Resumé

We examine how clearly citizens in Norway and the United States perceive the ideology and issue positions of parties and candidates. Following in the tradition of Campbell & Valen (1966) and Granberg & Holmberg (1988), we ask: Does the political system matter? We also consider the role of individual sophistication and whether sophistication has different consequences in the two cultural settings. Across a broad range of issues, we find that citizens in Norway have a sharper view of party positions than those in the United States. Sophistication influences the clarity of issue perception in both countries. However, the hypothesis that sophistication matters more in the US is not well supported.

Introduction

The issue and ideological commitments of parties and candidates provide policy meaning to the outcome of elections. In order to play this role, these commitments must be transmitted to the electorate. Even if every citizen voted strictly on the basis of issues, great variation in the perception of party positions would make the policy consequences impossible to discern. In contrast, a good deal of consensus on where the parties stand permits elections to serve as meaningful determinants of future policy.

How clearly do individuals perceive the issue positions of political parties and candidates? Our investigation will focus on Norway and the United States. We will be interested in seeing how the different political systems influence perceptual clarity across issues. We will also be concerned with differences across sophistication groups within countries, and the possible interaction between individual and systemic characteristics in determining how well party positions are understood.

Our work follows in the tradition of Campbell & Valen (1966) and Granberg & Holmberg (1988) in its concern with systemic effects on issue perception. When Campbell and Valen pursued their work in the early sixties, however, surveys did not ask individuals to place parties and candidates on issues. When Granberg and Holmberg compared Sweden and the United States, they investigated just two issues in each country. Our study will examine voter perceptions of parties and candidates over a much wider range of political issues.

The earlier studies pointed to sharper views of party issue positions in Scandinavia than in the United States. We anticipate replicating that finding here - the systems are different, and the disciplined multiparty system of Norway should provide more consistent and clearer cues to the electorate than the US system does. But perhaps the differences will be more muted than Campbell and Valen found, because some of the key factors have changed. In Norway the demographic basis of party choice has eroded considerably over the last three decades, as has the partisanship in the media (Listhaug 1989, n.d.). With less social solidarity and less overt partisanship in the media, cues about party position might be less readily obtained than in the past. Nevertheless, when one considers the rather chaotic candidate-centred politics of the US, it seems unlikely that voters in Norway would operate under the same level of ignorance as their American counterparts.

At the individual level, sophistication has been shown to be an important influence on political knowledge (Campbell et al. 1960; Luskin 1987). In a comparative context the interesting question is how sophistication and the political system interact. We suspect the differences between the two systems will be greatest among the least sophisticated; for these voters the environment is likely to have a marked impact on what knowledge they acquire. In contrast, we anticipate that the most sophisticated voters are likely actively to seek information, and thus the mediating role of the system will be less consequential.

Data

A serious problem in cross-national research is whether we are really comparing comparable issues and issue question formats. While no ideal solution exists, our approach is to select for the United States all the questions following the seven-point issue format on the 1988 NES presidential election study and for Norway all the questions following a similar ten-point format on the 1989 parliamentary election study. The virtue of this approach is that it allows comparison across a broad set of issues - issues that researchers within each country thought were important. This

is distinct from the Granberg-Holmberg approach which focused on a small subset of issues where each response represented a distinct policy option. The two approaches are best viewed as complementary and together should provide a richer picture of differential perceptions across cultures.

Both the American and Norwegian issue questions presented two policy alternatives - one at each end of a scale - and then asked the respondents to position themselves and the parties and candidates on the scale. Questions addressed in the US concerned liberal-conservative ideology, provision of government services, defence spending, government health insurance, government provision of jobs, the best way to deal with the Soviet Union, government aid to minorities, government aid to blacks, and the appropriate role for women in society. Questions in Norway concerned left-right ideology, agricultural support, environment versus growth, immigration policy, health care privatization, alcohol policy, and punishment for violent crime. The only exception to the format of the questions was that the ideology measure in the United States presented a specific cue for each point on the scale. The wording of the issue questions for the US is available in the 1988 NES codebook, and that of the Norwegian questions can be found in the Appendix to this article.

At the outset, we must confront the fact that the scales had different ranges in the two countries. There is inherently more variance in a scale with ten points than a scale with seven points, so to make the scales more comparable we transformed the values to give the Norwegian questions the same 1-7 range. When appropriate, we also reflected the scales to make them positively correlated with left-right (or liberal-conservative) ideology. Finally, we centred the scales so that a value with a minus sign designates a position on the "left" side of the issue and a plus sign designates a position on the "right" side. This gives the scales in both countries a —3 to +3 range.

We do not claim that these transformations make the two sets of questions identical. Even if scales with the same number of points had been used in the surveys, it would not be reasonable to argue that they assess the exact same thing because the issues themselves are different in the two societies. The difference in the scales thus serves to reinforce the point that comparisons should be made with caution. In our analysis we will follow two strategies. The first is to observe the extent of variation in the perceived positions of parties and candidates. For this to be meaningful we need to have some faith that the questions are broadly comparable across the two countries. The second strategy is to compare the variation in the placement of the parties to the variation in individual preferences. In general, if the questions were well selected by researches, we would anticipate real differences across citizens in how they stand on the issues. This need not be the case for the perception of a party or candidate. In the ideal-typic

case, opinion in the electorate would vary considerably while citizens would be in complete accord on where the parties stand. By examining patterns of relative variation in party placement compared to mass opinion, we will gain some leverage in assessing systemic differences.

Results

Let us begin our discussion with political ideology. Left-right and liberalconservative are the most substantively comparable questions on the two national surveys. Table 1 shows the mean and variance of respondent position and of the perceived position of the parties and candidates on ideology.

Norwegians have a clearer perception of where the parties stand on ideology than do their American counterparts. For example, George Bush has the most clearly perceived ideological position among the American objects, with a variance of 2.02. We can compare this to the least clearly perceived party position in Norway - 1.60 for the Liberals. The ratio of the two variances is approximately 1.25. Under the assumption of normality, the ratio of the variances conforms to an F distribution and can be assessed using a standard F test. This ratio is statistically significant at the 0.01 level. Since this is the smallest of the possible ratios, any comparison we made would show a significantly sharper perception in Norway.

These cross-country differences do not result from a fundamental difference in ideological consensus. Both countries, in fact, show rather similar variation in respondent position on ideology. Thus there is little question that with regard to ideology Norwegians have a clearer sense of where the parties stand.

The pattern observed for the ideological placements is not entirely replicated for the political issues. Ideology is notoriously ill formed and poorly understood in the United States (Converse 1964; Conover & Feldman 1981); hence that comparison was somewhat loaded in favour of more consistent imagery in Norway. An examination of the full set of issues (presented in Tables 2 and 3) reveals some instances in which the perception of a party or candidate in the US displays less variance than the perception of a party in Norway. Examples of relatively clear positions in the United States are Bush and the Republican Party on defence spending and the Democratic Party on aid to black people and relations with Russia. In Norway the Socialist Left Party is perceived fairly diffusely on agricultural support and environment versus growth and is joined by the Progressives on the latter issue. On the whole, however, the variances tend to be smaller in Norway. A simple count shows that the Norwegian parties have a variance below 2.0 more than 85 percent of the time (37 out of 42), while

this occurs for the parties and candidates in the US only 22 percent of the
time (10 out of 45).

Comparing absolute variances requires confidence that the issue questions are tapping similar levels of controversy in the two countries. Looking at the variances in the self-placements, we can see that the level of consensus varies from issue to issue. Interestingly, the most controversial issue in each country concerned health care. Punishment of violent crime stands out as an issue on which the Norwegian public was in great agreement. There is no issue of that sort on the US survey. To take into account the differences in issue consensus, we use a measure of "relative variance". This measure compares the clarity of the perceived party and candidate stands to the general level of social agreement on the issue.¹ With the exception of the crime issue, the relative variances are considerably smaller in Norway than in the US. If we take the value of 0.5 as a marker, we find that in Norway the party variance is less than half the respondent variance in 57 percent of the instances (24 out of 42). In the US a relative variance below 0.5 occurs only once out of 45 possible times - for George Bush with regard to medical insurance.

Party and Candidate Comparisons

Summarizing across issues, we find that every party in Norway displays lower variance in perception than any of the parties or candidates in the United States. This is not quite true of the relative variances. (See Table 4, which averages the variation over all issues including ideology for each of the political objects.) Citizens in Norway show the greatest uncertainty about the issue positions of the Socialist Left Party. Its relative variance of 0.67 is higher than that of either the Democratic or Republican Party or that of George Bush. But each of the remaining Norwegian parties has a relative variance substantially lower than that of the American parties or candidates. If we compare the Democrats with the Labour Party and the Republicans with the Conservative Party - the most similar parties in the two countries - the results suggest that party images are about 25 percent more diffuse in the US.

In Norway we have perceptual data only for the parties, but the US data include four candidates as well. Two of the candidates were the nominees for president in 1988 (Bush and Dukakis), and the other two were salient figures in the opposing parties (Jackson and Reagan). Within the US the variances follow a clear pattern. The two parties show the smallest perceptual variation, followed by the presidential nominees, followed by Jackson and Reagan: the differences between the two parties and their presidential nominees are small, particularly those between Bush and the Republican Party. The relative clarity of the party perceptions is

noteworthy. Given the wide variety of candidates who run under a party label, one could well imagine that the parties would project quite varied images compared to candidates who are single actors with single images; but that is not the case here. Of note, too, is the large variation in the perception of Reagan, especially since the survey was conducted as he was completing his eighth year in the White House.

Within each country the clarity of perceptions could be influenced by a tendency for more extreme parties and candidates to display higher variation. We can examine this possibility by looking at the values for "extremity" in Table 4. Extremity is the average squared distance from the centre point on the scale for each party and candidate.² In the US, variation in perception appears to be somewhat related to object extremity. In Norway, though, the most extreme party - the Progressive Party - does not show particularly high variance in perception.

The Role of Sophistication

Political sophistication impacts an individual's ability to follow and interpret political cues. We would expect then that sophistication would influence the clarity of issue perception of parties and politicians. In her analysis of liberal-conservative positions, Powell (1989) found a clear tendency for the mass public to underestimate the extremeness of congressional candidates because a large number of respondents guess in placing the candidates. Following the Powell logic, we expect two results as we examine perceptions by level of sophistication. First, we imagine that as voters become more politically aware their perceptions of party and candidate stands will be more sharply focused; second, we imagine that their placements of parties and candidates will tend to be more extreme.

Beyond this, we are interested in the potential interplay of the political system and individual sophistication. One way we would imagine that a strong party system such as that in Norway would differ from a weak partycandidate system such as that in the US would be in how easy it is to obtain information about the parties. Less sophisticated voters would be more dependent on the political system to provide cues as to where the parties or candidates stand on the issues. More sophisticated citizens would tend to find this out for themselves and would be less dependent on political institutions. Hence, we anticipate that sophistication will make a bigger difference in the United States than in Norway.

Table 5 indicates the extremity and variance of perceived party position by level of political sophistication. The measure of sophistication we use is fairly primitive and is based on the education and expressed political interest of respondents.³ In general, the low sophistication group includes those with low education and low political interest, the high group includes

those with high education and high interest, and the middle group covers
the rest of the sample.

The Powell hypotheses are borne out in both countries. Indeed, the pattern is monotonic for every one of the political objects. Each party and candidate is perceived more sharply as we move up the sophistication scale; it is also perceived to be more extreme.

The commonality across the systems in terms of the shift toward extremeness masks a fairly dramatic difference. In Norway, for each of the more established parties there is a fairly small change in perceived position, whereas in the United States there is a more marked shift. To see this, we can examine the parties that are substantively the most similar: Labour and Democrats; Conservatives and Republicans. The perceived extremity of the Labour Party changes from 0.85 to 1.07 across sophistication levels, a very small shift whether we consider the actual difference of 0.22 or the proportional change of about 25 percent. The extremity of the Democratic Party changes from 0.10 to 0.98, an actual difference of 0.88 and a proportional change of almost 900 percent. Similarly, for the Conservative Party the values range from 0.96 for the low group to 1.38 for the high group; for the Republican Party, from 0.35 to 1.10. In short, we observe pronounced changes in perception by level of sophistication in the American case - a result consistent with the view that it takes some real initiative and attention to receive political information in the United States.

The pattern allows for another explanation as well. In Norway, two parties show shifts in perceived extremity that are comparable to those observed for the parties in the United States. These parties - the Socialists and the Progressives - are, respectively, the most extreme party of the left and of the right in Norway. Perhaps parties that define one bound or the other of the ideological spectrum tend to elicit the greatest differences in perception across sophistication groups. This explanation would also encompass the US results and, if correct, would mean that the impact of sophistication is not so different in the two systems.

The need for caution is reinforced when we consider the variance in issue perceptions. We have already seen in both countries that citizens' views of party positions become more crystallized with increasing sophistication. Now we wish to gauge the extent of improvement. To do this, we compute the "low/high ratio" presented in Table 5. This ratio is the variance in perceived party position among the low sophisticates divided by the variance in perceived party position among the high sophisticates. The larger the ratio, the greater the improvement in clarity of perception. If sophistication has a stronger impact in the United States, we should see larger ratios in the US than in Norway. However, this is not the case. The ratios range from 1.53 to 2.58 in Norway and from 1.66 to 1.88 in the United States. Thus, while the variances themselves are always lower in Norway,

there is no evidence of an interaction effect. Clarity of perception seems
to respond to sophistication in much the same way in the two societies.

Conclusion

Our results provide support for the Campbell-Valen and Granberg view that the political system matters. Consistent with expectations we have found that perceptions of party positions on issues are clearer in Norway than in the United States. The differences are especially pronounced for ideology. We have also observed that sophistication matters. People who are more politically sophisticated locate parties in less central positions and place them more similarly than less sophisticated voters. While we find some support for the view that sophistication matters more in the United States, the evidence is not compelling.

Finally, a brief caveat is in order. From an analytic standpoint the results we have obtained could be a function of any factor that distinguishes the US and Norway. Even a short list of plausible factors would include: twoparty versus multiparty system, irresponsible versus responsible parties, federal versus unitary system, and - at a broader societal level - differences in population size and heterogeneity. To attempt to identify the causal agents would require further study based on a more systematically diverse set of countries. The results of the present study, combined with those of Granberg and Holmberg, simply give us confidence that real differences in the clarity of issue perception exist between the US and Scandinavian systems.

---

NOTES

1. Relative variance is the variance in perceived party (candidate) position divided by the variance in mass opinion on the issue.

2. The formula for calculating extremity is 2 (Nj mean²,/2 Nj), where Nj is the number of cases that responded to the issue. We use the squared values to be sensitive to party strategies which adopt fairly extreme positions on a few select issues, and to be more consistent with Euclidean distance which is the square root of the sum of the squared values.

3. In Norway the high sophistication group consists of those who had completed gymnasium and said they were either somewhat or very interested in politics. The low group

ACKNOWLEDGEMENTS

We are indebted to Bernt Aardal and Henry Valen for their willingness to include an extensive set of issue questions in the 1989 Norwegian Election Study. Data were made available through the Inter-university Consortium for Political and Social Research and the Norwegian Social Science Data Services. This research was supported by grants from the National Science Foundation (SES-9210825) and the Norwegian Research Council.

consists of those who had not completed gymnasium and who said they were a little or not at all interested in politics. In the United States the high group consists of those who had completed college and said they were somewhat interested or very interested in the political campaign and those with education beyond high school who said they were very interested in the campaign. The low group consists of all those with less than high school education and those with high school education who said they were not particularly interested in the political campaign.

REFERENCES

Campbell, A. et al. 1960. The American Voter. New York: Wiley.

Campbell, A. & Valen, H. 1966. "Party Identification in Norway and the United States", in
Campbell, A. et al.. Elections and the Political Order. New York: Wiley.

Conover, P. J. & Feldman, S. 1981. "The Origins and Meaning of Liberal-Conservative Self
Identification", American Journal of Political Science 25, 617-645.

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

Granberg, D. & Holmberg, S. 1988. The Political System Matters. Cambridge: Cambridge
University Press.

Listhaug, O. 1989. Citizens, Parties and Norwegian Electoral Politics 1957-1985: An Empirical
Study. Trondheim: Tapir.

Listhaug, O. N.d. "The Decline of Class Voting", in Strom, K. & Svåsand, L. G., eds..
Challenges to Political Parties: The Norwegian Case. Forthcoming.

Luskin, R. C. 1987. "Measuring Political Sophistication", American Journal of Political
Science 31, 856-899.

Powell, L. W. 1989. "Analyzing Misinformation: Perceptions of Congressional Candidates"
Ideologies", American Journal of Political Science 33, 272-293.