Cato Policy Analysis No. 3 August 24, 1981

Policy Analysis

An Alternative Analysis of Mass Belief Systems: Liberal, Conservative, Populist, and Libertarian

by Stuart A. Lilie and William S. Maddox

Stuart A. Lilie is an associate professor in, and the chairperson of, the Department of Political Science at the University of Central Florida, Orlando 32816. William S. Maddox is also an associate professor in that department, and for 1981-82 he is a visiting assistant professor in the Department of Political Science at the University of New Orleans, Louisiana 70122. An earlier version of this paper was presented at the annual meeting of the Midwest Political Science Association, Chicago, Illinois, April 1979.

Executive Summary

This study, based on the 1976 Center for Political Studies Election Study, suggests that at least four different belief systems exist among the American public. We treat the dimension of government intervention in economic affairs as distinct from the dimension of expansion of individual liberties. The liberal supports govern- ment economic intervention and the expansion of individual liberties; the conservative opposes both. The libertarian supports expanded individual liberties but opposes economic intervention. The populist supports economic intervention but opposes expansion of individual liberties. Some 72.4 percent of the sample can be classified as holding consistent or nearly consistent political beliefs using these categories. The relationships between our ideological categories and demographic group- ings are generally consistent with expectations about these groups. We also explore the relation- ships between our classifications and ideological self-identification, party identification, and presidential vote.

Political scientists use a variety of methods to measure the presence and direction of ideological thinking among mass publics. A standard approach is to code responses to open-ended "like and dislike" questions about candidates and parties according to ideological content. (1) Another method is to use correlations across issues positions (2) or across issue clusters. (3) Pierce (1970) added what he called the informational and affective dimensions of ideology, based on definitional questions about ideology and feeling thermometer reactions to the terms liberal and conservative. (4) Several studies use factor analysis to determine the number and nature of issue dimensions that are necessary to structure opinions in both masses and elites. (5) Finally, the ideological self-identification of respondents, often included in studies of particular elections, (6) has recently received attention as an independent predictor of presidential voting. (7)

Despite the great variety of techniques, all these studies share a common approach; a single liberal-conservative dimension is the primary tool for evaluating the presence and direction of ideological thinking among mass publics. Thus respondents whose attitudes do not fit the researcher's definition of liberal or conservative are categorized as nonideological or inconsistent. This approach of course assumes that liberal and conservative are the most meaningful and logical positions for a person to take. In recent years, however, researchers have become uncomfortable with this unidimensional approach, as is indicated by various caveats and disclaimers. For example, LeBlanc and Merrin admit that "there is no logical reason why voters could not be liberal on some issues and center or conservative on others," (8) while Nie, Verba, and Petrocik conclude that those whom they classify as inconsistent may be "consistent in ways we do not recognize." (9)

In spite of these caveats, analysts of mass belief systems continue to use a liberal-conservative continuum in their research. We contend that there are empirical and theoretical reasons to go beyond this unidimensional approach. Many of those respondents who are currently labeled "inconsistent" may in fact be consistent on other dimensions as significant as the present liberal-conservative continuum. By extending and refining ideological categories, we will more adequately understand the belief systems of the American electorate.

To achieve this end, we propose a two-dimensional approach as the basis for the analysis of mass belief systems. We measure attitudes toward economic intervention by government and attitudes toward economic intervention by government and attitudes toward individual liberties as separate dimensions and consider four ideological categories based on these two dimensions: liberal, conservative, libertarian and populist. Our definitions of liberal and conservative are generally consistent with current practice; there are also, we will argue, valid grounds for including the categories of libertarian and populist. Our approach, then, is an outgrowth and complement to current research in that it includes the liberal and conservative categories as traditionally defined, but attempts to account for many of those others who are "consistent in ways we do not recognize." Before developing specific definitions and categories, we will examine three arguments in support of the use of a two-dimensional approach to contemporary ideologies. (10)

First, among public-opinion researchers, as a recent writer notes, there is "an observation of great vintage: the content of mass political policy preferences is multidimensional...." (11) The origin of this recognition is the work in the fifties by Stouffer (1955) and Lipset (1959), which showed that working-class individuals are more "liberal" on economic issues but less "liberal" with respect to civil-liberties issues. Since then the nature of these dimensions and their relationship to social and economic status have been amplified and refined by many writers, including some who argue that this classic relationship is becoming inverted. (12) (10) The work of Milton Rokeach (1973) is an interesting exception to this reliance on a single continuum. He suggests that liberalism-conservatism is not a bipolar ideology, but must be considered as having two dimensions -- attitudes toward freedom and attitudes toward equality. This approach has value when analyzing competing ideologies from a global perspective, but as Rokeach himself says, its usefulness is limited in a given political system where there is a high degree of consensus, as in the United States, where certain basic freedoms are highly valued by almost everyone. Our proposal is a two-dimensional approach that clarifies ideological differences on economic and individual liberties issues, recognizing that these differences are relatively subtle because they largely fall within the parameters of the American consensus on these issues.

More recently, studies using factor analysis have also identified a number of separate dimensions in mass attitudes, including separate economic and social dimensions. (13) While the existence of these two dimensions is now an observation "of great vintage," little has been done to develop the full implications of this observation for categorizing individuals as liberal or conservative.

A second body of literature that demonstrates the inadequacies of a unidimensional approach is elite studies. For example, in his analysis of the Supreme Court, Schubert (1965) scaled liberalism into two basic dimensions, political and economic. He defined political liberalism in terms of "claims and personal (as distinguished from property) rights and freedoms" (p. 101) and economic liberalism as relating to "conflicts of interest between the economically affluent and the economically underprivileged" (p. 128). Schubert's analysis of the components of ideology is more sophisticated than that used here, but his work demonstrates the inadequacy of unidimensional definitions of liberalism and conservatism in elite analysis and again suggests that the two basic dimensions of ideology are economic issues and questions of personal liberty. Students of the U.S. Congress have found that while congressional roll-call votes show party as a strong predictive variable, a number of other dimensions are significant in explaining divisions in legislative votes, including several "policy domains" or "issue sets." (14) One student of Congress concludes that "different alignments form as the policy content changes . . . liberal and conservative ideologies do not provide the bases for many, many policy decisions." (15) Thus analyses of two significant segments of political elites, the Supreme Court and the Congress, call into question the assumption that elites rely simply on a liberal-conservative continuum in their decision Making.

A third reason for going beyond an unidimensional approach in the study of mass belief systems is that liberal and conservative are not the straightforward terms they are often assumed to be. While early-nineteenth-century liberal thought emphasized the expansion of individual liberties, including the absence of government intrusion in economic choices, by the late nineteenth century debate over the relationship of liberty and equality often focused on the proper role of government in acting to alleviate the perceived hardships of a capitalist economy. Indeed, the connection between government's economic role on the

one hand, and individual liberties on the other hand, has been the crux of much of the ideological debate in recent Anglo- American thought. Traditional theory provides no logically necessary reason to place attitudes about economic issues on the same dimension as attitudes about individual liberties. In fact, to do so obscures much of the theoretical debate of the past one hundred years.

If the advantages of supplements to the liberal- conservative continuum are so obvious, why has a unidimensional approach persisted? This question deserves more extended treatment, but two factors are notable here. One is the assumption by many researchers that liberal and conservative define the context of American politics. It is the basic division among elites and is therefore the relevant continuum for the voter. If mass publics hold other ideological positions, these are largely irrelevant to the dynamics of the political system. However, while it is true that political elites and the media use liberal-conservative terminology, scholars have found, as we have indicated, that close analysis of elite behavior often requires that they discard or modify the liberal-conservative continuum. A second factor is the methodological simplicity of a unidimensional approach. Conceptualization, question and scale construction, statistical analysis, and even graphic representation of data are simplified when working with a single dimension. Perhaps both these factors are reflected by the authors of a recent article, who conclude that it is necessary to analyze issue consistency on a liberal- conservative scale so that "electoral mandates could be easily interpreted." (16)

Empirical Definition of Belief Systems

Taken together, these three arguments provide theoretical and empirical bases for going beyond the traditional liberal- conservative dimension in the analysis of mass belief systems. They also suggest that the question of government intervention in the economy is separable from the question of individual liberties. To operationalize these two dimensions with available survey data, we selected three issue questions for each dimension from the 1976 Center for Political Studies Election Survey. As other political scientists have done, we selected issues that appear to represent each dimension; (17) we also found some empirical support for our selection of issues. The respondents in the survey tended to treat these issues as representative of distinct dimensions, as indicated by a statistical technique called factor analysis. Factor analysis identifies types of issue questions to which people responded in similar ways.

We performed a factor analysis with varimax rotation of twenty-five issue questions. This analysis revealed eight factors, of which two explained a tot a] of 60.7 percent of the variance. The two dimensions used here are defined by the three issues with highest loadings on those two factors that were the issue questions we originally chose. Factor I, economic intervention, is represented by government guarantee of jobs (with a loading of .57), government health insurance (.43), and progressive taxation (.39). Factor II, individual liberties, is represented by legalization of marijuana (.54), abortion (.43), and equal role for women (.33).

We combine the two major issue dimensions into a fourfold typo logy, which we use to approach the ideological orientation of the public. Perhaps the most straightforward of our four types is the libertarian who opposes government intervention but supports a high degree of individual liberties. This is the classical liberal position of the nineteenth century, represented by such figures as the early John Stuart Mill and (in extreme form) by Herbert Spencer. Because the mainstream of contemporary liberal thought has moved away from this position, it is confusing to label it liberal. However, the term libertarian is generally used today to describe this position. The libertarian position has recently received attention through the efforts of the Libertarian party (their 1980 presidential candidate received 1.1 percent of the popular vote). In the latter part of the nineteenth century such liberals as the later J. S. Mill and T. H. Green argued that if the liberal goal of promoting the independence and well-being of the individual were to be achieved, opposition to all forms of government economic intervention must be modified. Thus the state should take some responsibility for individual welfare through such measures as wage and hour laws, compulsory public education, and health and safety regulations. At the same time, liberals continued to support such personal liberties as freedom of speech and conscience. This position, sometimes called "welfare state liberalism" (although for obvious reasons liberals in the United States rarely use this term), is generally what is meant by liberalism in contemporary usage. Thus in terms of our two dimensions the liberal supports government intervention but also supports an expansion of individual liberties. On the other hand the conservative position as it has developed in the United States has been that human nature is sufficiently deficient that, without constraints imposed by society, the individual is likely to behave in deviant and socially harmful ways. Thus some restrictions on individual liberties are in principle desirable. In addition, American conservatives argue that in the economic realm, perhaps again because of man's selfish nature, the workings of the market should be relatively free from government intervention, in order to be efficient and productive. In terms of our two dimensions, the conservative supports restrictions on personal liberties but opposes government economic intervention. Finally the populist, while sharing the moralism of the conservative in regard to individual liberties, feels that an unregulated economy often means an unfair concentration of wealth at the expense of the poor. Thus the populist, while opposed to the expansion of individual liberties, does support government economic intervention.

Following earlier researchers, we recode the answers to each of the six questions (government guarantee of jobs, government health insurance, progressive taxation, legalization of marijuana, abortion, and equal role for women) into three categories: support, opposition, and centrist. A consistent liberal supports government action on all three economic issues and supports expansion of individual liberties on all three of those issues; the consistent conservative takes the opposite position on all six issues. The consistent libertarian opposes all three government actions in economic affairs and supports all three individual-liberty proposals. The consistent populist supports all three economic interventions but opposes all three of the expansions of individual liberties.

Because of the complex nature of our measure, only respondents who expressed an attitude on four or more of the six issues (90 percent of the sample) are considered for classi- fication as consistent, although the percentages of all tables are based on the total sample. Our findings support the earlier conclusions that perfect consistency is rare in the American public; the proportion we find as consistent (7.3 percent of the sample) is lower than the 17.8 percent found by LeBlanc and Merrin (1977), probably because our definition of consistency involves two dimensions.

LeBlanc and Merrin suggest that to measure perfect consistency is "too stringent a test" and use the term "nearly consistent" for those who deviate from consistency on only one issue or whose attitudes include liberal and centrist or conservative and centrist positions. (18) We follow the same strategy and define the "nearly consistent" as those in each category who are inconsistent on only one issue on each dimension. The nearly consistent populist, for example, takes the populist position on at least two of three economic intervention issues and on at least two of the three individual-liberties issues. The "deviant" attitudes may be a response in the non populist direction, a centrist response, or no response at all. Table 1 presents the distribution of the totally consistent ideologues, the nearly consistent, and the results of combining the consistent with the nearly consistent respondents.

We find that 72.6 percent of the sample can be classified as consistent or nearly consistent. The "traditional ideologies" of liberal and conservative do not appear to explain more belief

Table 1

Attitude Consistency in the American Public: 1976

Totally Consistent
Nearly Consistent
Hold fewer than 4 of 6 attitudes (inattentive)


NOTE: The data in this table and those that follow were made available by the Inter-University Consortium for Political Research. Neither the Center for Political Studies nor the consortium bears any responsibility for the analysis or interpretation presented here. systems than do the two additional orientations. The populist group is the largest, followed by the conservatives, liberals, and libertarians. Many of the "inconsistents" so long lumped together as a group in fact can be classified according to belief systems measured by a more complex standard than that used to identify only liberals and conservatives in previous research. Rather than a mass public consisting of a few ideologues and many inconsistents, we find a much more divided and diverse public.

Although LeBlanc and Merrin (1977) found that attitude consistency was related to expressing fewer attitudes, we find that this is true only for the populist group, whose proportion of the sample declines from 58.6 percent of those holding only four attitudes to 16 percent of those expressing six attitudes. For the other three ideological groupings, however, the reverse is true. Their representation is actually greater among those who express more attitudes. (19) Note, however, that all respondents categorized as holding a type of belief system expressed opinions on at least four of the six issues.

Social Groups and Belief Systems

Table 2 summarizes a demographic analysis of our four types of ideologues. In general, the demographics reflect the scholarly and practical assumptions about the relationship of belief systems and social groups. Of particular note are the relationships of our categories to income, education, age, and region. Analysis of place of residence, length of residence, religion, union membership, sex, and subjective social class revealed few differences between ideological types.

Income. Predictably there is a strong relationship between income and ideological type. The populist category virtually preempts the other ideologies in the lowest income category and is barely found in the highest income group. Part of this tendency may be explained by the dominance of populism among nonwhites, who also make up much of the lower income categories. The libertarian category is the largest for the two highest income groups, representing 33 percent of those with an income of $35,000 and over. Distinctions that might be overlooked or obscured with a liberal-conservative continuum are obvious in this table. For example, the high proportion of libertarians among the wealthy may in part explain the "inversion" of the class base of liberalism and conservatism analyzed by Ladd and Hadley (1978). Many of the issues that Ladd and Hadley use to indicate that upper income strata are becoming more liberal are in fact issues of individual liberty -- abortion, marijuana, and sexual relationships, for example.

Table 2

Demographic Characteristics of Ideological Types*

  Populist Conservative Liberal Libertarian Divided
National (N) 24% (565) 18% (425) 16% (392) 13% (313) 18% (442)
0-5,000 43 9 10 6 12
5-10,000 29 15 18 9 19
10-15,000 21 21 17 12 20
15-20,000 15 23 18 17 21
20-25,000 12 24 18 15 26
25-35,000 9 23 21 30 14
35 & Above 7 22 15 33 18
White 22 19 15 14 19
Nonwhite 38 4 25 4 11
Grade School 43 15 7 3 9
High School 27 19 13 11 20
Some College 12 18 23 19 23
College Degree 8 17 30 20 20
Advanced Degree 8 17 26 33 13
18-25 17 10 27 21 19
26-30 17 16 24 15 18
31-35 16 18 18 16 22
36 & over 28 20 11 10 28
Northeast 25 14 21 12 19
Midwest 22 22 14 15 19
South 29 17 12 9 15
West 14 16 22 17 23
Ideological Self-Classification**
Liberal 11 6 43 18 16
Moderate 20 27 23 34 34
Conservative 22 44 13 37 29
No Answer 47 23 20 11 21

* Percentages add across the rows except where noted (**). Inattentives and the small centrist categories are not listed here. Therefore, the percentages add up to less than 100 percent.

** Percentages add down the columns.

Education. As can be seen from table 2, education relates strongly to ideological type. The most obvious relation- ship is the predominance of the populist category among those with grade-school and high-school education, and the relatively small percentage of populists beyond high school. Conservatives are drawn about equally from each of the educational groups, while the proportions of liberals and libertarians increase quite significantly with higher levels of education. For those with college and advanced degrees, these two types represent about half the group.

Age. The data show interesting generational differences. If we look at the period in which people in each age category reached political maturity, the differences may be explained in terms of the political "milieu" of the time. Among the oldest group, many of whom matured during the depression, the predominant categories are populist and conservative. This reflects the New Deal conflicts, which were primarily about questions of government intervention; the expansion of individual liberties was not a primary division.

The second age group came to political maturity in the late fifties, often characterized as the period of "the end of ideology." People in this age group are clustered remarkably close together -- our four ideological categories are within 5 percent of each other in size. The next age group matured during the early sixties, a period symbolized by John Kennedy and a liberal sense of confidence and mission. This age group is distinguished by a greater number of liberals than the older groups, with the other ideological categories remaining about the same size.

Our youngest group reached maturity in the mid- seventies, with a general disillusionment with government, but also an increased concern with the "social issues" that relate to the individual-liberties dimension. This younger group seems to be distinguished by a commitment to individual liberties in that the two largest categories are liberal and libertarian. Thus this youngest group is divided over the effectiveness of govern- ment economic intervention, but not over a commitment to individual liberties, whereas the oldest group is divided over government intervention but is in agreement that individual liberties should not expand. Region. As one would expect, given the historical and economic factors in Southern politics, Southerners are more heavily populist than any other category, with conservative being the second largest category for that region. The South has a few liberals or libertarians. In the Midwest, conservatism is represented equally with populism while in the Northeast populism is the largest category with liberalism a close second. Perhaps the most interesting region is the West, which consists mostly of California residents in this sample. Here the largest category is the divided one, and no ideological category stands out as particularly large. The difficulty of building a coalition here is obvious. Some of the more bizarre aspects of California politics may reflect this lack of ideological agreement in the region.

The Political Behavior of Ideological Types in 1976

As Stimson (1975) argued, the identification of types of belief systems should be supplemented by a discussion of the behavior of those types of citizens. Do they behave in patterns that indicate that the belief system is in some way useful to them? Can we make predictions about the behavior of those in different belief categories? With the fourfold classification we have presented, these questions become even more complex. While the political world may be relatively well defined for liberals and conservatives, populists and libertarians may find the definition of American politics provided by the media more difficult to match with their own perspectives. We examine how these four types of ideologues respond to four major political choices: the liberal-conservative continuum as a means of self-classification, party identification, presidential vote, and the evaluation of political candidates.

Ideological Self-Classification. The comparison of the four ideological groups' responses to the liberal-conservative self-placement scale (trichotomized with the no-answer group presented separately) is at the bottom of table 2. The figures here add down the columns: We are interested in how members of each group classify themselves. The results generally confirm expectations. Almost the same proportion of liberals and conservatives correctly label themselves on this scale. In both cases about one-quarter choose to take the moderate way out, thus avoiding either of the two extreme labels. However, we cannot expect a perfect correspondence between classification based on issue positions and self-classification. (20) Just because a respondent holds a set of attitudes that analysis would describe as liberal, for example, does not mean that he understands the accepted definitions of that term or that he himself uses the same definition. Furthermore, in responding to the liberal-conservative scale, the person may be reacting more to the label and its connotations than to its representation of a set of issue positions. Some liberals, for example, may not wish to use the label (as in the 1976 campaign, when some liberal candidates or president suddenly became "progressive") because of negative connotations of big government and wasteful spending. Furthermore, Holm and Robinson suggest that some people in fact have an ideological framework for understanding politics but are not conscious of it. (21) The liberals and conservatives found here who choose the moderate or no-response option may be such people.

Self-classification is impossible for almost half of the populists; they have no obvious alternatives. Of those populists who do choose a label, almost four times as many choose either conservative or moderate as choose liberal, perhaps reflecting a hostility to the connotations of that term, despite the fact that they support major liberal economic initiatives. Libertarians do make a choice; their 11 percent non response is the lowest of any group (probably reflecting their higher education levels). Furthermore, they are more likely to choose conservative (37 percent) or moderate (34 percent) rather than liberal as their label. Perhaps the word "liberal" for populists too strongly suggests changing social values or welfare spending. "Liberal" may remind libertarians chiefly of government spending and increased taxation.

Party Identification. Table 3 presents evidence on the partisan choice of the four major belief systems. Again, the percentages add down, indicating the distribution of each ideological group across the three partisan choices. Libertarians are the group most likely to be independent of the parties. Some libertarians may reason that Republicans are the most likely to minimize government economic intervention and yet not drastically alter the state of individual liberties. Others, particularly the large group of younger libertarians, may reason that both parties are capable of deficit spending and increased government activity, but the Democrats are closer to their position on questions of individual liberties and changing lifestyles.

The strong Democratic identification among populists (47 percent) may reflect the appeal of Democratic candidates -- for example, Jimmy Carter in his support of economic intervention for the disadvantaged combined with lukewarm support for liberalization of individual liberties; or George Wallace, the candidate most often labeled a populist in recent years. Further, the endurance of party identification from its early formation and the fact that party identification may lag behind changing issue attitudes and realignments (22) suggests that populists, who may have formed partisan ties with the Democrats in earlier years when divisive social issues were not present (see our earlier discussion of age), should choose Democratic affiliation most often. The strength of candidates like Carter and Wallace in the presidential primaries of recent years is certainly not surprising given this finding. We must not overlook the substantial number of populists who call themselves independent and Republican, however. For someone with a populist set of issue positions, the choice between the two parties is a difficult one to make.

Liberals also are strongly Democratic (47 percent), but nearly as many identify themselves as independent. The conservative group is most evenly split across partisan choices, although the

Table 3

Relationship of Ideological Type to Party Identification, Controlling for Education, Income, and Race

(N) Libertarian (313) Conservative (425 Liberal (392) Populist (565)
Party Identification        
Democrat 23 31 47 47
Independent 42 32 41 30
Republican 35 36 12 21
Grade School (16) (56) (25) (163)
Democrat 20 39 62 60
Independent 40 21 14 19
Republican 40 37 14 18
High School (129) (220) (153) (318)
Democrat 25 38 50 45
Independent 45 30 41 35
Republican 30 31 9 19
College (175) (147) (215) (83)
Democrat 21 16 43 30
Independent 40 40 44 34
Republican 38 45 13 34
Low Income (69) (108) (133) (327)
Democrat 37 41 53 53
Independent 41 21 35 25 41 21 35 25
Republican 23 37 12 20
High Income (232) (294) (235) (205)
Democrat 20 28 45 37
Independent 41 38 42 39
Republican 39 34 12 22
Whites (298) (408) (321) (457)
Democrat 21 30 42 42
Independent 43 32 44 31

Republican 36 38 14 25

36 38 14 25

Republicans are strongest. The failure of Republicans to tap their "natural constituency" is indicated by the one-third of the conservatives who choose the independent label and the one-third who choose the Democratic party.

The relationship between ideological views and partisan identification may be contaminated by demographic factors, however. While there are some demographic differences in the degree to which ideological types identify with the parties or remain independent, we find in table 3 that the same pattern exists for most demographic categories. First, Democrats in the total sample are the strongest among liberals and populists; they are less well represented among conservatives and do poorly among libertarians. That same pattern exists among Democrats with grade-school or high-school education, among both Democratic income groups, and among white Democrats. Only when we compare ideologues among Democrats with college education do we find a slight variation. Democrats are stronger among libertarians than among conservatives, but, even there, liberals include more Democrats than does the populist group, and the populist group in turn is more Democratic than the two remaining groups. Secondly, independents in the total sample are strongest among libertarians and liberals, followed by conservatives and populists; subgroups of independents show some variations from this pattern. Among the lowest education group of independents, all types are notice- ably weaker except the libertarians. Among the high school educated group and those with low incomes, populists are stronger than conservatives among independents, hut otherwise the national pattern is repeated. Among independents with college education, the liberals, libertarians, and conservatives are virtually even, and in the high income group all four types are virtually even. Finally, Republicans are most strongly represented in the conservative and libertarian groupings in the total sample, followed by populists and liberals last. For all of the control groups except one, this same pattern is replicated. Among the high income segment, Republicans are slightly more represented among libertarians than among conservatives, but generally speaking the relationship between ideological orientation and Republican identification holds constant even when controlling for demographic factors.

Presidential Voting. In the 1976 election, most voters perceived Carter as a liberal and Ford as a conservative. (23) To the extent that liberals and conservatives share the majority perception they should have been able to make a clear choice. For populists and libertarians the problem was not just that the two candidates did not reflect their views on major issue dimensions but that the definition of the candidates was generally in liberal-conservative terms. Populists, however, could have responded to Carter because he de-emphasized lifestyle issues and focused on economic issues and trust in government. (As Miller 1978 points out, Carter tended to defuse those issues that had divided the Democrats in 1972.) Their tendencies toward Democratic identification obviously would have propelled many into Carter's column. For libertarians, many of whom are independents, the presidential choice was between two candidates whose pronounce- ments on questions of individual liberties did not greatly differ, although Carter probably more clearly supported women's rights and showed less enthusiasm for a "hard line" against legalization of marijuana and abortion. On the other hand, libertarians may have found Betty Ford's feminism and tolerance of changing lifestyles more appealing than Carter's Southern Baptist image. On the economic dimensions, however, Ford obviously opposed government intervention more than did Carter, whose campaign return to traditional Democratic liberalism may have muted his "un-Democratic" conservative economic views, which were noticeable earlier in the year. Thus a choice for Ford would have been the most defensible for the libertarian voter in 1976.

The proportion voting for Carter presented in table 4 indicates that our expectations are borne out. Sixty-six percent of liberals voted for Carter, slightly ahead of the 59 percent given him by populists. Conservatives gave him only 32 percent of their votes and libertarians supported Carter even less. The support for Carter among "self-identified" liberals is higher (79 percent) than the Carter support among the liberals as we have defined them. Similarly, the Ford vote among self-identified conservatives is higher than any conservatives as defined by our two-dimensional typo logy. We do not see this as a refutation of our classification, however. Those classified as liberals or conservatives in our method face a much more stringent test of classification than those classified by self-identification. Furthermore, the vote of "liberals for Carter" and "conservatives for Ford" by self-classification may mean little more than voter response to ideological labels that are repeatedly presented to them throughout a campaign, as both Field and Anderson (1969) and Pierce (1970) suggest happened during the 1964 campaign. Finally, our fourfold classification allows us to make predictions about two other groups, populists and libertarians, that are for the most part buried in the mountain of moderates and don't-knows of the self-classification scale.

We do not claim that membership in one of these ideological groupings is by itself the best predictor of a presidential vote. We suggest that a person's belief system may in some cases operate in conjunction with his partisanship to reinforce voting decisions or, in other cases, lead to defection. Further, the belief system should serve as a good predictor of presidential vote among self-identified independents. For evidence on this point, the lower part of table 4 presents the percentage voting for Carter in each ideological group while controlling for identification. As we expect, there are clear

Table 4

Relationship of Ideological Type and Vote for Carter Controlling for Education, Income, and Race
  Libertarian Conservative Liberal Populist
Vote for Carter 30% 33% 66% 59%
By Party Identification
Democrat 61 (52) 71 (105) 85 (143) 84 (176)
Independent 30 (91) 29 (94) 52 (116) 45 (107)
Republican 13 (95) 7 (132) 26 (31) 25 (93)
Grade School* 47 38 84 71
High School 37 39 68 57
College 25 26 63 44
Low Income 38 46 77 63


High Income

27 30 61 51
Whites 28 32 62 54

*N's for remaining categories are same as in table 2.

differences attributable to partisanship. All Democrats vote heavily for Carter regardless of ideology; all Republican identifiers supported Ford at high rates despite ideological classification. However, the voting tendencies of ideological groupings do appear even when party is controlled. Among Democratic identifiers, the highest rates of defection are among libertarians and then conservatives, while liberals and populists supported Carter at the highest rates. Among Republicans, more defections from Ford occur in the liberal and populist groups; Ford held 93 percent of the conservatives and 87 percent of libertarians. Furthermore, the figures among independents are similar in direction to those of the total sample. Party identification obviously was a better single explanation of presidential vote choice in 1976, but our ideological classi- fication offers an additional set of evidence about partisan defection rates and about the behavior of independents. The remainder of table 4 presents the vote for Carter by ideological classification, controlling for education, income, and race. Here we find that, although there is some variation in the magnitude of support for Carter across ideological groups when demographic factors are controlled, the same general directions are present in almost every case: The order of Carter's support goes from liberal, populist, conservative, to libertarian. Only in the case of grade-school-educated respondents do we find any variations, but the small cell size here may account for the findings.

Another means of testing the utility of our typo logy is to examine voter perception of the candidates' standing on the issues. The 1976 survey asked respondents to evaluate the presidential candidates' positions on the three economic- intervention questions and two of the individual-liberties questions. (No question was asked about candidate stand on abortion.) The data in table 5 indicate that, in general, the four ideological types evaluate candidates consistent with our expectations. Liberals and populists who voted for Carter generally perceived Carter as being more favorable to government economic intervention, while conservatives and libertarians who voted for Carter perceived him as being more opposed to these government actions. When we turn to two individual-liberties questions we find a different set of perceptions. Here liberals and libertarians who voted for Carter tend to perceive his position similarly while conservative and populist Carter sup- porters saw him as being more opposed to expanding individual liberties. Thus for Carter supporters the fourfold classi- fication provides an explanation for voter perceptions of the candidate that the liberal-conservative continuum by itself could not.

For Ford voters the patterns are not quite as clear. On government health insurance and government jobs, liberal and populist Ford voters perceived Ford as more in favor of government intervention, while conservatives and libertarian Ford voters saw

Table 5

Perception of Candidate Issue Positions* by Their Supporters, According to Ideological Type

  Carter Voters' Perception of Carter Ford Voters' Perception of Ford
Economic Issues
  Lib. Pop. Cons. Lbt. Lib. Pop. Cons. Lbt.
Progressive Taxation 2.91 3.28 3.73 4.13 4.05 3.84 4.33 4.40
Government Health Insurance 2.94 3.29 3.29 3.69 3.72 3.40 4.83 5.10
Government Jobs 3.13 2.97 3.61 3.50 4.50 3.69 5.04 4.93
Individual-Liberties Issues
  Lib. Pop. Cons. Lbt. Lib. Pop. Cons. Lbt.
Legislation Marijuana 4.40 3.90 4.97 5.37 4.21 4.57 4.98 5.35
Equal Role for Women 2.62 2.82 3.63 3.41 3.17 3.64 3.59 3.66


*Figures shown are the mean score for each group's perception of the candidate's positions on a 1-7 scale. The lower the score the more the candidate is perceived as supporting the policy in question him as being opposed. On progressive taxation, the distinctions in perceptions of Ford were minor and not consistent with our expectations. On legalization of marijuana, most voters saw Ford as opposing such a move, although the libertarians and populists clearly saw Ford's position quite differently. On women's rights, the Ford libertarians perceived Ford as being slightly favorable, while the other three groups perceived Ford's position as barely on the favorable side. In summary, the perception of the candidates by their supporters generally is predictable by a supporter's belief system. By a simple liberal-conservative distinction, liberals and conservatives should represent two opposing views of the candidates' positions. In fact, we find that liberal and populist voters perceive their candidates similarly on economic issues, with conservatives and libertarians' perceptions of the candidate clustered together in the opposite direction. When we look at individual-liberties issues, we find that libertarians and liberals perceive the candidates similarly (especially for Carter voters), while conservatives and populists share a different set of perceptions.


We have presented an alternative approach to the analysis of mass belief systems, showing that it is useful to conceptualize mass belief systems in terms of two dimensions -- government intervention in economic affairs, and expansion of individual liberties -- rather than in terms of the traditional liberal- conservative continuum. Using these two dimensions, we defined four ideological categories, which we label liberal, conservative, populist, and libertarian. We believe our approach more accurately reflects the complexities of the philosophical traditions of liberalism and conservatism as well as the realities of contemporary politics. The interrelationships of economic freedom and personal freedom have been much debated in the development of western political thought in the last two centuries; our use of a two- dimensional approach to mass belief systems more accurately reflects this debate. In addition, the literature analyzing elite behavior -- Congress and courts, for example -- often is forced beyond the single liberal-conservative dimension. We see our approach as a logical extension of previous research, rather than a totally new departure.

By clarifying the liberal and conservative labels and adding two new categories, we are able to more completely explain the behavior of the American electorate than can be done with a unidimensional approach. We find tendencies for our four groups to behave differently (even after controlling for various demo- graphic factors) in such areas as party identification, presi- dential vote, and evaluation of candidates' issue positions. Further, we can speculate that the existence of two major groups of people who hold political beliefs for which the traditional language and labels of American politics provide more confusion than clarity has long-range implications for the political system.

For example, growing evidence suggests that the traditional coalitions of the two parties are changing, that the number of independent identifiers is increasing, that split-ticket voting is growing, and that in general the electorate is more volatile and less predictable now than in the immediate past. Evidence of dissatisfaction with political parties (among other institutions) is also plentiful. While these changes relate in part to the impact of television, changing campaign styles, improved education, and -the cumulative impact of the perceived failures and scandals of political institutions in recent years, they may also be related to the presence of two ideological groups in society whose belief systems are not reflected by the Democratic or the Republican party nor by their candidates.

If our analysis is correct, the major parties and their candidates will have to deal with the presence of at least four, rather than two, ideological groupings in the American electorate. And those who seek to predict or explain voter behavior including voter apathy will especially have to recognize these four distinct groups.


Campbell, Angus; Converse, Philip; Miller, Warren; and Stokes, Donald. 1960. The American Voter. New York: Wiley.

Claus en, Aage. 1973. How Congressmen Decide: A Policy Focus. New York: St. Martin's Press.

Converse, Philip E. 1964. "The Nature of Belief Systems in Mass Publics." In Ideology and Discontent, ed. David Apter. New York: Free Press.

_____; Clausen, Aage; and Miller, Warren. 1965. "Electoral Myth and Reality: The 1964 Election." American Political Science Review 59 (1965): 321-26.

_____; Miller, Warren E.; Rusk, Jerrold G.; and Wolfe, Arthur C. 1969. "Continuity and Change in American Politics: Parties and Issues in the 1968 Election." American Political Science Review 63 (1969): 1083-1105

Field, John Osgood, and Anderson, Ronald E. 1969. "Ideology in the Public's Conceptualization of the 1964 Presidential Election." Public Opinion Quarterly 33 (1969): 380-98.

Free, Lloyd A., and Cantril, Hadley. 1967. Political Beliefs of Americans. New Brunswick, N.J.: Rutgers University Press.

Holm, John D., and Robinson, John P. 1978. "Ideological Identification and the American Voter." Public Opinion Quarterly 42 (1978): 235-246.

Knoke, David. 1979. "Stratification and the Dimensions of American Political Orientations." American Journal of Political Science 23 (1979): 772-91. Kritzer, Herbert M. 1978. "Ideology and American Political Elites." Public Opinion Quarterly 42 (1978): 484-502.

Ladd, Everett Carl, and Hadley, Charles D. 1973. "Party Definition and Party Differentiation." Public Opinion Quarterly 37 (1973): 21-34.

_____. 1978. Transformations of the American Party System. 2d ed. New York: Norton.

LeBlanc, Hugh L., and Merrin, Mary Beth. 1979. "Parties and Candidates in 1972: Objects of Issue Voting." Western Political Quarterly 32 (1979): 59-69.

_____. 1977. "Mass Belief Systems Revisited." Journal of Politics 39 (1977): 1082-87.

Levi tin, Teresa E., and Miller, Warren E. 1979. "Ideological Interpre- tations of Presidential Elections." American Political Science Review 73 (1979): 751-771.

Lip set, Seymour Martin. 1959. Political Man. Garden City, N.Y.: Doubleday.

Luttbeg, Norman. 1968. "The Structure of Beliefs among Leaders and the Public." Public Opinion Quarterly 32 (1968): 398-409.

MacRae, Duncan, Jr. 1958. Dimensions of Congressional Voting: A Statistical Study of the House of Representatives in the Eighty- First Congress. Berkeley: University of California Press.

May hew, David R. 1966. Party Loyalty among Congressmen. Cambridge: Harvard University Press.

Miller, Arthur H. 1978. "The Majority Party Reunited? A Comparison of the 1972 and 1976 Elections." In Parties and Elections in an Anti-Party Age, ed. Jeff Fishel. Bloomington: Indiana University Press.

_____, and Miller, Warren E. 1975. "Issues, Candidates, and Partisan Divisions in the 1972 American Presidential Election." British Journal of Political Science 5 (1975): 393-434.

Nie, Norman H., and Anderson, Kristie. 1974. "Mass Belief Systems Revisited: Political Change and Attitude Structure." Journal of Politics 36 (1974): 540-91.

_____; Verba, Sidney; and Petrocik, John P. 1976. The Changing American Voter. Cambridge: Harvard University Press.

Pierce, John C. 1970. "Party Identification and the Changing Role of Ideology in American Politics." Midwest Journal of Political Science 14 (1970): 25-42. Pomper, Gerald M. 1975. Voter's Choice. New York: Dodd, Mead.

Rokeach, Milton. 1973. The Nature of Human Values. New York: The Free Press.

Scammon, Richard, and Wattenberg, Benjamin. 1970. The Real Majority. New York: Coward McCann and Geoghegan.

Schubert, Glen don. 1965. The Judicial Mind: The Attitudes and Ideologies of Supreme Court Justices 1946-1963. University Press.

Stimson, James A. 1975. "Belief Systems: Constraint, Complexity, and the 1972 Election." American Journal of Political Science 19: 393-417.

Stouffer, Samuel. 1955. Communism, Conformity and Civil Liberties. Garden City, N.Y.: Doubleday.

Truman, David B. 1959. The Congressional Party: A Case Study. New York: Wiley.

Turner, Julius. 1951. Party and Constituency. Baltimore: Johns Hopkins Press.


(1) Campbell et al. 1960, Converse 1964, Field and Anderson 1969, Pierce 1970. (2) Converse 1964, Nie and Anderson 1974. (3) Stimson 1975, Kritzer 1978. (4) These were also used by Holm and Robinson 1978 in a different context. (5) Luttbeg 1968, Stimson 1975, and Kritzer 1978. (6) Converse et al. 1965, Converse et al. 1969, Miller and Miller 1975, and Miller 1978, for example. (7) Holm and Robinson 1978, Levitan and Miller 1979. (8) LeBlanc and Merrin 1979, p. 61. (9) Nie, Verba, and Petrocik 1976, p. 138.

(11) Knoke 1979, p. 772.

(12) Ladd and Hadley 1978. For a summary of this literature and a multi variate analysis of this relationship, see Knoke 1979.

(13) Miller and Miller 1975, Stimson 1975, Pomper 1975, for example.

(14) Turner 1951; MacRae 1958; Truman 1959; Mayhew 1966.

(15) Clausen 1973, p. 31

(16) LeBlanc and Merrin 1979, p. 61.

(17) See, for example, Knoke 1979.

(18) LeBlanc and Merrin 1977, pp. 1063-64.

(19) Data analysis relevant to this point is not presented here, as it is not our major concern. This information is available on request from the authors.

(20) See Free and Cantril 1967.

(21) Holm and Robinson 1978, p, 237.

(22) See Ladd and Hadley 1973, for example.

(23) Miller 1978.

Published by the Cato Institute, Policy Analysis is a regular series evaluating government policies and offer- ing proposals for reform. Nothing in Policy Analysis should be construed as necessarily reflecting the views of the Cato Institute or as an attempt to aid or hinder the passage of any bill before Congress. Contact the Cato Institute for reprint permission. Additional copies of Policy Analysis are $4.00 each ($2.00 in bulk). To order, or for a complete listing of available studies, write to: Policy Analysis, Cato Institute, 1000 Massachusetts Avenue NW, Washington, D.C. 20001

1981 The Cato Institute
Please send comments to webmaster