals are resocialized to new sets o f demands and expectations (Gla- ser & Strauss, 1971; Steckenrider & Cutler, 1989). Often, role adoption encourages certain attitudes and discourages others (Bogardus, 1927), and the adoption o f new roles can lead to changes in attitudes (e.g., Lieberman, 1950). The density o f role transitions in early and late adulthood may contribute to a general increase in susceptibility to attitude change as individu- als reconsider their relation to the social w o r d and redefine their social identities. Changes in social networks may also play a role in the relation between age and susceptibility to change. When discussing so- cial and political issues, people strategically select friends and discussion partners with whom they agree (Byrne, 1961; Huck- feldt & Sprague, 1987, 1988; Newcomb, 1961), and, as a result, discussions with such individuals reinforce their original atti- tudes. People also selectively misperceive the views o f their discussion partners when those views are at odds with their own (Huckfeldt & Sprague, 1987, 1988), further inflating people's perceptions o f social support for their attitudes. Thus, having large social networks with whom one comes into frequent con- tact may reduce susceptibility to change. 1404 VISSER AND KROSNICK And the sizes o f p e o p l e ' s social networks change over the life span. As people move from young adulthood to the middle adult years, they increasingly discuss important matters with a wide range o f people outside o f their families (Marsden, 1987). This trend reverses, however, at the end o f the life cycle: Relative to middle-aged adults, older adults have far fewer people with whom they discuss important matters (Burt, 1990; Lang & Cars- tensen, 1994; Marsden, 1987). Increased susceptibility to atti- tude change at the beginning and end o f the adult life cycle may be due in part to these changes in social support. A number o f other potential mechanisms may account for heightened susceptibility to attitude change during early adult- hood but not late adulthood. For example, because young adult- hood is typically the point o f entry into the political realm, attitudes are likely to be formed one at a time, in relative isola- tion from one another. But as a person ages, he or she is likely to come to see the interrelations among various attitudes and the applicability o f overarching principles such as ideologies or values. Consequently, among middle-aged adults, attitudes are more often embedded within overarching cognitive structures and are more strongly associated with other cognitive elements (Daniels & Clark-Daniels, 1995; Kirkpatrick, 1976). Such asso- ciations stabilize attitudes (Ostrom & Brock, 1969; Scott, 1968) and thus may partly account for the decline in susceptibility to change after early adulthood. Because young adults have had fewer life experiences, they may also have had fewer opportunities to behaviorally commit themselves to their attitudes (e.g., describing their attitudes to others or engaging in attitude-congruent behaviors). Behavioral commitment increases resistance to change (Brehm & Cohen, 1962; Hovland, Campbell, & Brock, 1957; Kiesler, 1971) and may partially account for the heightened susceptibility to atti- tude change at the beginning o f the adult life cycle. Still other processes may uniquely account for the decline in resistance to attitude change later in adulthood. For example, recent meta-analyses have revealed relatively large changes across the life span in cognitive functioning (Salthouse, 1992; Verhaeghen, Marcoen, & Goossens, 1993; Verhaeghen & Salt- house, 1997). Performance o f cognitive tasks declines beginning at approximately age 20, but the rate o f decline becomes espe- cially steep after age 50 (Salthouse, 1992, 1996; Verhaeghen et al., 1993). Thus, the oldest adults may find it hardest to generate effective counterarguments in the face o f persuasive appeals, increasing their susceptibility to attitude change~ Finally, late adulthood may see an increase in susceptibility to change because attitudes may become strikingly obsolete at that time. Young adults apparently form attitudes on a range o f political issues for the first time, and these attitudes remain through middle adulthood. Yet, during these middle adult years, the political world in which the attitudes were formed begins to change because o f technological and social innovations and the march o f time more generally. Thus, for example, people who came o f age during the Great Depression probably found that as the c o u n t r y ' s economic conditions improved, the atti- tudes they formed early in adulthood were less and less suitable to new world conditions. Because social reform and structural change occur slowly, dramatic discrepancies between o n e ' s early-formed attitudes and current world conditions are not likely to occur until late in adulthood. And this may make people especially open to change at that time. In addition, the very meaning o f some attitude objects may change over the several decades that divide early and late adult- hood. In the contemporary political context, for example, the term welfare calls to mind a certain set o f programs that assist a particular subset o f the population. In the 1950s, welfare referred to an entirely different set o f programs that assisted a different subset o f people. Likewise, support for civil rights in the early part o f the 20th century meant support for integration o f workplaces, neighborhoods, schools, and other social institu- tions. But in recent years, support for civil rights has meant support for redistribution o f economic resources and opportuni- ties to offset past inequalities (Schuman, Steeh, & Bobo, 1985). Thus, some people who formed attitudes toward civil rights decades ago may find that those attitudes are no longer applica- ble, therefore opening them up to change. And this should occur most among the oldest adults, whose attitudes were formed longest ago. A l l of these processes, summarized in Table 6, represent plau- sible mechanisms for the observed relation between age and susceptibility to attitude change, and each implies something different about which o f the myriad social, psychological, and biological aspects o f aging are responsible for changes over the life cycle in susceptibility to attitude change. We look forward to future research that investigates the role o f each o f them, along with other potential mechanisms. One possibility that we were able to test with the data from Study 2 turned out not to be a plausible mechanism o f the relation between age and susceptibility to change. That data set included a measure o f the number o f adults living in respon- dents' households, providing one index o f domestic social sup- Table 6 Potential Mechanisms of Decreasing and Increasing Susceptibility to Attitude Change During Early and Late Adulthood Early adulthood Late adulthood Increasing recognition of the relevance of public issues to one's material interests, values, and reference groups, resulting in increased attitude importance Acquisition of knowledge about the social and political world Increasing certainty with which attitudes are held Frequent role transitions Growing social networks Increasing attitude constraint Increasing opportunities for behavioral commitment to attitudes Decreasing attitude importance Decreasing retrievable knowledge Decreasing certainty with which attitudes are held Frequent role transitions Shrinking social networks Declining cognitive resources with which to generate effective counterarguments Increasing obsolescence of attitudes Changes over time in the meaning associated with particular attitude objects ATrlTUDE STRENGTH 1405 port. Linear and nonlinear ordinary least squares regression models revealed that the proportion o f respondents who lived alone increased sharply from young adulthood to middle adult- hood (from about 16% to about 30%), remained steady across the middle adult years (at about 30%), and rose steeply during the late adult years (to about 40% among 65-74-year-olds and about 50% among those 75 years old or older). Thus, social support decreased at roughly the same time that susceptibility to attitude change increased, raising the possibility that the for- mer may mediate the relation between age and susceptibility to change at the end o f the life cycle. However, this measure o f social support was unrelated to attitude change, both in the full sample (B = 0.00, ns) and among the subsample o f respondents 50 years old or older (B = 0.06, ns). Social support in the form o f living companions, therefore, cannot account for the age-related differences in sus- ceptibility to change observed in Study 2. This surprising result illustrates the importance o f formally testing the possible mech- anisms o f the relation between age and susceptibility to change. Other intuitively appealing mechanisms o f this relation, many o f which we outlined earlier, may also be rejected when sub- jected to empirical scrutiny. Societal Implications Our findings have interesting implications for social change across history, both in the past and in the future. To understand some o f these implications, it is useful to begin with Figure 11, which displays the distribution o f age in the American adult population starting in 1950 and projected out to 2050 (U.S. Bureau o f the Census, 1953, 1993). The width o f each band in this figure represents the proportion o f adults in each age group in each year. The consistent widening o f the solid band at the top represents the consistent growth of the 65-year-old and older age group, a process often referred to these days as the " g r a y - ing" of America. The hump in the band at the bottom between 1970 and 1980 represents the baby boom generation, which began to reach age 18 in the late 1960s. That hump is also apparent in each o f the older age groups, but it moves progres- sively to the right (representing the aging o f the baby boomers). If the perpetual susceptibility hypothesis had been true, these changes in the age distribution o f Americans would have no implications at all. And if the impressionable years hypothesis had been true, the growth o f the oldest age group would have had no implications for change in the susceptibility o f the Ameri- can public to attitude change across this 100-year period. But because the life stages hypothesis seems correct, the growing band o f older Americans signals a likely steady increase in the susceptibility o f the population to attitude change. Therefore, increased volatility in public opinion may be apparent in the future, especially during periods o f intense public debate (e.g., during presidential election campaigns). Because the 18-24-year-old group was disproportionately large between 1970 and 1980, and because these individuals were presumably highly susceptible to change, the malleability o f the aggregated public's views on political issues was probably a bit higher during those years than before or after. The national turmoil o f the late 1960s and early 1970s is no doubt attributable primarily to dramatic events such as the Vietnam War and Wa- tergate, but the reverberation o f such events throughout those years (and even today) may have been especially loud and sus- tained because an easily influenced age group was unusually large at that time. Impact o f Education, Gender, and Race on Susceptibility to Change Social psychologists have had long-standing interest in the relation o f other demographics to attitude change susceptibility, most notably gender (e.g., Coope~; 1979; Eagly, 1978; Mac- c o b y & Jacldin, 1974). In a meta-analysis o f 148 studies that explored gender differences in influenceability, Eagly and Carli (1981) found that women were significantly more open to change than were men. Consistent with Eagly and Carli's find- Figure 11. Age distribution of U.S. adults. 1406 VISSER AND KROSNICK ings, our own meta-analysis combining Studies 1, 2, and 3 revealed that women manifested attitude change significantly more than men. In line with ZaUer's (1990) findings, less educated people manifested significantly more attitude change than more edu- cated people (d = - 0 . 0 9 , z = 2.06, p = .04). Because education is strongly correlated with intelligence (Ceci, 1991), this finding may reflect the well-documented greater resistance to attitude change among more intelligent people when a persuasive mes- sage is easy to comprehend and equally exposed to everyone, as was the case here (e.g., Eagly & Warren, 1976; Rhodes & Wood, 1992). However, educational attainment is correlated with many other individual differences as well (see, e.g., Hyman, Wright, & Reed, 1975; Nie, Junn, & Stehlik-Barry, 1996). Thus, our finding encourages further study of the psychological sources of this relation. Given that the oldest age cohort was less educated and more predominantly female than the younger age cohorts, the relation of education to susceptibility would have been systematically distorted if we had not controlled for these characteristics. In contrast, the combined effect of race on susceptibility was not significant (d = 0.02, z = 0.32, p = .75), so controlling for this. variable here had no effect. But our documentation of the expected relation between race and age suggests that other inves- tigators of aging should be attentive to this aspect of changing cohort composition over the life course. Structure o f Attitude Strength One Construct or Many? To the extent that antecedents of attitude strength are all mani- festations of a single underlying construct, they should exhibit similar trajectories across the adult years. Whereas susceptibility to attitude change, attitude importance, and perceived knowl- edge manifested nearly identical relations with age, attitude cer- tainty increased less dramatically early in the life cycle and declined more slowly, beginning at an earlier age. More strik- ingly, different attitude attributes correlated differently with gen- der, race, and education. Averaging across Studies 4, 5, a n d 6 , attitude importance was significantly higher among women than among men ( d = 0.06, z = 3.43, p < .01), whereas men reported higher levels of knowledge, d = 0.24, t(1456) = 5.60, p < .01, and certainty, d = 0.14, t(1691) = 2.96,p < .01. Non-Whites reported higher attitude importance (averaging across Studies 4, 5, and 6; d = 0.14, z = 5.65, p < .01) and higher certainty than Whites, d = 0.13, t(1691) = 2.70, p < .01, but Whites reported having more knowledge than non-Whites, d = 0.i2, t(1456) = 2.83, p < .01. And more educated people reported higher attitude importance (averaging across Studies 4, 5, and 6; d = 0.07, z = 3.31, p < .01 ) and much higher knowledge, d = 0.43, t (1456) = 9.93, p < .01, but education was completely unrelated to certainty. Thus, no two of these attributes manifested the same relations with all three demographics. Taken together, then, this body of evidence challenges the single-construct view of attitude strength. In fact, the distinction between importance and perceived knowledge is inconsistent with even the modest suggestion that importance and perceived knowledge reflect a single latent factor (e.g., Erber et al., 1995; Pomerantz et al., 1995). It therefore seems wise, when studying attitude strength, to measure multiple dimensions, each with multiple items (to eliminate measurement error), and to assess the effects of each dimension individually, no matter what a factor analysis suggests about the potential to combine dimen- sions into a single meta-dimension. Meta-Attitudinal Versus Operative Indexes o f Strength Although attitude change susceptibility fell after early adult- hood and rose again late in life, people are apparently not aware of this, Rather, in Study 4, we observed a negative linear relation between age and perceived susceptibility to attitude change. Therefore, people were partly correct, in that they perceived susceptibility to drop after early adulthood, but they were incor- rect in their beliefs about the rest of the life cycle. Likewise, partial accuracy appeared when gauged by correlations between susceptibility to change and other demographics. Women cor- rectly recognized that they were more susceptible to attitude change than men, and people recognized that there was no robust relation of race to attitude change. However, highly educated people perceived themselves to be especially open to attitude change, whereas, in fact, they were unusually resistant to change. All of this reinforces the claim that meta-attitudinal measures of attitude properties can differ significantly from operative mea- sures of those same properties (see also Bassili, 1996; Krosnick et al., 1993). Thus, there seems to be theoretical value in further exploration of the origins and consequences of both meta-attitu- dinal and operative indexes, because the conclusion one reaches in any given investigation may hinge on which type of measure is used. Limitations Aging Effects Versus Birth Cohort Effects The age-related differences in attitude attributes documented here could be due to progress through the life cycle, as we have presumed, or they could be due to differences between birth cohorts in the nature of their socializing experiences (Mann- helm, 1952). That is, the young adults in our samples may manifest constant attitude attributes throughout their lives, and the older adults may manifest different but constant attributes throughout their lives. The differences we have documented may be attributable not to changes over the life course but to differ- ences in the early life experiences of these individuals. It is impossible to rule out this alternative explanation with the data we analyzed, because age and cohort effects are inevitably confounded. However, many longitudinal studies conducted over 30 years indicate that attitude change has been more common among young adults than among older adults, reinforcing the generalizability of the early life cycle pattern we saw (e.g., Alwin et al., 1991; Krosnick & Alwin, 1989). And even the relatively small number of attitude change data sets we analyzed were collected over a span of more than 10 years, long enough for members of the youngest cohort in the earliest data set to move into middle adulthood when the latest data set was col- lected. The replication across the three data sets provides at least tentative support for the generalizability of our effects across cohorts. As additional survey data are collected from ATTITUDE STRENGTH 1407 different birth cohorts at different points in their life cycles v i a methods such as those used in Studies 1, 2, and 3 here, it will be possible to further explore the generalizability o f our results regarding the life cycle. O b j e c t D o m a i n s Another useful direction for future research would be to ex- plore the relation between age and susceptibility to attitude change in other domains o f objects. Our studies and most in this literature have focused on political attitudes. Early adulthood is typically the period during which many individuals begin at- tending to and becoming behaviorally engaged in the political process, whereas people may become cognitively and socially disengaged from politics late in life. Some issues become espe- cially relevant to people as they reach the end o f the life cycle (e.g., government spending on social security), but most social and political issues probably seem increasingly irrelevant to people as they grow old. This dynamic may account for the life cycle shifts we observed. In contrast, attitudes toward other sorts o f objects may crys- tallize much earlier in life and may be no more susceptible to change during early adulthood than later in life. Food prefer- ences, for example, may be established during childhood and may remain relatively invulnerable to change throughout adult- hood. And some attitudes may remain central to p e o p l e ' s lives throughout old age, so they may become no more open to change late in the life cycle. Thus, the life stages hypothesis may be correct regarding only some attitudes. Alternatively, the crystallization o f any given attitude may not occur independently o f other attitudes. Instead, life experiences across a broad range o f domains may contribute to an increase in the stability o f all attitudes early in adulthood. That is, through the accumalation o f diverse experiences, individuals may arrive at a constellation o f attitudes and beliefs that capture a sense o f order and understanding o f the social environment. Once a consonant system o f attitudes and beliefs i s established, individ- uals may resist changes to any element within the system. Such resistance may be driven b y the motivation to avoid dissonance within the system or by a general sense o f mastery o f o n e ' s social environment and confidence in the validity of o n e ' s views. And late in life, as people have more restricted behavioral op- portunities and interact with more narrow groups o f people, all o f their preferences may weaken. Partly consistent with this latter view, several diverse lines o f research suggest that early adulthood is a period o f relative plasticity across many domains. For example, personality traits gradually increase in stability during young adulthood (Nessel- roade & Baltes, 1974) and become highly stable across the remaining adult years (Caspi, Bem, & Elder, 1989; I~osta, McCrae, & Arenberg, 1983; McCrae & Costa, 1994; Schuerger, Tait, & Tavernelli, 1982; Schuerger & Witt, 1989; Schuerger, Zarrella, & Hotz, 1989). Similarly, self-images exhibit consider- able instability during early adulthood but show much greater stability later in life (Mortimer, Finch, & Kumka, 1982). Many social identities, too, show this pattern o f stability. Identification with a political party, for example, tends to be substantially less stable during young adulthood than it is during middle adulthood ( A l w i n et al., 1991; A l w i n & Krosnick, 1991). Even antecedents o f attitude strength, such as attitude-relevant knowledge, are less stable in early adulthood than they are later in adulthood (Jennings, 1996). The findings o f each o f these lines o f inquiry, along with our own results, provide convergent evidence o f the malleability o f young adulthood. These additional indexes o f malleability suggest that the pattern we observed here regarding early adulthood may describe a wide range o f psychological phenomena. Much less attention has been devoted thus far to careful study o f these processes in representative samples o f adults late in the life cycle, and we look forward to such work. C o n c l u s i o n Aging is a complex, multifaceted process (Birren & Schaie, 1996). The passages from infancy to childhood, childhood to adolescence, adolescence to young adulthood, and across the adult years are marked by m y r i a d changes in mental, physical, and social processes. Beneath what may appear to be smooth, even linear transitions in thinking styles and behavior over the life cycle is undoubtedly rich and elaborate interplay o f these various processes. Therefore, the social and psychological mech- anisms underlying the effect o f aging we have highlighted may be complex and challenging to identify and to document. But the work we have reported here provides at least some i m p e t u s to initiate such inquiry in the future. R e f e r e n c e s Abelson, R. E (1988). Conviction. American Psychologist, 43, 267- 275. Alwin, D. E, Cohen, R. L., & Newcomb, T M. ( 1991 ). The women of Bennington: A study o f political orientations over the life span. Madi- son: University of Wisconsin Press. Alwin, D. E, & Krosnick, J. A, ( 1991 ). Aging, cohorts and the stability of socio-political orientations over the life-span. American Journal o f Sociology, 97, 169-195. Bassili, J. N. (1996). Meta-judgmental versus operative indexes of psy- chological attributes: The case of measures of attitude strengfla. Jour- nal o f Personality and Social Psychology, 71, 637-653. Berelson, B.R., Lazarsfeld, E E, & McPhee, W.N. (1954). Voting: A study of opinion formation in a presidential campaign. Chicago: University of Chicago Press. Birren, J. E., & Schaie, K. W. (1996). Handbook o f the psychology of aging. San Diego, CA: Academic Press. Bogardus, E. S. (1927). Personality and occupational attitudes. Socio- logical Social Research, 12, 73-79. Boninger, D. S., Krosnick, J. A., & Berent, M. K. (1995). Origins of attitude importance: Self-interest, social identification, and value rele- vance. Journal of Personality and Social Psychology, 68, 61-80. Boninger, D.S., Krosnick, J.A., Berent, M.K., & Fabrigar, L.R. (1995). The causes and consequences of attitude importance, In R. E. Petty & J.A. Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 159-189). Mahwah, NJ: Erlbaum. Brehm, J. W., & Cohen, A. R. (1962). Explorations in cognitive disso- nance. New York: Wiley. Brim, O.G., & Kagan, J. (1980). Constancy and change in human development. Cambridge, MA: Harvard University Press. Burt, R. S. (1990). Kinds of relations in American discussion networks. In C. Calhoun, M. W. Meyer, & W. R. Scott (Eds.), Structure ofpower and constraint: Papers in honor o f Peter M. Blau (pp. 45-67). New York: Cambridge University Press. Byme, D. (1961). Interpersonal attraction and attitude similarity. Jour- nal of Abnormal and Social Psychology, 62, 713-715. Campbell, J. P., Dunnette, M. D., Lawler, E. E., & Weick, K. E. (1970). 1 4 0 8 VISSER AND KROSNICK Managerial behavior, performance, and effectiveness. New York: McGraw-Hill. Carlsson, G., & Karlsson, K. (1970). Age, cohorts and the generation of generations. American Sociological Review, 35, 710-718. Caspi, A., Bern, D. J., & Eider, G. H., Jr. (1989). Continuities and conse- quences of interactional styles across the life course. Journal of Per- sonality, 57, 375-406. Ceci, S. J. ( 1991 ). How much does schooling influence general intelli- gence and its cognitive components? A reassessment of the evidence. Developmental Psychology, 27, 703-722. Centers for Disease Control and Prevention. (1995). Mortality pat- terns--The United States, 1992. Journal of the American Medical Association, 273, 100. Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum. Converse, P. E. (1976). The dynamics of party support: Cohort analyz- ing party identification. Cambridge, MA: Harvard University Press. Converse, P. E., & Markus, G. B. (1979). Plus ~a c h a n g e . . . : The new CPS election panel study. American Political Science Review, 73, 3 2 - 49. Conway, M., & Ross, M. (1984). Getting what you want by revising what you had. Journal of Personality and Social Psychology, 47, 738 -748. Cooper, H.M. (1979). Statistically combining independent studies: M_eta-analysis of sex differences in conformity research. Journal of Personality and Social Psychology, 37, 131-146. Costa, P. T., McCrae, R. R., & Arenberg, D. (1983). Recent longitudinal research on personality and aging. In K. W. Schaie (Ed.), Longitudinal studies of adult psychological development (pp. 222- 263 ). New York: Guilford Press. Crew, I., Sarlvik, C., & A l t , J. (1977). Partisan dealignment in Britain 1964-1974. British Journal of Political Science, 7, 129-190. Cutler, N.E., & Kanfrnan, R.L. (1975). Cohort changes in political attitudes: Tolerance and ideological nonconformity. Public Opinion Quarterly, 39, 69-81. Daniels, R. S., & Clark-Daniels, C. L. ( 1995, May). Age, conservatism, and sophistication: A reexamination. Paper presented at the annual meeting of the Midwest Political Science Association, Chicago, IL. Davies, J. C. (1965). The family's role in political socialization. Annals of the American Academy of Political and Social Science, 361, 1 0 - 19. Dawson, R. E., & Prewitt, K. (1969). Political socialization. Boston: Little, Brown. Eagly, A. H. (1978). Sex differences in influenceability. Psychological Bulletin, 85, 86-116. Eagly, A. H., & Carli, L. L. (1981). Sex of researchers and sex-typed communications as determinants of sex differences in influenceability: A recta-analysis of social influence studies. Psychological Bulletin, 90, 1-20. Eagly, A. H., & Warren, R. (1976). Intelligence, comprehension, and opinion change. Journal of Personality, 44, 226-242. Easton, D., & Dennis, J. (1969). Children and the political system: Origins of political legitimacy. New York: McGraw-Hill. Erber, M. W., Hodges, S. D., & Wilson, T. D. (1995). Attitude strength, attitude stability, and the effects of analyzing reasons. In R. E. Petty & J.A. Krosnick (Eds.), Attitude strength: Antecedents and conse- quences (pp. 433-454). Mahwah, NJ: Erlbanm. Gergen, K.J. (1980). The emerging crisis i n life-span developmental theory. In P. B. Baltes & O. G. Brim, Jr. (Eds.), Life-span development and behavior (pp. 6 0 - 8 3 ) . New York: Academic Press. Glaser, B., & Strauss, A. (1971). Theory ofstatas passage. Chicago: Aldine. Glenn, N. O. (1974). Aging and conservatism. Annals of the American Academy of Political and Social Science, 415, 176-186. Glenn, N. O. (1980). Values, attitudes, and beliefs. In O. G. Brim & J. Kagan (Eds.), Constancy and change in human development. Cam- bridge, MA: Harvard University Press. Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102, 4-27. Gross, S. R., Holtz, R., & Miller, N. (1995). Attitude certainty. In R. E. Petty & J.A. Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 215-245). Mahwah, NJ: Erlbaum. Herrmann, R. K., Tetlock, P.E., & Visser, P. S. (1997). Mass public decisions on going to war: A cognitive-interactionist framework. Un- published manuscript, Ohio State University, Columbus. Hess, R. D., & Tomey, J. V. (1967). The development of political atti- tudes in children. Chicago: Aldine. Hovland, C. I., Campbell, E. H., & Brock, T. C. (1957). The effects of "commitment" on opinion change following communication. In C. I. Hovland (Ed.), The order of presentation in persuasion (pp. 2 3 - 3 2 ) . New Haven, CT: Yale University Press. Huckfeldt, R., & Sprague, J. (1987). Networks in context: The social flow of political information. American Political Science Review, 81, 1197-1216. Huekfeldt, R., & Sprague, J. (1988). Choice, social structure, and politi- cal information: The informational coercion of minorities. American Journal of Political Science, 32, 467-482. Hyman, H. H., Wright, C. R., & Reed, J. S. (1975) ..The enduring effects of education. Chicago: University of Chicago Press. Jennings, M. K. (1996). Political knowledge over time and across gener- ations. Public Opinion Quarterly, 60, 228-252. Jennings, M. K,, & Markns, G. B. (1984). Partisan orientations over the long haul: Results from the three-wave political socialization panel study. American Political Science Review, 78, 1000-1018. Jennings, M. K., & Niemi, R.G. (1978). The persistence of political orientations: An overtime analysis of two generations. British Journal of Political Science, 8, 333-363. Jennings, M. K., & Niemi, R. G. (1981). Generations and politics: A panel study of young adults and their parents. Princeton, NJ: Princeton University Press. Johansson, C.B., & Campbell, D.P. (1971). Stability of the Strong Vocational Interest Blank for men. Journal of Applied Psychology, 55, 34-37. Kelman, H. C. (1958). Compliance, identification, and internalization: Three processes of attitude change. Journal of Conflict Resolution, 2, 51-60. Kiesler, C, A. (1971). The psychology of commitment: Experiments linking behavior to belief. San Diego, CA: Academic Press. Kirkpatrick, S. A. (1976). Aging effects and generational differences in social welfare attitude constraint in the mass public. Western Political Quarterly, 29, 43-58. Krosnick, J. A., & Abelson, R. P. (1992). The case for measuring atti- tude strength in surveys. In J. M. Tanur (Ed.), Questions about ques- tions: Inquiries into the cognitive bases of surveys (pp. 177-203). New York: Russell Sage Foundation. Kxosulck, J.A., & Alwin, D. E (1989). Aging and susceptibility to attitude change. Journal of Personality and Social Psychology, 57, 416-425. Krosnick, J. A., Boninger, D. S., Chuang, Y. C., Berent, M. K., & Carnot, C. G. (1993). Attitude strength: One construct or many related con- structs? Journal of Personality and Social Psychology, 65, 1132- 1151. Krosnick, J. A., & Petty, R. E. (1995). Attitude strength: An overview. In R. E, Petty & J. A. Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 1 - 2 4 ) . Mahwah, NJ: Erlbaum. Lang, E R., & Carstensen, L. L. (1994). Close emotional relationships in later life: Further support for proactive aging in the social domain. Psychology and Aging, 9, 315-324. Lastovicka, J. L., & Gardner, D. M. (1979). Components of involve- ment. In J.C. Maloney & B. Silverman (Eds.), Attitude research ATTITUDE STREblGTH 1 4 0 9 plays for high stakes (pp. 5 3 - 7 3 ) . Chicago: American Marketing Association. L e r n ~ R.L. (1984). On the nature of human plasticity. New York: Cambridge University Press. Lieberman, S. (1950). The effects o f changes in roles on the attitudes of role occupants. Human Relations, 9, 385-403. Lydon, J., Zanna, M.P., & Ross, M. (1988). Bolstering attitudes by autobiographical recall: Attitude persistence and selective memory. Personality and Social Psychology Bulletin, 14, 78-86. Maccoby, E. E., & Jacklin, C. N. (1974). The psychology of sex differ- ences. Palo Alto, CA: Stanford University Press. Mannheim, K. (1952). The problem of generations. In P. Kecskemeti ( Ed.), Essays on the sociology of knowledge (pp. 276-322). London: Routledge & Kegan Paul. Markides, K. S., & Black, S.A. (1996). Race, ethnicity, and aging: The impact of inequality. In R. H. Binstock & L. K. George (Eds.), Handbook of aging and the social sciences (4th ed., pp. 153-170). San Diego, CA: Academic Press. Markus, G. B. (1979). The political environment and the dynamics of public attitudes: A panel study. American Journal of Political Science, 23, 338-359. Marple, C. H. (1933). The comparative susceptibility of three age levels to the suggestion of group versus expert opinion. Journal of Social Psychology, 4, 176-186. Marsden, P. V. (1987). Core discussion networks of Americans. Ameri- can Sociological Review, 52, 122-131. McCrae, R. R., & Costa, P. T. (1994). The stability of personality: Ob- servation and evaluations. Current Directions in Psychological Sci- ence, 3, 173-175. Mortimer, J.T., Finch, M.D., & Kumka, D. (1982). Persistence and change in development: The multidimensional self-concept. In P. B. Baltes & O. G. Brim, Jr. (Eds.), Life-span development and behavior (Vol. 4, pp. 263-312). New York: Academic Press. Mueller, J. E. (1973). Wars, presidents, and public opinion. New York: Wiley. Neijens, P. (1987). The Choice Questionnaire. Amsterdam: Free Univer- sity Press. Nesselroade, J. R., & Baltcs, P. B. (1974). Adolescent personality devel- opment and historical change: 1970-1972. Monographs of the Society for Research in Child Development, 39( 1, Serial No. 154). Newcomh, T M. ( 1961 ). The acquaintance process. New York: Holt, Rinehart & Winston. Newcomb, T. M., Koenig, K. E., Hacks, R., & Warwick, D. P. (1967). Persistence and change: Bennington College after 25 years. New York: Wiley. Nie, N. H., Junn, J., & Stehiik-Barry, K. (1996). Education and demo- cratic citizenship in America. Chicago: University of Chicago Press. Nisbett, R. E., & Wilson, T D. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review, 84, 2 3 1 - 259. Nurm, C. Z., Crockett, H. J., & Williams, J. A. (1978). Toleranoefor nonconformity. San Francisco: Jossey-Bass. Ostrom, T. M., & Brock, T. C. (1969). A cognitive model of attitudinal involvement. In R. P. Abelson, E. Aronson, W. J. McGulre, T. M. New- comb, M. J. Rosenberg, & P. H. Tannenbanm (Eds.), Theories of cog- nitive consistency: A sourcebook (pp. 373-383). Chicago: Rand McNally. Petty, R. E., & Krosnick, J. A. (Eds.). (1995). Attitude strength: Ante- cedents and consequences. Mahwah, NJ: Erlbaum. Pomeramz, E.M., Chaiken, S., & Tordesillas, R.S. (1995). Attitude strength and resistance processes. Journal of Personality and Social Psychology, 69, 408-419. Reifman, A., Klein, J. G., & Murphy, S. T. (1989). Self monitoring and age. Psychology and Aging, 4, 245-246. Rhodes, N., & Wood, W. (1992). Self-esteem and intelligence affect influenceability: The mediating role of message reception. Psychologi- cal Bulletin, 111, 156-171. Ross, M., McFarland, C., & Fletcher, G. LO. (1981). The effect of attitude on the recall of personal histories. Journal of Personality and- Social Psychology, 40, 627-634. Ryde~ N.B. (1965). The cohort as a concept in the study of social change. American Sociological Review, 30, 843-861. Salthouse, T.A. (1992). Reasoning and spatial abifities. In E I. M. Cralk & T A. Salthouse (Eds,), Handbook of aging and cognition (pp. 167-211). Hillsdale, NJ: Erlbaum. Salthouse, T. A. (1996). The processing-speed theory of adult age differ- ences in cognition. Psychological Review, 103, 403-428. Schuerger, J. M., Tait, E., & Tavemelli, M. (1982). Temporal stability of personality by questionnaire. Journal of Personality and Social Psychology, 43, 176-182. Schuerg~ J.'M., & Witt, A. C. (1989). The temporal stability of individ- ually tested intelligence. Journal of Clinical Psychology, 45, 2 9 4 - 302. Schuerger, J.M., Zarrella, K: L., & Hotz, A. S. (1989). Factors that influence the temporal stability of personality by questionnaire. Jour- nal of Personality and Social Psychology, 56, 777-783. Schuman, H., Steeh, C., & Bobo, L. ( 1985 ). Racial attitudes in America: Trends and interpretations. Cambridge, MA: Harvard University Press. Scott, W. A. (1968). Structure of natural cognitions. Journal of Person- ality and Social Psychology, 12, 261-278. Sears, D. O. (1975). Political socialization. In E I. Greenstein & N. W. Polsby (Eds.), Handbook of political science (Vol. 2, pp. 93-153). Reading, MA: Addison-Wesley. Sears, D. O. (1981). Life stage effects on attitude change, especially among the elderly. In S. B. Kiesler, J. N. Morgan, & V. K. Oppenhei- mer (Eds.), Aging: Social change (pp. 183-204). New York: Aca- demic Press. Sears, D. O. (1983). The persistence of early political predispositions: The role of attitude object and life stage. In L. Wheeler (Ed.), Review of personality and social psychology (Vol. 4, pp. 7 9 - 1 1 6 ) . Beverly Hills, CA: Sage. Sears, D. O., & Funk, C. L. ( 1996, October). Thepersistenoe of political attitudes through the life span: The Terman longitudinal study. Paper presented at the annual meeting of the Society for Experimental Social Psychology, Old Sturbridge, MA. Sears, D. O., & Gahart, M. T ( 1980, September). The stability of racial prejudice and other symbolic attitudes. Paper presented at the 88th Annual Convention of the American Psychological Association, Mon- treal, Quebec, Canada. Sniderman, P. M,, & Grob, D. B. (1996). Innovations in experimental design in attitude surveys. Annual Review of Sociology, 22, 377-399. Sniderman, P. M., & Piazza, T L. (1993). The scar of race. Cambridge, MA: Belknap Press of Harvard University Press. Steckenrider, L S., & Cutler, N.E. (1989). Aging and adult political socialization: The importance of roles and role transitions. In R. S. Sigel (Ed.), Political learning in adulthood (pp. 5 6 - 8 8 ) . Chicago: University of Chicago Press. Stem, L. W (1907-1908). Literatur zur psych, d. aussage. Zeits. E Angw. Psych, 1, 429-450. Strate, J.M., Parrish, C . L , Elder, C.D., & Ford, C. (1989). Life span civic development and voting participation. American Political Sci- ence Review, 83, 443-464. Stukat, K. G. (1958). Suggestibility: A factorial and experimental anal- ysis. Stockholm: Almquist & Wiksell. Tesch, S.A. (1983). Review of friendship development over the life span. Human Development, 26, 266-276. Tesser, A., Martin, L., & Mendulia, M. (1995). The impact of thought on attitude extremity and attitude-behavior consistency. In R.E. Petty & J.A. Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 7 3 - 9 2 ) . Mahwah, NJ: Erlbaum. 1410 VISSER AND KROSNICK Tyler, T R., & Schuller, R. A. (1991). Aging and attitude change. Jour- nal of Personality and Social Psychology, 61, 689-697. U.S. Bureau of the Census. (1953). Current population reports (Series P-20, No. 45). Washington, DC: U.S. Government Printing Office. U.S. Bureau of the Census. (1993). Current population reports (Series P-20, No. 474). Washington, DC: U.S. Government Printing Office. Verhaeghen, P., Marcoen, A., & Goossens, L. (1993). Facts and fiction about memory and aging: A quantitative integration of research find- ings. Journal of Gerontology: Psychological Sciences, 4& P157- P171. Verhaeghen, P., & Salthouse, T A. (1997). Meta-analyses of age-cogni- tion relations in adulthood: Estimates of linear and nonlinear age effects and structural models. Psychological Bulletin, 122, 231-249. Verplanken, B. (1989). Involvement and need for cognition as modera- tors of beliefs-attitude-intention consistency. British Journal o f So- cial Psychology, 2& 115-122. Whipple, G. M. (1909). The observer as reporter: A survey of the 'psy- chology of testimony.' Psychological Bulletin, 6, 153-170. Wood, W., Rhodes, N., & Biek, M. (1995). Working knowledge and attitude strength: An information-processi'ng analysis. In R. E. Petty & J.A. Krosnick (Eds.), Attitude strength: Antecedents and conse- quences (pp. 283-313). Mahwah, NJ: Erlbanm. Zaller, J. R. (1990). Political awareness, elite opinion leadership, and the mass survey response. Social Cognition, 8, 125-153. Received January 5, 1998 Revision received July 12, 1998 Accepted July 22, 1998 • Call for N o m i n a t i o n s The Publications and Communications (P&C) Board has opened nominations for the editorships of the J o u r n a l of Abnormal Psychology, J o u r n a l of Comparative Psychology, J o u r n a l of Exl~rimental Psychology: Learning, Memory, and Cognition, Journal of Personality and Social Psychology: Attitudes and Social Cognition,Professional Psychology: Research and P r a c t i c e , Psychological Review, and Psychology, Public Policy, a n d L a w for the years 2001-2006. Milton E. Strauss, Phi3; Charles T. Snowdon, Phi); James H. Neely, Phi); Arie W. Kruglanski, PhD; Patrick H. DeLeon,.PhD, JD; Robert A. Bjork, PhD; and Bruce D. Sales, JD, PhD, respectively, are the incumbent editors. Candidates should be members o f APA and should be available to start receiving manuscripts in early 2000 to prepare for issues published in 2001. Please note that the P&C Board encourages participation by members o f underrepresented groups in the publication process and would particularly welcome such nominees. Self-nominations are also encouraged. To nominate candidates, prepare a statement of one page or less in support of each candidate. Send nominations to the attention of the appropriate search c h a i r - - • David L. Rosenhan, PhD, for J o u r n a l o f A b n o r m a l Psychology Lauren B. Resnick, Phi), for J o u r n a l of Comparative Psychology Joe L. Martinez, Jr., PhD, for j E P : Learning, Memory, and Cognition Sara B. Kiesler, PhD, for J P S P : Attitudes a n d Social Cognition Judith P. Worell, PhD, for Professional Psychology: Research and P r a c t i c e Lyle E. Bourne, Jr., Phi), for Psychological Review Lucia A. Gilbert, PhD, for Psychology, Public Policy, a n d Law - - t o the following address: c/o Karen Sellman, P&C Board Search Liaison Room 2004 American Psychological Association 750 First Street, NE Washington, DC 20002-4242 The first review of nominations will begin December 7, 1998.