The imagined ideal citizen in a democracy is informed about the issues of the day, closely follows the positions that candidates take on those issues, and weighs those positions when deciding for whom to vote. In an electorate filled with such citizens, the candidates' positions on important social issues would be a critical determinant of election outcomes. Furthermore, citizens would consider whether an issue was something that was truly in the domain of presidential action. For example, the importance of the economy as an issue would be tempered by the understanding that the economy is far too complex for a president to manage, and that business cycles will occur regardless of the actions of the president.
When political scientists began using large-scale national surveys to study political attitudes and voting behavior in the 1950s, what they actually discovered about voters was that many lacked even a minimal level of knowledge about political issues. Furthermore, there was little evidence that a voter's position on issues, as measured by survey questions, influenced the voter's choice of candidates (Campbell, Converse, Miller, and Stokes 1960). In fact, many voters seemed to have no opinions whatsoever regarding prominent national issues.
In a longitudinal survey that asked respondents the same identically-worded questions regarding their opinions about how the federal government should handle such issues as education, housing, and employment, only about thirteen out of twenty people gave the same responses in successive surveys at two-year intervals (Converse 1964, 239). This result leads one to suspect that many respondents simply had no opinions regarding the issues and answered the questions randomly. The inability of political scientists to predict voting behavior using public opinion on issues has led to a search for new explanations for voting behavior.
If voters do not choose among candidates on the basis of issues, how then do they decide for whom to vote? A group of researchers at the University of Michigan offered an answer to this question that has framed the debate within the study of voting behavior ever since. They argued that the voting behavior of most voters can be explained by a long-term "psychological identification" with one of the two major parties. Because the Republican and Democratic parties are "the most enduring objects of the political environment," voters' perceptions of candidates are filtered through voters' perceptions of the parties (Campbell et al. 1960, 27-28).
The high rate of correlation between the partisanship of children and their parents led the authors to speculate that partisan orientations originate in childhood1 (Campbell et al. 1960, 86). Panel studies indicated that individuals were considerably more likely to cross party lines in a particular election than they were to change their partisan identifications.2 Furthermore, those with the most intense partisan preferences were the most likely to vote, and the most likely to vote in accordance with their partisan preference (Campbell et al. 1960, 52-53).
The Michigan School has been challenged by others who argue that subsequent survey research has revealed a better-informed and less partisan public than did the National Elections Studies of the 1950s and early 1960s. Two issues emerged in the 1960s-race and Vietnam-that created sharp divisions among the populace and led the public to form more coherent views than had been the case in the 1950s (Nie, Verba, and Petrocik 1976, 104). At the same time, the percentage of strong partisan identifiers declined, and the percentage of independents increased from 1964 to 1974 (Nie et al. 1976, 49, 71-72). As a result, there was increasing correlation between voters' attitudes on issues and their votes, and a decline in the relationship between partisan identification and the vote (Nie et al. 1976, 164-66). These findings, however, have not gone unchallenged.
Carmines and Stimson (1980) argue that an increase in issue voting during the 1960s was not the result of a better informed electorate. They differentiate between "easy issues" and "hard issues." "Easy issues" are those issues that even uninterested and uninformed voters have opinions about. Three characteristics of an easy issue are the following: (1) it is "symbolic rather than technical," (2) it is "more likely to deal with policy ends than means," and (3) it is "an issue long on the political agenda" (Carmines and Stimson 1980, 80). Hard issues, on the other hand, require that voters weigh a variety of policy alternatives.
Carmines and Stimson found that less than one-quarter of the electorate in 1972 could be classified as "hard-issue voters" (1980, 86-87). Because much of the increase in issue voting in the 1960s and 1970s was based on "easy issues," Carmines and Stimson contend that issue voting was not indicative of a well-informed electorate. Instead, issue voting is most likely to occur "when parties engage in the hazardous behavior of staking out opposing positions on a deeply felt issue" (1980, 88). Polarizing issues, such as race and the Vietnam War, divided the parties. More ideological candidates, such as Barry Goldwater and George McGovern, provided easier opportunities for issue voting.3
The evidence for a decline in partisan identification also depends on how partisan identification is measured. The number of self-identified independents has clearly grown since the 1950s. However, independent "leaners," who identify themselves as closer to one party or the other, vote as consistently or more consistently for that party than do "weak" identifiers. Pure independents, who lean toward neither party, are not increasing4 (Keith et al. 1992, 64-65). Even in the 1992 election, independent candidate Ross Perot was only slightly more successful with independent leaners than with weak partisans. The increase in the number of self-identified independents has not been matched by a corresponding increase in the number of voters who act like independents.
Survey data demonstrate the enduring ability of partisanship to explain voting decisions. Table 1 shows that, in 1992, President Clinton won the votes of 93.5 percent of strong Democrats, and more than two-thirds the votes of weak Democrats and those who lean toward the Democrats. By contrast, Clinton was supported by only 3.2
percent of strong Republicans, and less than 15 percent of weak Republicans and those who lean toward the Republicans.
Table 2 shows that the partisanship of parents and the partisanship of children remains highly correlated. Of children whose parents were both Democrats, 73.4 percent either identified with or leaned toward the Democratic Party, while of those whose parents were both Republicans, 74.2 percent either identified with or leaned toward the Republican Party. Furthermore, partisanship is not simply a shortcut for those who are uninterested in politics. People who report that they follow public affairs "most of the time" are more likely to be partisan-identifiers, and are more likely to vote consistent with their partisan identification than are less attentive respondents (American National Election Studies [ANES] 1992).
The conclusions of the Michigan School have also been challenged from the perspective of a rational choice approach. In Anthony Downs's influential model of voting behavior, "each citizen casts his vote for the party that he believes will provide him with more benefits than any other"5 (Downs 1957, 37). Voters, however, must make this "rational choice" decision without perfect information, and the most reliable and easily accessible information at hand is the performance of the current party in power. Such voters are interested in outcomes and are less interested in the policies used to attain those outcomes (Fiorina 1981, 11).
For rational choice theorists, the undeniable link between a voter's partisan identification and his or her voting behavior is not an indication of "psychological attachment." Rather, party preferences are rooted in "past performance, not wishful thinking or expressions of team loyalty" (Popkin 1991, 57). A voter may be unaware of a party's position on current issues, but may nevertheless support that party due to favorable evaluations of its performance in office over time. Fiorina argues that "what we think of as separable parts are actually intertwined, with current issues and experiences becoming the stuff of retrospective evaluations that cumulate into a kind of long-term party judgment that in turn affects the interpretation of current issues and experiences" (1981, 56).
In theory, each voter can define "benefits" according to his or her own criteria. Empirically, however, it is nearly impossible to determine how individual voters determine their "benefits." Consequently, rational choice theorists generally begin with the highly questionable assumption that voters are motivated primarily by economic self-interest. In fact, political scientists have found that a voter's opinion regarding the national economy as a whole is a better predictor of his or her vote than is a voter's opinion regarding his or her own economic circumstances.
Markus, for example, found that national economic conditions had an impact on voting decisions in presidential elections from 1956 through 1984, even when a voter's personal economic circumstances were held constant (1988, 151). This seems to have been the case in 1992 as well. Table 3 shows that President Bush failed to win a plurality of those who described their own economic condition as much better than it was a year before. On the other hand, as Table 4 shows, Bush won more than 70 percent of the votes of those who responded that the national economy was better than it had been a year before.
Neither the Michigan School nor the rational choice approach views social issues as important determinants of presidential elections. The Michigan school argues that most voters will vote for the candidate from the party to which they have a life-long psychological attachment. Although the outcome of the election will be determined by short-term forces that favor one party or the other, the general lack of knowledge regarding issues among the majority of American voters precludes them from making decisions based on issues. Retrospective voters are interested only in outcomes. If the incumbent administration has performed satisfactorily, particularly with regard to the economy, the incumbent is likely to be reelected. If this is not the case, the opposition party's candidate will win.
Why is issue voting so difficult? It is a process with three prerequisites. First, the voter must have some knowledge about the issue, and must have preferences regarding the proper government response to it. Second, there must be a perceptible difference between the candidates regarding the issue, and the voter must have knowledge of the candidates' positions. Third, the voter must feel strongly enough about the issue to cast a vote based on it. Few issues meet the three criteria for issue voting.
Voter Knowledge and Types of Issues
The authors of The American Voter write: "If an issue is to motivate a voter, he must be aware of its existence and must have an opinion about it. Although this statement is obvious, it draws our attention to the fact that many people know of the existence of few if any of the major issues of policy" (Campbell et al. 1964, 98-99). In addition, in order for a problem to be viewed as a national problem, it must be connected in the minds of voters to the actions of the government. Issues such as crime, drug use, and health care clearly affect the daily lives of many Americans, but it is often difficult to determine how the actions of elected officials relate to personal experiences (Popkin 1991, 24-28). Issue voting is also easier when there are only two identifiable positions. It becomes more difficult when there are a variety of alternatives and the voter must choose the candidate whose position is the closest to the voter's preferences (Nie et al. 1976, 158-59).
The easy issue/hard issue dichotomy is useful, but it fails to capture key differences among "easy issues," and does not acknowledge that an "easy issue" for one voter may be a "hard issue" for another voter. Easy issues may be polarizing issues, such as abortion or homosexuality. These issues provide stark choices among alternatives that touch on deeply held moral principles. Issues can also be "easy issues" when there is a consensus among the electorate that a particular outcome is desirable. For example, nearly everyone agrees that there should be less crime or that the education system should be improved; however, there is little agreement regarding the means to achieve these ends. Candidates will attempt to portray themselves as responsive to voters' demands for fighting crime or improving the quality of education, but this portrayal is often more symbolic than substantive, and the actual policy differences between candidates are obscured. For voters who have clear policy preferences and measure the candidates' positions against these preferences, the same issues can be "hard issues."
Candidates and Issues
Political candidates are aware of the difficulties that voters have in choosing among multiple and complex policy alternatives, and will often focus their campaigns on issues where there is a direct link between actions and results-such as the death penalty-instead of more difficult issues (Popkin 1991, 100). Politicians can raise the salience of a particular issue by discussing it in their campaigns or by taking action if they are elected officials. For example, in January 1994, shortly after the failure of the Clinton Administration's efforts to reform the health care system, 31 percent of respondents in one poll mentioned health care as the most important problem facing the United States. A May 1996 poll showed the number had declined to 9 percent, despite that fact that virtually none of the problems associated with the American health care system had been solved in the preceding eighteen months. (Gallup Poll May 1996)
Whether political campaigns enhance or obscure public knowledge about issues is a subject of debate.
Popkin argues that campaigns increase overall knowledge about the issues that are discussed, increase the perceived importance of an issue," "strengthen the connections between an issue and an office," and "increase the perceived differences between the candidates on an issue" (1991, 39, 108). Others see political campaigns as spectacles in which candidates or officeholders present simple stories in which they appear "exceptionally decisive, tough, courageous, prescient, or prudent."
The complexity of governing is reduced to a series of symbolic confrontations that purport to illuminate the candidate's character but in fact obscure the difficult choices of governing (Miroff 1994, 274-79).
Intensity of Voter Preferences
The final condition for issue voting is that the voter must feel strongly enough about the issue or issues in question to base his or her vote on those issues. Clearly, some people feel strongly enough about a particular issue to base their voting decisions on that issue alone. These groups, or "issue publics," may be highly knowledgeable about both the issue itself and the positions of the candidates (Popkin 1991, 28-29). Two types of issues are likely to have highly-motivated issue publics: issues that are viewed in terms of moral choices, such as abortion, and issues that have a direct impact on attentive groups, such as Medicare.
In the case of moral issues, typically there are activists on both sides of the issue. These activists may be very important in low-turnout elections, such as party primaries and school board elections, but among the electorate at large, they tend to be small groups that offset each other. In the case of issues with attentive constituencies, such as Medicare among the elderly, candidates are careful to avoid alienating such groups, and the lines between the parties will tend to blur. Although recent Republican proposals to cut Medicare spending were not drastic, those efforts may harm Republican candidates in the 1996 elections. On the other hand, when a group is inattentive and unorganized-for example, welfare recipients-their interests can be safely ignored. Neither party has shown any concern that major cuts in welfare programs for the poor will cause poor people to change their voting behavior.
The Issues of 1996
What issues are most likely to influence people's voting decisions in the 1996 presidential campaign? According to a Gallup poll taken in May 1996 (see Table 5), 22 percent of respondents mentioned crime as the "most important issue facing the country." Crime, of course, has a major impact on the lives of millions of Americans. Many people have been or know victims of crime.6 People alter their behavior to avoid situations in which crime is likely by avoiding dangerous areas, locking doors, buying alarm systems, or possibly arming themselves.
Crime has an enormous economic cost, as governments must build prisons and pay law enforcement officers, while individuals must buy insurance, replace stolen property, and pay medical expenses.7 Crime is a daily feature in the news, as well as in entertainment. If one candidate could make a convincing case that he or she could reduce crime, or at least handle the problem more successfully than his or her opponents, the electoral benefits would be enormous. For a variety of reasons, however, this is unlikely to happen.
First, people often lack basic knowledge about the issue. Crime rates are declining, albeit modestly, at the same time that the number of people who consider crime to be the most important problem is increasing. The rate of decrease is most notable in the category of violent crime.8 In addition, the responsibility for preventing crime belongs more to state and local governments than to the federal government. The police officer who responds to the typical crime, whether it is a stolen bicycle or a murder, is likely to be an employee of a state or local government. If an arrest is made, the accused is likely to be tried in a state or municipal court. Important decisions, such as whether capital punishment will be an option, are made by state legislatures and governors. Nevertheless, crime is a perennial issue in presidential campaigns.
Second, crime is a complex social phenomenon, and there is little consensus among social scientists regarding the proper measures for reducing crime. When the issue is discussed in a presidential campaign, the candidates tend to focus on only the simplest proposals: more police officers, tougher sentences for convicted criminals, the death penalty, and gun control. In 1996, of these proposals, only gun control seriously divides Republican and Democratic candidates for the presidency. Considering that there are millions of guns already in circulation, one does not have to be a member of the National Rifle Association to believe that a law to limit the availability of guns would do little to reduce overall rates of crime in the United States.
Presidential candidates continue to surround themselves with police officers and pledge to be tough on crime, but this is little more than posturing. There is no reason to believe that this pattern will change in the 1996 campaign. In his speech accepting the Republican presidential nomination, Bob Dole warned that if he wins the election, "the lives of violent criminals are going to be hell." Yet the candidate offered just two specific crime-fighting ideas-one that few would oppose, a nationwide check on the criminal records of potential gun buyers, and one that the president has no power to implement, abolishing parole at the state level.
By contrast, the abortion issue is well suited for issue voting. For those who believe that abortion is the equivalent of murder, there is no question about what the proper policy of the federal government should be: abortion should be banned. Likewise, for those who believe that abortion is a private decision, there is no question that the government should ensure that abortion remains legal and available to women. Of course, many people who oppose abortion qualify their positions by making exceptions for rape victims or for the life of the mother, and many who support abortion rights would permit those rights to be conditioned by parental or spousal consent. Although these subissues may be morally difficult, such issues are finite and readily comprehensible. In addition, the Republican and Democratic candidates for president have taken opposite stances on the abortion issue in all recent elections.9
The abortion issue, however, is not likely to change the outcome of a presidential election. Surveys consistently show that abortion is the "most important issue" for only roughly 1 percent of the American public. Furthermore, even among those whose personal opinions on abortion are uncompromising, many will vote for candidates who advocate the opposite point of view. For example, in 1988, 46 percent of those who said that abortion should never be permitted voted for the pro-choice candidate, Michael Dukakis (ANES 1988). Despite the strong feelings that surround the abortion issue, it does not seem to influence the presidential choices of many voters.
Taking the Long View
As a proximate cause of voting behavior, it is difficult to argue that social issues have the impact of either partisan identification or economic conditions in most presidential elections. Nevertheless, these issues do influence voters' perceptions of political parties over time. The image of the Democratic Party as the protector of the poor and the elderly is rooted in Franklin D. Roosevelt's New Deal and Lyndon B. Johnson's Great Society. Similarly, the Republican Party has achieved great success by portraying itself as tough on crime and the guardian of traditional family values. The Reform Party, meanwhile, struggles to portray itself as the party of deficit reduction.
One can question the accuracy of any of these perceptions in 1996.
President Clinton seems to have no suggestions for reducing poverty. Bob Dole seems to have no suggestions for reducing crime. And Ross Perot's plans for deficit reduction lack specificity. In 1996, there is no social issue important to a large percentage of the electorate on which this particular set of candidates is "staking out opposing positions."10
Some future presidential election is likely to present starker choices on social issues. There are potential candidates, such as Pat Buchanan, who would not hesitate to base their campaigns on divisive positions on social issues, and history will inevitably yield new crises that challenge the parties to make tough choices that will shape their images. Without a highly salient, polarizing issue, however, the majority of American voters lack the necessary information to choose their president on the basis of issues.
1 Of course, partisan identification does not remain indefinitely stable even if it is transmitted from parent to child. When a party succeeds or fails during a period of crisis, people's views of the party are forever altered. For example, the Democratic Party became a majority party in the 1930s because Franklin D. Roosevelt's New Deal attracted many new voters to the party.
2 The National Election Study has conducted only two panel studies, and both lasted only four years. Some question remains about the long-term stability of partisan identifications (Petrocik 1989, 45). If individuals frequently change their partisan identification, the concept has little utility for political science.
3 The racial unrest of the summer of 1963 provides an example of how events can quickly change the perceived importance of an issue. Between March and September of 1963, the percentage of respondents who mentioned "civil rights/integration" as the most important problem in the Gallup poll increased from 4 percent to 52 percent. (Nie et al. 1976, 99)
4 The American National Election Study asks two questions. First, they ask whether the respondent thinks of himself or herself as "a Republican, a Democrat, an Independent, or what." Then, as a follow-up, the respondent is asked whether he or she thinks of himself or herself as a "strong" or "weak" Democrat or Republican. If the respondent identifies himself or herself as an Independent, then he or she is asked whether he or she is "closer" to the Democrats or to the Republicans.
5 This rule applies only to two-party systems. In multiparty systems, rational voters should often vote for their second or third choice in order to prevent even less desirable candidates from winning.
6 The Bureau of Justice Statistics reported that 4.9 million households had a member victimized by violent crime (rape, robbery, or assault) in 1992. See "Selected Findings from the Bureau of Justice Statistics: Violent Crime: National Victimization Survey," April 1994 NCJ-147486.
7 The Bureau of Justice Statistics reported that victims of violent crime lost $1.4 billion dollars in 1992. The Federal Bureau of Investigation estimated that property valued at $15.6 billion was stolen in 1994. See "Crime in the United States," 1994.
8 Both the Bureau of Justice Statistics, which measures the rate of crime through victimization surveys, and the Federal Bureau of Investigation, which records the number of crimes reported to law enforcement officers, have found declines in violent crime in the 1990s. The Bureau of Justice Statistics data suggest that crime rates peaked in the early 1980s and that 1992, the year for which the most recent data are available, had the lowest recorded rate of crime since the victimization survey began in 1975. By contrast, FBI data suggest that crime rates have been declining only since the early 1990s and that crime rates today remain higher than those of the mid-1980s. Both the Bureau of Justice Statistics and the FBI maintain Internet sites with considerable data on crime statistics. See http://www.fbi.gov/ucr/ and http://www.usdoj.gov/bjs/.
9 The ability of a president to actually implement his or her abortion policies is less clear. The appointment of Supreme Court Justices is one important area of influence. Presidents can lobby Congress to pass constitutional amendments. More likely, presidents can make incremental changes that influence the availability of abortion by regulating abortion providers and by funding decisions.
10 The issue of crime and the issue of race are inextricably linked in American politics, and there were clearly racial overtones to the Republican's crime ads in 1988. It is beyond the scope of this article to address the role of racism in presidential politics, but I have little doubt that it remains considerable.
Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. The American Voter. New York: Wiley, 1960. (references are to abridged edition, 1964).Carmines, Edward G., and James A. Stimson. "The Two Faces of Issue Voting." American Political Science Review 74 (1980): 78-91.Converse, Philip E. "The Nature of Belief Systems in Mass Publics." In Ideology and Discontent, edited by David E. Apter, 206-58. New York: Free Press, 1964.---. "The Concept of a Normal Vote." In Elections and the Political Order, edited by Angus Campbell et al., 9-39. New York: Wiley, 1966.Downs, Anthony. An Economic Theory of Democracy. New York: Harper, 1957.Fiorina, Morris P. Retrospective Voting in American National Elections. New Haven: Yale University Press, 1981.Keith, Bruce E., David Magleby, Candice Nelson, Elizabeth Orr, Mark Westlye, and Raymond Wolfinger. The Myth of the Independent Voter. Berkeley: University of California Press, 1992.Markus, Gregory B. "The Impact of Personal and National Economic Conditions on the Presidential Vote: A Pooled Cross-Sectional Analysis." American Journal of Political Science 32 (1988): 137-54.Miroff, Bruce. "The Presidency and the Public: Leadership as Spectacle." In The Presidency and the Political System, 4th ed., edited by Michael Nelson, 273-96. Washington, D.C.: CQ Press, 1994.Nie, Norman H., Sidney Verba, and John R. Petrocik. The Changing American Voter. Cambridge: Harvard University Press, 1976.Petrocik, John R. "An Expected Party Vote: New Data for an Old Concept." American Journal of Political Science 33 (1989): 44-66.Popkin, Samuel L. The Reasoning Voter: Communication and Persuasion in Presidential Campaigns. Chicago: University of Chicago Press, 1991.Survey data are from the 1988 and 1992 American National Election Studies conducted by the Survey Research Center and the Center for Political Studies at the University of Michigan with support from the National Science Foundation and disseminated by the Interuniversity Consortium for Political and Social Research (ICPSR). All data analysis is solely the responsibility of the author.
Scott Piroth is a Ph.D. candidate at American University in Washington, D.C.