102 Are Citizens Pliable?
psychology as we overview theories of attitude formation and change that have greatly influenced how political scientists study public opinion.
Are Americans' Attitudes Stable?
One way to assess whether people's attitudes remain the same over time is to survey people at one time about their political opinions and then ask them about the same opinions later. Recall from Chapter 2 that this approach, known as a panel study, was the one used by Kent Jennings and Richard Niemi to assess attitudes among high school seniors and their parents at various ages for both groups. Across the years, numerous panel studies have been conducted as part of the American National Election Study (ANES) scries, allowing researchers to study attitude stability and change among the general public. ANES surveys have been carried out at least every two years, coinciding with presidential and congressional elections, since 1948.3 Although in most years the ANES is a cross-sectional study—a survey using a new representative sample of adults—periodically since 1948 panel studies have been conducted whereby previous respondents arc rcin-terviewed.
Individual Attitude Change
One of the very first empirical analyses of political attitude stability was conducted by Philip Converse using data from the ANES 1956,1958, and i960 panel study.4 This study questioned respondents about salient political issues of the day in each of the three years. These issues included domestic and foreign policy issues such as school desegregation, federal aid to education, the creation of a fair employment practices commission to prevent racial discrimination in employment, and military aid to fight communism. Respondents were also asked their party identification. Comparisons of respondents' political attitudes in 1958 with their attitudes in i960 are presented in Table 4-1. To measure the degree to which individuals' attitudes were stable, Converse calculated tau-b correlation coefficients, which are presented in the first column of Table 4-1. When tau-b equals 1.0, everyone's attitude was the same in i960 as it had been in 1958. The smaller the value of tau-b, the more people's attitudes fluctuated over these two years. Another way to compare attitude stability over time is to determine how many people kept the same opinion at two times. We present our results from such an analysis in the second column of the table. The figures here are the percentage of respondents whose opinions stayed on the same side of an issue or whose party allegiance was to the same party (or who were politically independent) from 1958 to i960. Political attitudes are considered stable by this measure even if someone changed from strong to weak agreement or from strong to weak partisanship over time. The final column of the table presents the percentage of respondents who expressed no opinion toward the policy issues in either 1958 or i960, or in both years.
Attitude Stability and Attitude Change 103 Table 4-1   Stability of Individual Political Altitudes from 1958 to i960
Political attitude	Correlation' between attitudes in 1958 and i960 (tau-b)	Percentage holding the same attitude in 1958 and i960	Percentage having no opinion in at least one year
Party identification	.73	85.7	N/A
School desegregation	.43	57.5	15.8
Employment discrimination	.41	60.0	17.9
Guaranteed employment	.41	56.5	15.2
Isolationism	.39	59.6	16.3
Fedetal aid to education	.38	57.2	13.8
Foreign economic aid	.34	48.0	21.4
Foreign military aid	.32	56.7	24.9
Federal housing	.29	40.7	33.9
Sources; Philip E. Converse, "The Nature of Belief Systems in Mass Publics," in Ideology and Discontent, ed. David E. Apter (New York: Free Press, 1964), 240; Analysis of American National Election Studies 1956-58-60 Panel Data File.
As we see from the results in 'Table 4-1, party identification was the most stable political attitude over these two years. In fact, the value of tau-b for parry identification is significantly higher than the tau-b for any of the other attitudes. For most people, their party affiliation remained the same from 1958 to i960. Converse attributes this to the fact that attitudes and affect toward groups, such as political parties, help to organize the political opinions of many people. Because party identification was the only attiLude surveyed "rhat touches on pure affect toward a visible population grouping," it came as no surprise to Converse that these attitudes remained quite stable over two years.5
In conrrast, attitude instability was much more common for the policy issues, with between 40 and 60 percent of the public holding the same opinion over time, versus nearly 86 percenr for party allegiance. Among the policy issues queried, citizens' attitudes toward racial employment discrimination, isolationist foreign policy, and school desegregation were the least likely to change. Opinions about foreign aid and whether the federal government should be involved in providing housing and electric power were especially likely to fluctuate. To explain why there were these differences in attitude stability across issues, Converse points to the fact that some issues, even domestic ones, were less significanr to citizens in the late 1950s.6 Stable attitudes are more likely toward objects that tend to influence people's everyday lives (such as jobs or schools) or toward salient groups (such as blacks). When issues are more remote, attitude instability is more common.
104 Are Citizens Pliable?
People's political attitudes could change for a number of reasons. Citizens' opinions are susceptible to change, for instance, in the face of real-world events. Current events can provide new information to citizens as well as perhaps new understandings of issues, both of which can produce changes in attitudes. Converse considered this possibility but ruled it out as an explanation for the attitude instability he observed. In the late 1950s, high-profile news events did occur that could have influenced attitudes toward some issues that Converse studied, but not all. The public witnessed standoffs and violence in school desegregation cases in the southern United States, such as the integration of Central High School in Little Rock, Arkansas, in 1957 when army soldiers were required to escort black students into the formerly all-white high school. News also broke of waste in foreign aid spending during these years. Yet for one of the issues where the public's attitudes were the least stable, the role of the federal government in providing housing, there were no changes in federal policy or other relevant newsworthy items during these years that might have produced public attitude change. Converse also witnessed that the average correlations between people's attitudes were the same, and were similarly low, between 1956 and T958, between 1958 and i960, and between 1956 and i960. We might expect that attitudes would fluctuate more the longer the interval between measuring the attitudes. This was not the case, leading Converse to turn to explanations for instability other than responses to changing events.
Ultimately, Converse concludes that many citizens' policy opinions are meaningless and might be better characterized as rtonattitudes than attitudes.7 Converse argues that "large portions of an electorate do not have meaningful beliefs, even on issues that have formed the basis for intense political controversy among elites for substantial periods of time."8 If public attitudes were well developed, carefully considered, and based on detailed information, according to Converse, they probably would not change as much as he had observed. Converse's statements might seem extreme and unnecessarily harsh, especially when you consider that, according to Table 4-1, a majority of the public did continue to hold opinions on the same side of most issues between 1958 and i960. Examining temporal attitude stability was only one portion of Converse's research, however, and his conclusions were based on a variety of other analyses. In particular, Converse demonstrated that most people's opinions are not well grounded in broader beliefs (such as liberalism or conservatism) and are not well organized.9 We discuss these other conclusions more thoroughly in Chapter 5.
Yet, even thinking only about Converse's evidence of attitude change, it is fair to ask how many Americans should have stable attitudes for us to conclude that the public's policy opinions are meaningful. Is our standard 100 percent? If so, then any deviation from this might lead to conclusions similar to Converse's. In fact, it was empirical evidence such as this that fostered the development of the
Attitude Stability and Attitude Change 105
theories of democratic elitism and pluralism. Recall from Chapter r that these theorists compared empirical research about the public's capabilities with classical democratic theorists' expectations about the public and concluded that the public was not living up to the classical model. By using a different standard, however, perhaps one that recognizes some policy issues arc complex and are not central to most people's daily lives, we may conclude that levels of attitude stability apparent in the late 1950s seem reasonable and not too low.
Collective Attitude Stability
We turn now to an examination of the stability of collective public opinion, that is, the aggregate political opinions of the public. In the most detailed analysis of collective opinion over time, Benjamin Page and Robert Shapiro analyzed results from public opinion surveys conducted over fifty years. Page and Shapiro's conclusions paint a very different picture than that emerging from studies of individuals. In their own words, "the American public, as a collectivity, holds a number of real, stable, and sensible opinions about public policy and . . . these opinions develop and change in a reasonable fashion, responding to changing circumstances and to new information."'0
Page and Shapiro's evidence for this conclusion consists primarily of comparisons of responses to survey questions that were asked at least two different times between 1935 and 1990. For ;8 percent of these questions, aggregate public opinion did not change significantly over time. For instance, in 1942, approximately 75 percent of the American public believed that the United States should have an active rather than isolationist foreign policy. Fourteen years later, the percentage of Americans holding this attitude was also about 75 percent. Opinions on this topic did fluctuate a bit in the intervening years, but within a narrow range of 72-80 percent supporting activism over isolationism.
Turning to domestic matters, Page and Shapiro observed that opinions about government spending on a variety of programs did not fluctuate very much during the 1970s and 1980s. Consistently high percentages of the public (68-77 percent) felt that the government was spending too little to fight crime, while consistently low percentages (12-25 percent) felt that government was spending too little on welfare (see Figure 4-t). Attitudes toward government spending on Social Security were especially stable during the late ry8os (the years for which opinion data are available). The item displayed in Figure 4-1 that shows the most variation was beliefs about education spending. Whereas about 49 percent of the public felt the government spent too little on education in the early 1970s, this percentage gradually increased during the time period, reaching 68 percent in 1989.
While Page and Shapiro found many examples of collective attitude stability, they also uncovered some instances of attitude change. Unlike what we might