Issue Importance and Performance Voting Patrick Fournier, André Blais, Richard Nadeau, Elisabeth Gidengil, and Neil Nevitte *** Soumis à Political Behavior *** Issue importance mediates the impact of public policy issues on electoral decisions (Krosnick, 1988, 1990) Individuals who consider that an issue is important are more likely to rely on their attitudes toward that issue when evaluating candidates and deciding whom to vote for The logic behind the link between issue importance and issue voting should translate to a link between issue importance and performance voting Incumbent performance evaluations regarding an issue should have a stronger impact on the vote choice of individuals who find that issue important The analysis demonstrates that there is a significant interaction between performance evaluations and issue importance People concerned about an issue assign more weight to their evaluations of the government on that issue when making up their mind Issue importance is a significant mediator of issue voting (Krosnick 1988, 1990) The impact of various policy attitudes on candidate appraisal and vote choice is stronger among those individuals who feel that the issue in question is important Although this interaction has large implications for our estimates of issue voting and our understanding of voting behavior, it has received little attention since the publication of Krosnick s studies Rarely is issue importance integrated into models of political decision This paper seeks to expand the work of Krosnick on the mediating role of issue importance It determines whether the impact of incumbent performance evaluations on vote choice is mediated by issue importance The analysis draws on the 1997 Canadian Election Study, which contains detailed measures of citizens perceptions of the Liberal government performance on various issues, and of their opinions on the importance of these issues in the election The analysis demonstrates that there is a
significant interaction between performance evaluations and issue importance: people who consider that an issue is important are more likely to base their vote choice on their evaluation of the government s performance on that issue The Context Issue importance affects the impact of public policy issues on electoral decisions (Krosnick, 1988, 1990) Individuals who feel that an issue is important are more likely to rely on their attitudes toward that issue when evaluating candidates and deciding whom to vote for Krosnick found that: "Policy attitudes that citizens consider important are highly accessible in memory, are highly resistant to change, are highly stable over time, are extensively linked to and consistent with individuals basic values, instigate polarized perceptions of competing presidential candidates policy attitudes, and are powerful determinants of candidate preferences (Krosnick, 1990: 70)" Put differently, issue importance is a significant source of interpersonal heterogeneity in political decision-making (Rivers, 1988; Sniderman, Brody & Tetlock, 1991; Fournier, 2000) Since the effects of issues are not uniform across all voters, it is inappropriate to follow the homogeneity assumption and estimate the impact of issues across the entire population Krosnick suggests that the logic that links issue importance to issue voting should extend to concepts other than issue voting, notably government performance evaluations: "This article has focused exclusively on policy attitudes, but the theory of importance presented here may be useful for understanding the impact that other sorts of factors have on candidate preference as well () Similar extensions of the theory may be made to assessments of the incumbent's performance in office, political
party affiliation, and other such consequential variables (1990: 86)" Surveys often ask respondents to rate the performance of political actors currently in office Performance ratings are sometimes used as dependent variables capturing approval of a political actor s job handling (Neustadt, 1960; Mueller, 1973; Edwards, 1990; Brody, 1991) They are also used as measures of retrospective judgment on the incumbent's accomplishments, and included in models of candidate appraisal and vote choice (Downs, 1957; Butler & Stokes, 1969; Fiorina, 1981; Rose & McAllister, 1990; Flanigan & Zingale, 1994; Abramson, Aldrich & Rohde, 1999) The reasoning behind the link between issue voting and issue importance should translate to a link between performance voting and issue importance Individuals who are concerned about an issue should be more able to rate the performance of the government on that issue, they should have more polarized evaluations of the government s performance on that issue, and they should link their evaluations of the government on that issue more closely to their vote choice This paper tests these hypotheses The Study We draw on data from the 1997 Canadian Election Study (CES) 1 In the campaign wave of the CES surveys, respondents were asked to rate the performance of the incumbent Liberal government on 1 The 1997 Canadian Election Study was funded by the Social Sciences and Humanities Research Council of Canada The surveys contain a rolling cross-sectional component and a panel component A representative sample of 80-120 respondents (110, on average) was interviewed each day of the 36-day campaign to track the dynamics of vote intentions Respondents from the campaign wave were reinterviewed twice after the election, once by telephone, and once by mailback questionnaire A total of 3949 eligible voters were surveyed during the campaign, 3170 during the post-election telephone interview, and 1851 with the mailback questionnaire The response rate for the campaign survey was 59 percent, and the average daily sampling error is about 11 percentage points The survey was conducted by the Institute for Social Research at York University Copies of the questionnaires, technical documentation and data can be obtained at wwwisryorkuca/isr and through the ICPSR
six specific policy issues: preserving national unity, reducing the deficit, creating jobs, fighting crime, protecting social programs, and keeping election promises Table 1 presents the distribution of the performance evaluations Table 1 shows that the incumbent government was not judged uniformly across all issues The Liberals were perceived to have done a good job of reducing the deficit, and a bad job of creating jobs and keeping election promises Table 2 reports the average performance score for each group of party identifiers It indicates that some projection is at play The average performance score among Liberal identifiers is typically positive Conversely, partisans of the other parties generally give the Liberals negative scores However, the results do not solely reflect projection In two of the six issues, the average performance evaluation by Liberal identifiers drops below the neutral point into negative appraisals Furthermore, supporters of opposition parties often attribute neutral and even positive marks to the government For instance, despite the fact that they identify with another party, Conservative, NDP and Bloc partisans were willing to state that the Liberals had done a good job of reducing the deficit All in all, the evidence shows that government performance evaluations are closely related to partisan attachment, but that they are not strictly the product of projection To operationalize issue importance, Krosnick (1988) used closed-ended questions which inquired about the importance that individuals attach to several specific issues 2 There is evidence that responses to close- and open-ended importance question differ only in minor and predictable ways (Schuman, Ludwig & Krosnick, 1986) 3 In this study, we rely on both types of measures of issue 2 For instance, Krosnick used the following question from the 1984 NES: 'How imporant is it to that the federal government do what you think is best on this issue of X? Extremely important, very important, somewhat important, or not important at all?' 3 Using split sample experiments, Schuman, Ludwig & Krosnick (1986) found that all issues proposed in the closeended format were mentioned more often than they were in the open-ended form, and that minor issues experienced the greatest increases Although that study deals with questions about the importance of problems facing the country, its conclusions should extend to questions about the importance of issues in an election
importance First, respondents were asked an open-ended question about the most important issue of the election for them personally Second, respondents were also asked to rate the importance they attached to each of the performance issues specifically in close-ended questions The distribution of the two sets measures is presented in Table 1 for each issue 4 We created variables that combined each issue's two indicators of issue importance 5 The analysis proceeds with two tests First, it examines whether issue importance increases the likelihood of evaluating the performance of the government and of giving the government extreme performance ratings Then, it uses interactive terms between the issue importance indexes and the performance ratings to determine whether issue importance mediates the effect of performance ratings on Liberal vote choice 6 The Results First, it is clear that issue importance influences the capacity to rate the government s performance, and the propensity to give very positive or very negative evaluations of the performance of the government (see Table 3) 7 Individuals who feel that an issue is important are more likely to evaluate the government on that issue The average proportion of Don t know responses to the performance evaluations is substantially smaller among voters who deem the issue very important (4%) 4 This table also shows that the distribution of issue importance disadvantaged the incumbent in the 1997 election: the Liberals were hurt by the fact that the issues on which they had a negative record were deemed important (jobs and promises) while those on which they fared positively were considered less important (deficit) 5 We simply computed the mean of the open-ended (0: not mentioned as most important election issue / 1: mentioned) and the close-ended scales (0: not very important / 5: somewhat important / 1: very important) to form a variable with a range of 1 (0,25,5,75,1) The means of the indexes run from 4 (national unity) to 6 (jobs) 6 Rose & McAllister (1990) also use interactions between issue importance and performance ratings in their models of vote choice in Britain, but they do not show the contribution of issue importance to the explanation 7 The combined index of issue salience was used to distinguish respondents
than among those who feel it is not at all important (10%) Individuals who feel that an issue is important are also, on average, more likely to have extreme ratings of the government on that issue (26%) than those who consider it unimportant (21%) It should be noted, however, that the greater polarization among the high importance group occurs only at one end of the performance scale Citizens who feel an issue is important are particularly more likely to express very negative performance evaluations (21%) than those who feel it is not (13%) This relationship does not exist for very positive evaluations: individuals who say an issue is not important are actually more likely to give the government a very positive rating (8%) than those who deem it important (6%) 8 The more crucial test, though, concerns the impact of issue importance on the link between performance ratings and vote choice Does issue importance affect the impact of government performance evaluations on vote choice? Table 4 reports the estimates of two logistic regressions of the vote for the incumbent government party in 1997, the Liberal Party of Canada Controls for party identification, political values, issue positions, economic evaluations, strategic considerations and sociodemographic variables were also included in both models, but these coefficients are not reported to lighten the results 9 Model A shows the effects of the government performance evaluations on vote choice when issue importance and the interactive terms are not present in the estimation All six performance evaluations have a positive and statistically significant impact on support for the Liberals: favorable evaluations lead to greater Liberal support Once the issue importance variables and interactive terms are taken into account (Model B), however, the positive direct effects of performance evaluations dissipate completely The six interactive terms retain the sole positive and statistically significant impact on vote choice 8 There are six issues times four comparisons (very low versus very high importance): 17 of the 24 differences of means have the right sign, and 12 are statistically significant at the 05 level
These results indicate that issue importance mediates the impact of government performance evaluations on vote choice Individuals who felt an issue was important were more likely to rely on their evaluations of the government s performance on that issue when making up their mind about whom to vote for In fact, those who felt an issue was important were essentially the only ones to weigh government performance on that issue when deciding whether or not to vote for the incumbent 10 Conclusions This paper shows that issue importance has a mediating effect on the link between government performance evaluations and vote choice Individuals who feel that an issue is important assign more weight to their evaluations of the government on that issue when making up their mind As a result, the performance of the government on important issues affects Liberal support more strongly than its performance on unimportant issues Thus, negative performance evaluations on important issues hurt the incumbent more than negative performance evaluations on unimportant ones, and positive performance evaluations on important issues bring in more votes than positive performance evaluations on unimportant ones Furthermore, questions about issue importance and government performance appear to illicit meaningful responses from voters The performance of the incumbent is judged differently for each issue These evaluations of government performance are partly independent of party identification, there are not solely the product of projection or rationalization They contribute to the explanation of voting 9 There are listed at the bottom of Table 4 10 Simulations reveal that the difference in the probability of voting Liberal between an individual who rates the government's performance negatively (-1) and one who rates it positively (+1) is, on average across all six issues,
behavior even when controlling for partisan identification, political values, issue positions, and economic assessments And the issues which voters feel are more important have more weight on their decision about 16 percentage points among those who feel the issue is important (75), but only about 5 percentage points among those who consider the issue less important (25)
References Abramson, Paul, John Aldrich, & David Rohde 1999 Continuity and Change in the 1996 Election, Washington: CQ Press Brody, Richard 1991 Assessing the President: The Media, Elite Opinion, and Public Support, Stanford: Stanford University Press Butler, David, & Donald Stokes 1969 Political Change in Britain,, London: Macmillan Downs, Anthony 1957 An Economic Theory of Democracy, New York: Harper & Row Edwards, George C III 1990 Presidential Approval: A Sourcebook, Baltimore: John Hopkins University Press Fiorina, Morris P 1981 Retrospective Voting in American National Elections, New Haven: Yale University Press Flanigan, William, & Nancy Zingale 1994 Political Behavior of the American Electorate, Washington: CQ Press Fournier, Patrick 2000 Everything You've Always Wanted to Know about Heterogeneity: The Nature, Extent, Sources, Dynamics, and Consequences of Interpersonal Differences in Decision- Making, Unpublished Doctoral Dissertation, Vancouver: University of British Columbia Krosnick, Jon A 1988 The Role of Attitude Importance in Social Evaluation: A Study of Policy Preferences, Presidential Candidate Evaluation, and Voting Behavior, Journal of Personality and Social Psychology, 55: 196-210 Krosnick, Jon A 1990 Government Policy and Citizen Passion: A Study of Issue Publics in Contemporary America, Political Behavior, 12: 59-92 Mueller, John 1973 War, Presidents, and Public Opinion, New York: Wiley Neustadt, Richard 1960 Presidential Power, New York: Wiley Nevitte, Neil, André Blais, Elisabeth Gidengil, & Richard Nadeau 2000 Unsteady State: The 1997 Canadian Federal Election, Toronto: Oxford University Press Rivers, Douglas 1988 Heterogeneity in Models of Electoral Choice, American Journal of Political Science, 32: 737-757 Rose, Richard, & Ian McAllister 1990 The Loyalties of Voters:A Lifetime Learning Model, London: Sage
Sniderman, Paul M, Richard A Brody, & Philip E Tetlock 1991 Reasoning and Choice: Explorations in Political Psychology, Cambridge: Cambridge University Press
Table 1 Distribution of the Performance Evaluations and the Importance Assessments Mean Liberal Percent Said Percent Mentioned Performance Very Important as Most Important Evaluation in Closed Question in Open Question National unity 04 555 % 91 % Deficit 15 598 % 72 % Jobs -25 839 % 288 % Crime -04 688 % 18 % Social programs -04 598 % 157 % Election promises -27 749 % 27 %
Table 2 Average Performance Evaluations by Party Identification Liberal Conservative NDP Reform Bloc Preserving national unity 22-02 04-20 -09 Reducing the deficit 33 13 14-22 10 Creating jobs -08-30 -40-40 -44 Fighting crime 11-07 -08-37 -11 Protecting social programs 11-04 -24-03 -24 Keeping election promises -03-35 -36-55 -44
Table 3 Non-response and Extreme Government Performance Ratings by Issue Salience Individuals who feel the issue is: Not Important Very Important Average % of 'Don't know' to performance ratings 103 35 Average % of extreme performance ratings 206 258 Average % of very negative performance ratings 129 205 Average % of very positive performance ratings 78 58
Table 4 Logistic Regression of Liberal Vote (Unstandardized Coefficients and Standard Errors) Model A Model B Liberal Performance Preserving national unity 10 (13) -37 (26) Reducing the deficit 31 (15) ** -16 (30) Creating jobs 45 (16) *** -14 (35) Fighting crime 28 (14) ** -28 (37) Protecting social programs -16 (15) -133 (31) *** Keeping election promises 29 (15) ** -18 (35) Issue Importance National unity - 62 (33) * Debt and deficit - -75 (43) * Jobs and unemployment - -19 (32) Crime - 71 (52) Social programs - 44 (37) Election promises - -61 (47) Importance * Performance National unity - 117 (51) ** Debt and deficit - 121 (60) ** Jobs and unemployment - 81 (51) * Crime - 147 (80) ** Social programs - 250 (57) *** Election promises - 106 (78) * Constant -110 (58) -98 (72) 2 Log Likelyhood 12152 11689 Number of cases 1490 1490 Control variables included in the estimation but not reported: partisan identification, ideological variables (women, regional alienation, outgroups, Canada-US, cynicism, moral traditionalism, free enterprise), issues (spending, taxes, deficit, jobs, crime, gun control, immigration), economic perceptions (fear of losing job, personal finance retrospective, personal finance prospective, unemployment rate retrospective, unemployment rate prospective, national economy retrospective, national economy prospective, provincial economy retrospective, provincial economy prospective), strategic considerations (government, opposition), and numerous sociodemographic variables For more details on the construction of these variables, see Nevitte, Blais, Gidengil & Nadeau, 2000)