Resumé

Data on self-placement and political perception on an ideology scale from surveys of adults in Sweden and the U.S. were analyzed and compared. Consensus in political perception, the correlation between perception of the party's and its leader's position, distinctiveness and polarization, subjective agreement, and a rational democratic coefficient were all substantially higher in Sweden. In spite of these differences, evidence of assimilation and contrast effects in perception was observed within both countries.

Introduction

As political systems, Sweden and the U.S. are very different. Sweden has a disciplined multiparty system with a unicameral parliament with seating based on proportional representation. The U.S., by contrast, has a weak, relatively undisciplined two-party system with a bicameral legislature and an independent presidential branch, and elections based on a principle of winner-take-all. The vagueness or ambiguity of politics in the U.S. is implied in the expression, 'fishing in muddy waters', often used to describe politics and political activities in the U.S. By contrast, politics and politicians in Sweden have the reputation of being more clear and open. Most political issues in Sweden cohere with a well-known left-right dimension (Holmberg 1974, 1984). This difference may not be unrelated to the much higher participation in Sweden than in the U.S., with the voting rates being roughly 90% and 55%, respectively.

Overall, it may be that among the Western democracies, Sweden and the U.S. are about as different as any two systems. If so, comparative analyses between them may hold considerable potential. For example, if processes are observed to occur similarly in these two very different systems, it is not unreasonable to suppose that resultant propositions may have considerable

generality. On the other hånd, if an effect were observed to occur in one
country but not the other, this would tell us that it has more limited
generality.
Most research on political perception, beginning with the 1948 election
(Berelson, Lazarsfeld, & McPhee 1954), has been done in the U.S. Insofar
as this research has been guided by theoretical considerations, the conceptual
tools provided by Sherif & Hovland's (1961) assimilation-contrast model and
by Heider's (1958) balance theory have been most often employed.

Citizens, it is reasoned, will be motivated to achieve and maintain a subjective state of agreement with candidates or parties they like and disagreement with candidates or parties they dislike (Page & Brody 1972). They can do this by rationally choosing on the basis of proximity or similarity, by altering their own position to coincide more closely with that of a preferred candidate or party, or by subjectively distorting the actual position of the candidate or party in the process of political perception. If the latter course is followed, it is presumed that people would show assimilation (overestimating the similarity between self and other) when perceiving the position of a liked candidate, and contrast (overestimating the difference between self and other) when perceiving the position of a disliked candidate. All this can be incorporated quite handily within Heider's P-O-X cognitive consistency or balance model.

The evidence to date is quite supportive of this approach. Considerable evidence from numerous elections has been examined (Granberg & Brent 1974; Granberg & Jenks 1977; King 1978; Luttbeg 1981; Sherrod 1972), resulting in an empirically grounded set of propositions pertaining to the circumstances under which assimilation and contrast effects occur in political perception (Brent & Granberg 1982; Granberg 1982; Granberg & Seidel 1976).

Until recently there was virtually no hint as to whether such processes occur in a similar manner in countries other than the U.S. Now there are some indications that assimilation and contrast effects in political perception do indeed generalize to a country as different from the U.S. as Sweden (Granberg 1983; Holmberg 1981). However, the published reports make precise comparisons difficult. In some cases, the issues considered may be quite different in salience (Granberg & Holmberg 1986a). In other cases, an 11-point scale is used in placement judgments of the parties in Sweden, while a 7-point scale is used for judgments of the candidates in the U.S. (Granberg 1985a; Granberg & Holmberg 1986b).

The data used in the present paper are considerably more comparable. People in Sweden were asked to place themselves and to give their perceptions of the positions of the five parties and the leaders of those parties on a 1-7 left-right scale. People in the U.S. were asked to place themselves

and to give their perceptions of the positions of the two parties and the
nominees of those parties on a 1-7 liberal-conservative scale.

The central proposition in social judgment theory being applied here is that one's own position and one's attitude toward a party exert an effect on perception of the position of that party. That is, affect influences cognition. In addition, we also shall consider an alternative, more cognitively based theory that has recently been put forth in a series of articles (Conover 1981; Conover & Feldman 1982, 1986; Feldman & Conover 1983; Granberg 1985b). Political cue theory depicts political perception occurring as a result of information processing which may be done in a manner that is (more or less) affectively neutral. People develop schemas or general concepts of what goes together and how things work. Cues can then be used to infer features that go beyond the information that is given. In this approach, considerable emphasis is placed upon the political party that can be used as an abiding anchor from which to infer positions of particular candidates. This approach has emphasized and capitalized on the strong correlation that does exist in the U.S. between where people place a party on a scale and where people place the nominee of that party.

It has been shown that experimentally manipulating cues or the salience of certain cues can have an effect on political perception (Conover 1981; Granberg 1985b). Also, in analysis of survey data, the person's perception of the party's position and the person's own position are used to jointly predict perception of a preferred candidate's position. When this is done, the party cue variable appears to have a stronger effect, and the erstwhile assimilation effect, while still statistically significant in most instances, is reduced very substantially.

Method

The data from in-person interviews in two surveys are now analyzed. The Swedish data are from a survey conducted by the Svenska Institute! for Opinionsundersokningar (SIFO) in June, 1982 prior to the parliamentary election of that year. The Swedish sample (N = 1021) was drawn to be representative of adults in Sweden aged 18 to 70. Respondents were shown a card with a left-right (vanster-hoger) scale going from 1-7. They were then told, 'In politics, people talk about left and right. Here is a scale from left to right. How would you place the party leaders on the scale?' After giving their perceptions of the leaders of each of the five major parties (Ulf Adelsohn, Ola Ulisten, Thorbjorn Fålldin, Olof Palme, and Lars Werner), people were then asked, 'How would you place the parties on the scale?' After they had given their perceptions of each of the five parties (Conservative, Liberal, Center, Social Democratic, and Communist), they were asked, 'How would you place yourself on the scale?' (Zetterberg 1982). Party preference was

measured by asking people which party they liked best. Party preference and voting are remarkably stable in Sweden, and such a simple measure is a highly accurate predictor of how people will later vote in the election (Granberg & Holmberg 1986b; Holmberg 1981).

The U.S. data are drawn from the 1984 National Election Study (N = 2257) conducted by the Center for Political Studies of the University of Michigan. The sample is representative of adults in the U.S., aged 18 and older. In the pre-election interviews done in September and October of 1984, people were shown a card with a 1-7 scale with the ends labeled extremely liberal and extremely conservative. The instructions were, 'We hear a lot of talk these days about liberals and conservatives. Here is a 7-point scale on which the political views people might hold are arranged from extremely liberal to extremely conservative. Where would you place yourself on this scale, or haven't you thought much about this?' After answering this, they were asked a series of four questions regarding their perceptions of the positions of the candidates and the parties, 'Where would you place Ronald Reagan (Walter Mondale, the Democratic Party, the Republican Party) on this scale?' Party identification was measured by asking people, 'Generally speaking, do you usually think of yourself as a Republican, a Democrat, an Independent, or what?' Candidate preference was measured by asking people for whom they intended to vote in the 1984 election. The relationship between party identification (using a 7-point scale measuring strength of identification if Democrat or Republican and whether they think of themselves as closer to the Democratic or Republican Party, if Independent) and voting behavior was very strong in 1984, reaching an all-time high of .73 (Eta), among comparable surveys going back to 1952. The relationship between voting intention and voting behavior was unusually strong in 1984 (r = .91), with about 96% of the respondents later reporting voting in a way that was consistent with previously stated intentions.

Results

Self-Placement

The distributions for self-placement and perception of the positions of the leaders, candidates, and parties on the ideology scale are given in Table 1. As shown in Fig. 1, self-placement does not occur identically in Sweden and the U.S. The average self-placement for Sweden was slightly to the left of center (3.8), while in the U.S. it was slightly to the right of center (4.2). In both countries, the modal position was in the center (4), but the percentage choosing this centrist position was larger in the U.S. than in Sweden (34% to 27%). In terms of content, issues, and connotations, it obviously means something quite different to be in the center of the political spectrum in Sweden from what that would mean in the U.S.

It may also be noted that there is a rather large difference in the percentage of the sample who place themselves at some point on the 1-7 scale. Most of this difference is probably attributable to the 'easy out' provision in the U.S. question ('...0r haven't you thought much about this?'). The General Social Surveys of adults in the U.S., conducted by the National Opinion Research Center of the University of Chicago, does not provide this 'easy out' and get rates of missing data on the 1-7 ideological self-placement scale that are very similar to that for Sweden in Table 1. Overall, however, one is struck by the similarity in the distribution of self-placements on the ideology scale in the two countries, as shown in Fig. 1. In both cases, the resemblance to a bell-shaped curve is rather strong.

Perceptual Consensus

If the political system of Sweden is less ambiguous in ideological terms than in the U.S., this ought to be evident in the distribution of placement judgments of the parties, the party leaders, and nominees. Table 1 presents supportive evidence in that regard. For each stimulus, a perceptual consensus coefficient was calculated by (a) dividing the obtained standard deviation by the standard deviation that would result if the placement judgments had

been made randomly by that number of people using that number of alternatives and then (b) subtracting the result from one (Granberg & Holmberg 1986a, 1986b). In Sweden, the perceptual consensus coefficients vary from a low of .49 for Thorbjorn Fålldin, Prime Minister and leader of the Center Party, to a high of .59 for the Communist Party. As might be expected, perceptual consensus is somewhat higher for the extreme right and extreme left parties than for the parties of the center. In the U.S., the perceptual consensus coefficients were comparatively low: .16 for Reagan, .22 for the Democratic Party, .23 for the Republican Party, and .24 for Walter Mondale. These coefficients probably represent somewhat more consensus than would occur by chance but not much more. An interesting sidelight here is to note that in the U.S. there was actually more collective consensus in self-placement than in perception of the parties and candidates. In Sweden, by comparison, there was less consensus in self-placement than in perceptions of the parties and party leaders.

Another way of inferring the relative ambiguity of the two systems is by examining the frequency of 'wrong side errors'. A wrong side error in the U.S. would occur when someone places Mondale or the Democratic Party to the right of center or places Reagan or the Republican Party to the left of center. In Sweden, a wrong side error in perception occurs when one of the Socialist parties (Communist or Social Democratic) or its leader is placed to the right of center or when one of the bourgeois parties (Center, Liberal, or Conservative) or its leader is placed to the left of center. Wrong side errors occurred in about 22% of the perceptual estimates in the U.S., far more than the approximately 3% in Sweden.

Overall, it is safe to say that perceptual consensus, that is, collective agreement as to where on the ideology scale the parties and party leaders should be placed, is much higher in Sweden than in the U.S. This is not unexpected, but it does confirm that differences appear at a social psychological level of analysis in a way that relates to differences between the political systems. It should also be added that the higher perceptual consensus in Sweden is not limited to the extreme left and right parties. The centrist parties in Sweden were also placed with more perceptual consensus and fewer wrong side errors than the parties and nominees in the U.S.

Perceptions of the Positions of Parties and Their Leaders and Nominees

Given that Sweden has a strong party system while the U.S. does not, this also ought to have a counterpart empirical finding at the social psychological level of analysis involving individual perceptions. Table 2 summarizes the relationship between perception of the position of the party and the party's leader for Sweden and between perception of the party and the party's

presidential nominee for the U.S. There it is evident that people in Sweden are far more likely to put the two at exactly the same place on the 1-7 ideology scale than is true for the U.S. Overall, in about 85% of the cases in Sweden, people put the party and the party leader at the same position on the scale, while this was true in only about 49% of the cases in the U.S. This means that people in the U.S. are more likely to perceive a difference between the party's position and that of the party's nominee than people in Sweden are to perceive a difference between the party's position and that of the party leader. An associated question concerns whether a perceived difference is systematic as to direction. Some slight differences in averages were detected by a series of related sample t-tests. Reagan was perceived overall to be somewhat to the right of the Republican Party (4.99 to 4.90, t = 3.20, p<.o\), but the difference was really quite slight. The Democratic Party was seen as slightly farther from the center than Mondale (3.35 to 3.45, / = 2.86, p<.ol). This difference is so small as to be substantively unimportant. In the four U.S. presidential elections in which comparable data are available (1972-1984), the largest gap occurred in 1972 when George McGovern was

seen as being rather considerably to the left of the Democratic Party's
position.

In the case of Sweden, three of the overall tests yielded nonsignificant results. Specifically, in comparing the perceptions of the Communist, Center, and Liberal parties and their respective leaders, the averages did not differ significantly. Olof Palme was perceived as slightly to the left of the Social Democratic Party (2.34 to 2.41, t = 4.74, p<.00l), and the Conservative Party was perceived as being slightly closer to the extreme right position than was Ulf Adelsohn (6.57 to 6.50, t = 3.77, ;?<.001). Again, these differences are very slight and seem to depend on the specific circumstances. That is, we cannot detect any consistent tendency to perceive the individual candidate or leader as closer to the center than the party or vice versa.

Perceived Distinctiveness and Polarization

V.O. Key's (1966) concept of an echo chamber implied that there ought to be an empirical relationship between how distinct the alternatives are from among which the voters are choosing and the degree to which voter groups are polarized or divided on that same dimension (Granberg, Harris, & King 1981; King 1978). The presumption is that voter groups will echo, perhaps in a slightly muted or modulated form, the alternatives which they are presented in elections. Thus, if the voter groups are not divided or polarized on an issue or on ideology, this may often be traceable back to a lack of distinctiveness of the alternatives.

It is obvious from Table 1 that not only the number but also the range of alternatives is very considerable in the case of Sweden. It would also appear from the averages in Table 1 that the range of the alternatives for people in the U.S. is by comparison much more constricted.

Just how distinctive the alternatives were perceived to be in the U.S. can be observed more precisely by superimposing the distribution of perceptions of Reagan's and Mondale's positions on the ideology dimension. Then, as shown in Fig. 2, it becomes a matter of determining the area of nonoverlap. Obviously, maximum perceived distinctiveness would be implied if the two distributions did not overlap at all. If they overlapped completely (i.e., coincided), this would imply minimum or zero distinctiveness. From this procedure, a perceived candidate distinctiveness coefficient of .46 is inferred for the U.S. in 1984 on the ideology dimension. If we use the same procedure for the Social Democratic and Conservative parties (which have been the two largest parties in the past three elections and which together received 72% of the five-party vote in 1982), the perceived party distinctiveness coeficient for Sweden in 1982 was .95.

Is this large difference in perceived distinctiveness reflected in the degree
of voter group polarization on the same ideology dimension? To determine

this, the distribution of self-placement on the ideology dimension by Mondale supporters is superimposed on the comparable distribution of Reagan supporters on the same dimension (Fig. 2). Again, the degree of nonoverlap can be used, this time to infer the degree of voter group polarization. When this was done, the resultant polarization coefficient was much higher for Sweden than for the U.S. (.78 to .37). Thus, there is strong support here for the implication derived from Key's concept of an echo chamber. Provide

voters a choice between a distinct set of alternatives, and they will divide
themselves accordingly (Page 1978).

Effect of Candidate or Party Preference on Perceptions

The next matter concerns the effect of candidate or party preference, per se, on political perceptions. Prior research suggests the main effect of candidate or party preference on political perceptions will not be large (Carlson & Habel 1969; Granberg & Brent 1974; Granberg & Holmberg 1986b; Sigel 1964). The pertinent results are in Table 3. There it is seen that candidate or party preference does exert a significant main effect in some instances and not in others.

Reagan supporters perceived Mondale and the Democratic party as
departing farther from the center than did Mondale supporters. However, this
effect was not symmetrical. That is, Mondale supporters were no more likely

than Reagan supporters to perceive Reagan and the Republican Party
departing farther from the center.

as

In the case of Sweden, party preference exerted a significant main effect on perceptions of each party and its leader except in the case of the Conservative Party and its leader Adelsohn. One might suppose that people would be inclined to displace one's own party toward the center out of an intuitive sense that that is where most of the voters are, and therefore, that is where one's party ought to be. The Communists tend to displace all parties toward the right of themselves, including their tacit ally, the Social Democratic Party. The Communists also tend to see their own party as not so extremely toward the left as it is regarded by the other party preference groups. Overall, it is apparent from Table 3 that candidate and party preference by itself can exert some effect on political perception. This effect, however, does not appear to be especially strong and it is difficult to discern any regular or predictable pattern.

Subjective Agreement with One's Preferred Candidate or Party

Subjective Agreement Preferred Party If the alternatives available to the citizen in Sweden cover a broader range than the alternatives provided by the major parties in the U.S., it follows that, at the social psychological level of analysis, we ought to be able to observe a greater degree of subjective agreement in Sweden than in the U.S. By subjective agreement in this context, we mean the degree of correspondence between a person's own ideological position and that which the person attributes to a preferred party or candidate (Granberg & Holmberg 1986b.).

Subjective agreement is measured by using one's self-placement as the independent variable and placement of one's preferred party, candidate, or leader as the dependent variable. The Eta coefficient from such analyses can be regarded as a subjective agreement coefficient. The results of such analyses for Sweden and the U.S. are shown in Fig. 3. Regardless of whether placement of preferred party or preferred leader or nominee is the dependent variable, it is obvious that subjective agreement is much stronger in Sweden than in the U.S.

Rational Democratic Processes and Perceptual Distortion

It will be readily recognized that the subjective agreement coefficient is a rather crude indicator in that the same degree of subjective agreement could result from a variety of underlying processes acting singly or in conjunction with one another. For instance, if the alternatives presented to citizens corresponded to the positions held by citizens, rational selection on the grounds of objective proximity could by itself, at least in theory, produce complete subjective agreement. Second, people might also alter their selfplacement so as to coincide more closely with the position of their preferred

party or candidate. Third, people may engage in psychological distortion by altering or bending their perception of the preferred party's or candidate's position to make it seem more similar to their own position than it actually is. Any or all of these three processes might be producing a given level of subjective agreement.

There is probably no adequate method available for completely disentangling the effects of these interrelated processes using cross-sectional data. However, it may be possible to identify the effect of the first two, relatively more rational processes, and compare this with the presumed collective effect of all three processes that is evident in the subjective agreement coefficient.

Assume that the subjective agreement coefficient (shared variance between self-placement and placement of preferred party or candidate), as described above, is an adequate measure of the joint effect of all three processes. Then, if people rationally select a candidate or party on the criterion of objective proximity, or if they alter their self-placement to coincide with the position of a preferred party or candidate, this ought be reflected in shared variance between party or candidate preference and self-placement on the ideology scale. For lack of a better term, let us call this the rational democratic coefficient. It is measured by the Eta resulting from using party or candidate preference as the independent variable and self-placement on the ideology scale as the dependent variable.

Comparing the subjective agreement coefficient and the rational democratic coefficient may indicate the relative contribution of the first two processes and the third process to the observed level of subjective agreement. For instance, if subjective agreement is high and the rational democratic coefficient is as high or nearly so, this would imply that the first two processes

are essentially producing the subjective agreement. On the other hand, if the level of subjective agreement is high but the rational democratic coefficient is very low, this would imply that the level of subjective agreement is probably being produced by irrational psychological distortion.

The rational democratic coefficients have been given in Table 3 and the subjective agreement coefficients in Table 2. These coefficients for Sweden and the U.S. are compared to each other in Fig. 4. There it is evident that both coefficients are much higher for Sweden than for the U.S., but the difference between the two is quite similar for both countries. This implies that psychological distortion in placing a preferred candidate or party (i.e., assimilation) may be occurring to roughly the same extent in both countries.

Assimilation and Contrast Within Voting Preference Groups The subjective agreement coefficients in Fig. 3 are based on a composite of all respondents placing their preferred party, candidate, or leader, whichever one that happens to be for a particular person. In that way, the subjective agreement that is observed could be due to psychological distortion, but it could also be due to rational selection or a persuasion effect. However, when

we look within a party or candidate preference group, the party or candidate whose position is being estimated is, in effect, being held constant. Therefore, in such analyses, relationships between self-placement and perception of a party's or candidate's position within a preference group must be due to psychological distortion (i.e., assimilation or contrast).

The correlations between self-placement and perception of the parties' and party leaders' positions within party preference groups in Sweden and between self-placement and perception of the candidates' and parties' positions within candidate preference groups in the U.S. are shown in Table 4. If one looks at the rows in which people are not selected on the basis of candidate or party preference (rows 1 and 7), the correlations are mostly negative but hover near .00. However, when control is instituted so that the people, whose perceptions and self-placements are being correlated, are homogeneous as to party or candidate preference, systematic and strong effects appear. Within the U.S., Reagan's supporters show a rather strong tendency to assimilate in their perceptions of the positions of Reagan and the Republican Party, and to contrast in their perceptions of Mondale and the Democratic Party. The mean values for perceptions of the candidates' positions as a function of candidate preference and self-placement are shown in Fig. 5.

In Sweden, the Communists showed a strong tendency to assimilate in perceptions of the position of their preferred party and its leader. They also showed a mild assimilative tendency in regard to the party that was ideologically adjacent to them (Social Democrats). They showed a similar tendency in relation to the Center Party, which had shared a strong antinuclear power position with the Communists, an issue which tends to be independent of left-right ideology in Sweden (Holmberg 1981; Holmberg & Asp 1984).

The Social Democrats showed a substantial tendency to assimilate in their perceptions of their preferred party and its leader, a mild tendency to assimilate when estimating the position of the Communist Party, and a mild tendency toward contrast when estimating the position of the three bourgeois parties. People in the right-of-center parties each showed a significant tendency to assimilate in perceptions of their respective preferred party and its leader, but their only contrast occurred when supporters of the extreme right Conservative party were perceiving the position of the moderately leftist Social Democratic Party. The average perceptions for supporters of the Social Democratic and Conservative Parties are plotted in Fig. 5. Although the analyses may not be altogether unequivocal, the evidence in Table 4 and Fig. 5 strongly suggests that the dual processes of assimilation and contrast, identified by Sherif & Hovland (1961), indeed do occur in political perceptions in both Sweden and the U.S. on the ideology dimension.

Inferring Leaders' and Nominees' Positions

Finally, we will consider the results of comparable regression analyses in which perception of the party's position and the person's self-placement are used to predict perception of the party leader's or nominee's position. We have already seen in Table 2 that the correlation between perception of party and perception of party leader, as a bivariate relationship, is much stronger in

Sweden than the correlation between perception of party and perception of nominee in the U.S. Indeed, we could simulate a situation in which that relationship is perfect by selecting only those people who place the party and its leader at the same position on the scale. Sweden under natural circumstances approaches that situation, though not perfectly, as about 15% of Swedish respondents saw some difference in their perception of the party's and the party leader's position on the ideology scale.

Table 5 shows the results of the regression analyses for the data from Sweden and the U.S. If one were to take these results at face value, they would indicate that the party cue variable operates strongly in both systems, but somewhat more strongly in Sweden. The data for the U.S., consistent with prior reports, imply that the erstwhile assimilation effects, though still statistically significant, are not very substantial. The erstwhile assimilation effects in Sweden, with the exception of the data from the Communists, appear to be reduced to nonsignificance in the regression analyses.

There can be no question that perception of the party's position and perception of the party leader's or nominee's position are strongly correlated both in Sweden and in the U.S. It is not so easy to interpret the meaning of that relationship. The evidence in Table 5 does not really prove that people infer a leader's or nominee's position from their prior knowledge of the party's position, although it is plausible to suppose that it probably does work that way for some people.

Given the abundance of evidence presented previously in this paper, it is still reasonable to maintain that systematic displacement effects (i.e., assimilation and contrast) do occur in political perception in both the U.S and Sweden. When people have a positive link with a political stimulus, whether it is a party, party leader, or nominee, it appears likely that assimilation in political perception will occur.

Concluding Remarks

Like other efforts at comparative analyses, especially those utilizing surveys not expressly designed with such analyses in mind, the data used here provide comparative analyses which are less than exact. Two differences ought to be kept in mind. First, the U.S. data labeled the end points 'extremely', while the Swedish data did not. It is difficult to say what effect this variation might have had. Second, the U.S. survey had more missing data than the Swedish survey, as shown in Table 1. As indicated before, this is most likely due to the 'easy out' feature in the U.S. question on selfplacement.

Had people in the U.S. not been given this, it is quite likely that a higher
percentage of people would have answered the self-placement and perception

questions within the 1-7 range and that this would add some additional 'noise' or random variation into the U.S. data. What this would seem to imply is that some of the observed differences between Sweden and the U.S., as in the much larger rational democratic coefficient in Sweden, would actually be even larger if the questions had been more similar. That is, if people in both countries or in neither country had been given the 'easy out' provision in the self-placement question, then several of the differences between Sweden and the U.S. would be even larger than what appear in the tables and figures of this article.

The assimilation-contrast model of Sherif & Hovland (1961) makes the rather difficult prediction that people placing or estimating the position of a series of stimuli will displace some toward their own position (assimilation) while displacing others away from their own position (contrast). In their original work, assimilation was expected when people were estimating the actual position of stimuli that were only mildly discrepant from their own position and at least somewhat ambiguous. Contrast was expected when people were estimating the position of a stimulus which was rather substantially distant from their own position and yet somewhat ambiguous. Their research dealt with judgments of anonymous statements or speeches, a rather different task from that faced by the people in the present study.

Nevertheless, the people in the Swedish survey were asked to estimate the position of a series of 10 stimuli. They did show assimilation of some and, perhaps to a lesser extent, contrast of others. The crucial factor here, however, is that the people had an attitude toward each of the parties and leaders whose positions they were being asked to estimate. This factor (P's attitude toward 0 in Heider's balance model) seems to be decisive in determining whether assimilation or contrast will occur in placement of a particular party or leader.

---

NOTE

1. This project was initiated while the author was a guest scholar at the Sociologiska Institutionen of G6teborgs Universitet with support from the Svenska Institutet, and while on sabbatical leave from the University of Missouri-Columbia. The author is currently Guest Professor in the Department of Political Science, University of Goteborg. Data from the 1982 Swedish survey were obtained from the Svenska Institutet for Opinionsundersokningar with a grant from the Graduate Research Council of the University of Missouri-Columbia. Data from the 1984 National Election Study of the U.S.A., collected by the Center for Political Studies, were obtained through the Interuniversity Consortium for Political and Social Research at the University of Michigan. The author wishes to thank Gregory Johnson, William Jacoby, Birol Yeshilada, Karin Busch, Bent Rembe, Hans Zetterberg, Sten Olsson, Bertrand Granberg, and Patricia Shanks for their advice and assistance. The author is solely responsible for the analyses and interpretation.

REFERENCES

Berelson, 8., Lazarsfeld, P. & McPhee W. 1954. Voting: A Study of Opinion Formation in a Presidential
Campaign. Chicago: University of Chicago Press.

Brent, E. & Granberg, D. 1982. 'Subjective Agreement with the Presidential Candidates of 1976
and 1980', Journal of Personality and Social Psychology 42, 393-403.

Carlson, E. & Habel, D. 1969. 'The Perception of Policy Positions of Presidential Candidates',
Journal of Social Psychology 79, 69-77.

Conover, P. 1981. 'Political Cues and Perception of Candidates', American Politics Quarterly 9,
427-448.

Conover, P. & Feldman, S. 1982. 'Projection and the Perception af Candidates' Issue Positions',
Western Political Quarterly 35, 228-244.

Conover, P. & Feldman, S. 1986. 'The Role of Inference in the Perception of Political Candidates',
pp. 127-158 in R. Lau & D. Sears, eds., Political Cognition. Hillsdale, New Jersey: Lawrence
Erlbaum.

Feldman. S. & Conover, P. 1983. 'Candidates, Issues and Voters: The Role of Inference in Politi
cal Perception', Journal of Politics 45, 812-839.

Granberg, D. 1982. 'Social Judgment Theory', Communication Yearbook 6, 304-329.

Granberg, D. 1983. 'Preference, Expectations, and Placement Judgments: Some Evidence from
Sweden', Social Psychology Quarterly 46, 363-368.

Granberg, D. 1985a. 'An Assimilation Effect in Perceptions of Sweden's Political Parties on the
Left-Right Dimension', Scandinavian Journal of Psychology 26, 88-91.

Granberg, D. 1985b. 'An Anomaly in Political Perception', Public Opinion Quarterly 49, 504-516.

Granberg, D. & Brent, E. 1974. 'Dove-Hawk Placements in the 1968 Election: Application of So
cial Judgment and Balance Theories', Journal of Personality and Social Psychology 29,
687-695.

Granberg, D., Harris, W. & King, M. 1981. 'Assimilation but little Contrast in the 1976 U.S.
Presidential Election', Journal of Psychology 108, 241-247.

Granberg, D. & Holmberg, S. 1986a. 'Political Perception in Sweden and the U.S.: Analyses of
Issues with Explicit Alternatives', Western Political Quarterly 39, 7-28.

Granberg, D., & Holmberg, S. 1986b. 'Subjective Ideology in Sweden and the U.S.', Research in
Political Sociology 2, 107-143.

Granberg, D. & Jenks, R. 1977. 'Assimilation and Contrast in the 1972 Election', Human Rela
tions 30, 623-640.

Granberg, D. & Seidel, J. 1976. 'Social Judgments of the Urban and Vietnam Issues in 1968 and
1972', Social Forces 55, 1-15.

Heider, F. 1958. The Psychology of Interpersonal Relations. New York: Wiley.

Holmberg, S. 1974. Riksdagen representerar Svenska folket: Empiriska studier i representativ
demokrati. Lund: Studentlitteratur.

Holmberg, S. 1981. Svenska våljare. Stockholm: Liber Forlag.

Holmberg, S. & Asp, K. 1984. Kampen om kårnkraften. En bok on våljare, massmedier ochfolkomrostningen
1980. Stockholm: Liber Forlag.

Key, V. 1966. The Responsible Electorate. Cambridge: Harvard University Press.

King, M. 1978. 'Assimilation and Contrast of Presidential Candidates' Issue Positions, 1972',
Public Opinion Quarterly 41, 515-522.

Luttbeg, N. 1981. 'Balance Theory as a Source of where Political Parties Stand on the Issues',
pp. 137-145 in N. Luttbeg, ed., Public Opinion and Public Policy. Itaska, Illinois: F.E. Peacock.

Page, B. 1978. Choices and Echoes in Presidential Elections: Rational Man and Electoral De
mocracy. Chicago: University of Chicago Press.

Page, B. & Brody, R. 1972. 'Policy Voting and the Electoral Process: The Vietnam Issue', American
Political Science Review 66, 979-995.

Sherif, M. & Hovland, C. 1961. SocialJudgment: Assimilation and Contrast Effects in Communication
and Attitude Change. New Haven: Yale University Press.

Sherrod, D. 1972. 'Selective Perception of Political Candidates', Public Opinion Quarterly 28,
483-496.

Sigel, R. 1964 'Effect of Partisanship on the Perception of Political Candidates', Public Opinion
Quarterly 35, 554-562.

Zetterberg, H. 1982. 'Hoger-vånster', Stfo Indikator: Stfos Politiska Nyhetsbrev 9, 9-10.