Attitudinal Stability on Short- and Long-term Issues

Goal

The goal of this exercise is to explore the differences in the stability of attitudes about long term and short term issues. Correlation coefficients, comparisons of correlations, and T-statistics will be used./p>

Concept

Attitudinal stability refers to consistency in a single attitude over time. For example, if a survey respondent said that he preferred lower taxes and fewer government sponsored social programs during the first interview and then said that he preferred higher taxes and more government sponsored social programs during the second interview, one could say that the respondent's attitudes on this topic are unstable. If, on the other hand, the respondent gave the same answer to the question both times, one could say that his attitudes on this topic are stable.

Political science literature suggests that attitudinal stability in an individual indicates that he or she has a deep and meaningful opinion regarding the given issue. That is, the person's opinion is important and makes sense to them. Stability suggests that the person in question has mulled over the topic, thought thoroughly about it, and has reached a somewhat definitive conclusion on the issue.

Attitudinal instability, on the other hand, can be interpreted in one of two ways. First, it may be that the respondent in question has gained an important piece of information between the first and second interviews that has caused an attitude to change. In this case, instability would indicate an important and meaningful change in an attitude. Second, it may be that the respondent has a "non-attitude" on the issue. A non-attitude suggests that a respondent does not hold the issue as personally important and that he or she has not put it a great deal of thought into generating an opinion.

Long term issues are those that have been central to politics and campaigns for years or even decades. One might consider federal tax policies a long term issue, for example, because they have been central to every election and every campaign platform in recent memory. Political science research suggests that, due to their extended and continuing importance, people are able to form "gut" reactions to these long term issues. Political parties and candidates base a significant portion of their platforms around these types of issues, and political elites are constantly reminding voters of how their party or candidate stands. This constant reinforcement results in attitudinal stability.

Short term issues, as you may have guessed, are issues that have been important to politics for a relatively short period of time. These issues are not engrained into political debate, so they are not engrained into individuals' belief systems. Consistency on these issues requires one to think deeply and form meaningful conclusions without much help from political parties and elites. In short, these issues are cognitively complex, and the meaningful attitudes about them require more work by the respondent. Attitudes on short term issues are often less stable than attitudes on long term issues.

Examples of possible research questions about attitudinal stability include:

  • How does attitudinal stability on short term issues compare to attitudinal stability on long term issues?
  • What other factors contribute to attitudinal stability?
  • What types of issues are particularly stable? Which issues are less stable?

Data for this exercise come from the American National Election Study, 2004: Pre- and Post-Election Survey. The American National Election Studies (ANES) grew out of the Survey Research Center and the Center for Political Studies of the Institute for Social Research at the University of Michigan. These organizations, together, have been covering elections since 1948. ANES conducts surveys of the American electorate in election years. ANES produces high quality data on voting, public opinion, and political participation to serve the research needs of social scientists, teachers, students, policy makers and journalists who want to better understand the theoretical and empirical foundations of national election outcomes.

Respondents in the 2004 ANES were interviewed before and after the November election. Questions asked of respondents during the survey cover a broad range of topics including demographics, attitudes toward candidates and parties, attitudes on different segments of the American public, attitudes on foreign policy matters, and political behavior. The ANES uses random sampling in order to produce the most representative data possible on the American electorate.

This exercise will use the following variables:

  • Liberal/Conservative Pre-election (V043085)
  • Liberal/Conservative Post-election (V045117)
  • Government Spending and Services Pre-election (V043136)
  • Government Spending and Services Post-election (V045121)
  • Interventionism by Diplomacy/Military Pre-election (V043107)
  • Interventionism by Diplomacy/Military Post-election (V045124)
  • Education (V043254)

In this exercise, you will use correlations to measure the consistency of respondents' answers to questions before and after the 2004 election. A correlation is a measure of the strength of association between two variables. It is used to predict the value of one variable given the value of the other. Correlations are expressed on a scale from -1.0 to +1.0, with the strongest correlations found at both extremes of the scale. Correlation coefficients closer to zero indicate a weak relationship between variables. The closer the correlation coefficient is to +1.0, the more stable responses are between the pre-election survey and the post-election survey.

Ideological Identification

In both waves of data collection, respondents were asked to place themselves on a seven-point scale, where 1 equals "extremely liberal," two equals "liberal," three equals "slightly liberal," four equals "moderate; middle of the road," five equals "slightly conservative," six equals "conservative," and seven equals "extremely conservative." Look at the stability of ideology between the pre-election survey and the post-election survey. How stable do the ideology responses seem to be? Is this the result you would expect for ideological identification?

Government Spending and Services

Respondents were asked their opinion about government spending and services using a seven-point scale, where one equals "government should provide many fewer services" and seven equals "government should provide many more services." Consider the stability in attitudes toward government spending and services between the pre-election survey and the post-election survey. Are attitudes toward government spending and services stable between the pre-election survey and the post election survey? Is this what you expected, given the issue's centrality to elections?

Interventionism

To measure attitudes toward government intervention into foreign affairs, respondents gave their attitudes on a seven-point scale, ranging from "should solve with diplomacy" (1) to "must be ready to use military force" (7). Look at the stability of opinions on government interventionism between the pre- and post-election surveys. How stable are the opinions on government interventionism? Is this what you expected given the issue's novelty in the political arena?

Education and Attitudinal Stability

To find out whether attitudinal stability varies by level of education, you will use comparison of correlations. For ease of analysis, we recoded education level (v043254) into three categories: "less than a high school diploma," "high school diploma, no college degree," and "college degree or higher." The new variable is called "edu.cats."

Comparison of correlations calculates the correlation coefficients between two variables separately for each category of the row variable. In each analysis, the pre-election response will be the independent variable, the post-election response will be the dependent variable, and education will be the row variable.

T-statistics measure whether or not the correlation in one cell of the table is different than what is expected (the overall correlation). When the absolute value of the T-statistic is large, the probability that you would obtain these results if the variables are not related at all is small. Higher absolute values of the T-statistic suggest that it's more likely that the variables are related.

Turn your attention to the relationship between education and stability in ideological identification. Do respondents with higher levels of education have more stable ideology than those with lower levels of education?

Now look at the relationship between education and stability in attitudes about government spending. Do respondents with higher levels of education have more stable attitudes about government spending than those with lower levels of education?

Next consider the relationship between education and stability of attitudes about government interventionism. Do respondents with higher levels of education have more stable attitudes about government interventionism than those with lower levels of education?

Think about your answers to the application questions before you click through the interpretation guide for help in answering them.

Weights (mathematical formulas) are often used to adjust the sample proportions, usually by race, sex, or age, to more closely match those of the general population. The analyses used in this guide did not use any weights, which may reduce the generalizability of the findings, but the resulting tables are accurate descriptions of the relationships found between these variables among these respondents.

Ideological Identification

How stable are the ideology responses? Is this the result you would expect for ideological identification?

Government Spending and Services

Are opinions on government spending and services stable between the pre-election survey and the post election survey? Is this what you expected, given it's centrality to elections?

Interventionism

How stable are opinions on government interventionism between the pre-election survey and the post-election survey? Is this what you expected given the issue's novelty in the political arena?

Education and Attitudinal Stability

Do respondents with higher levels of education have more stable ideological attitudes than those with lower levels of education?

Do respondents with higher levels of education have more stable attitudes about government spending than those with lower levels of education?

Do respondents with higher levels of education have more stable attitudes about government interventionism than those with lower levels of education?

Interpretation

Things to think about in interpreting the results:

  • It is important to look at the amount of missing data in each relationship and think about the ways in which that might affect the generalizability of the results - some analyses have relatively little missing data, while others may have a great deal. In general, results from this dataset should be fairly representative of the general population because it is a national probability sample.
  • Reading the results: A correlation is a measure of the strength of association between two variables. It is used to predict the value of one variable given the value of the other. Correlations are expressed on a scale from -1.0 to +1.0, with strongest correlations at both extremes and providing the best predictions. The closer the correlation coefficient is to +1, the more stable the responses were between the pre-election survey and the post-election survey. Correlation coefficients closer to zero indicate that there is a weak relationship between the responses.
  • When the absolute value of the t-statistic is large, the probability that the observed difference in scores (correlation coefficients in this case) is due to sampling error alone is small. Thus, higher absolute values of the t-statistic suggest that it's more likely that the variables are associated.

The analyses show the following:

  • The correlation coefficient associated with pre-election and post-election ideological identification is 0.82 which indicates a strong positive relationship. Remember that there is no direct interpretation of the coefficient (the value does not represent a proportion), but it reasonable to say that ideological identification for a very high percentage of respondents did not change or at least did not change much.
  • The correlation coefficients associated with pre- and post-election attitudes regarding government spending (0.54) and attitudes foreign policy/interventionism (0.55) are also indicative of a strong positive associations between pre- and post-election attitudes, but it appears that the stability of these attitudes are not as consistent as ideological identification. On the one hand, this makes sense because identities change more slowly than positions on issues. On the other hand, debates on government spending have long been part of the American political landscape and it would seem that attitudes would have coalesced to a greater degree. Still, the coefficients suggest that although there is not the same level of stability as seen with ideological identification, how the respondents felt about these issues after the election is pretty similar to how they felt before the election.
  • Respondents with a college degree or higher appear to have more stable ideological identifications than those with less education. Although we can see large differences in the correlation coefficients, we are better able to interpret the t-statistics found just below the correlation coefficient in each cell. For those with a college degree we see that the t-statistic is 8.2 which suggests that this correlation is much, much, much higher than we would expect if there was no relationship between education and the stability of ideological identification. We see the t-statistics for the other cells are -2.5 and -2.7 which suggests that the correlation coefficients are much lower than we would expect if there was no relationship between education and the stability of ideological identification.
  • Note that the correlation coefficient associated with those without a high school diploma is .43 compared to the .77 associated with those who have completed high school, but not a college degree, yet the t-statistics are very similar. This is because there are far fewer respondents who have reported that they have less than a high school diploma and the t-statistic is very sensitive to sample or group size.
  • It does appear that respondents with a college degree also exhibit greater stability on the issues of government spending and foreign policy/interventionism. The "large" t-statistics and the dark red shading of the cell in the "college degree" rows in both tables suggests that the correlation coefficient is significantly higher than we would expect if education was not associated with the stability of these attitudes.

Summary

The goal of this exercise was to explore attitudinal stability in different types of political issues. As a whole, one can draw the conclusion that long term issues are, on average, more stable than short term issues, and education matters in attitudinal stability. However, the confidence in this conclusion should be limited due to the small number of questions available for analysis.

CITATION: Inter-university Consortium for Political and Social Research. Attitudinal Stability on Short- and Long-term Issues: A Data-Driven Learning Guide. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2009-04-16. Doi: https://doi.org/10.3886/attitudinalstability

Creative Commons License This work is licensed under a Creative Commons Attribution-Noncommercial 3.0 United States License.