CREST PAPER NO 61 : POLITICAL TALK AND THE STABILITY AND CONSISTENCY OF POLITICAL ORIENTATION
BY
GEOFFREY EVANS AND MANSUR LALLJEE
ABSTRACT
This paper investigates the relationship between talk and political orientation. Two indices of talk are considered: frequency (the extent to which a person talks politics) and discussant homogeneity (whether the person(s) with whom they talk politics support the same party as they do). Two aspects of political orientation are examined: party identification (viz. the party to which the person feels closest) and political attitudes (along the left-right and libertarian-authoritarian dimensions). It is hypothesised that the frequency of political talk will be related to the stability and consistency of a person's political attitudes and to the stability of their party identification. It is also expected that discussant homogeneity will be related to the stability of party identification and to the stability of political attitudes. The data were drawn from a panel study which used a stratified national sample of over 200 respondents interviewed in 1985 and 1986. The results show that frequency of talk is clearly related to the consistency of political attitudes; and also related to the stability of political orientation. Discussant homogeneity is clearly related to the stability of party identification but not to attitudinal stability. The results offer general support for the ideas presented which emphasised the importance of talk as a way of expressing a person's political identity and the interpersonal processes involved in its maintenance.
This paper is primarily concerned with the interpersonal processes underlying the consistency and stability of political orientation. Political orientation consists of two main features: party identification and political attitudes. Party identification refers to the party that a person feels closest to, an affective link between a person and a political party (Campbell, Converse, Miller & Stokes, 1960). Political attitudes are normally conceptualised along two separate dimensions: the left - right dimension (which is broadly concerned with social inequality and stresses the importance of actions that redress such inequality); and the libertarian - authoritarian dimension (which is broadly concerned with individual freedom and conversely with punishment for people who break the law) (Evans, Heath & Lalljee, 1996).
The main hypotheses of the current investigation are that the stability and the consistency of political orientation are related to the frequency of political talk. These hypotheses stem partly from the view that talk is one of the ways in which people express their political identity; and further that talk sets up expectations in the interlocutor which the speaker strives to fulfil. These interpersonal functions of attitude expressions have been mentioned by a range of writers (e.g. Baumeister,1982; Lalljee, Brown & Ginsburg, 1984; Petty & Cacioppo,1982; Schlenker,1982), but have not played a central role in the psychological study of attitudes. Here we suggest that making a statement in some sense commits the speaker to a point of view, and that the speaker is expected to make similar statements on future occasions. If a person claims to support a particular party, we would expect them to make similar claims in the future. The more frequently they claim to identify with that party, the more these expectations will be cemented and the greater the obligations on the speaker to live up to them. The stability of a person's party identification and of their attitudes towards specific political issues will be affected through talk in the same way. If a person makes a statement emphasising the exploitative nature of big business, then they would be expected to make other negative statements about big business. The more frequently the person makes such statements, the more they are committed to making similar statements in future. Thus the stability of a person's party identification and of their political attitudes should be mediated through the frequency of their political talk.
Besides stability, the frequency of talk should also be related to the consistency of a person's political attitudes. Not only is a speaker expected to make similar statements about (for instance) big business on future occasions, but would also be expected to make other statements consistent with it. The concept of consistency is one of the trickier social psychological concepts. For political scientists, the notion refers to the conventional coherence of a person's attitudes on major dimensions such as left - right attitudes. The relationship between different beliefs on each of these dimensions is not one of logic, but of social expectation. Conventional ideological thinking would lead to the expectation that a person who believes that big business exploits the workers would also believe that the state should play an active role in re-distributing income. There is no logical relationship between these two propositions, but given conventional thinking, one would be surprised if someone agreed with one of these statements and disagreed with the other.
In the political science literature, the debate about the extent of ideological thinking has its origins in the work of Cambpell et al. (1960), and was later elaborated and extended by Converse (1964). The general point made by Converse was that few people in the USA organised their thinking about politics in terms of general ideological principles (such as left and right) from which attitudes towards particular issues or policies could be derived. Our goal here is not to argue about the extent of ideological thinking, which surely will depend upon a range of specifics (see Kinder & Sears, 1985, for a review), but to analyse some of the processes that may underlie such coherence. Converse himself related the extent of ideological thinking to education and to political involvement, the latter finding being further explored and amplified by Jennings (1992).
Returning to the relevance of talk to the construction of consistency, we suggest that making a statement about the exploitative nature of big business leads to expectations that the speaker will also make other statements that are conventionally consistent with this view. Again, the more frequently a person makes such statements the greater would be the expectation that they will make more statements of this sort. Thus the frequent elaboration of political attitudes in talk should lead to greater stability and consistency of a person's political attitudes.
The stability of a person's political orientation (their party identification and their political attitudes) can also be related to another aspect of their interpersonal environment - that of discussant homogeneity. This refers to whether the respondent thinks that the person or persons with whom they talk about politics generally share their views. Given the importance of the views of others for reinforcing the subjective validity of our beliefs (see Festinger,1950, 1954) it would follow that if a person generally talks about politics to others whom they perceive as agreeing with them, their political orientation and their political attitudes are more likely to be stable than if they talk to people who disagree with them.
The importance of discussant homogeneity for the stability of voting intention was originally shown by the pioneering work of Lazarsfeld and colleagues who studied the American presidential elections of 1940 (Lazarsfeld, Berelson & Gaudet, 1948) and 1948 (Berelson, Lazarsfeld & McPhee, 1954). More recently, comparable results have been reported by Huckfeldt and Sprague (1995). While that work looks at the stability of voting intention during periods of election campaigns, the data reported here looks at different indices (viz.party identification and political attitudes) and examines their stability over the period of a year in between general elections. While we expect discussant homogeneity to be related to the stability of both partly identification and attitudes, there is no clear basis for expecting a relationship between discussant homogeneity and the consistency of political attitudes. We see consistency as a function of the elaboration of attitudes that occurs in frequent talk, while simply believing that others around one share the same attitudes does not necessarily lead to such elaboration.
To recapitulate: the following hypotheses are investigated:
1. that frequency of political talk is positively related to the stability of political attitudes;
2. that frequency of political talk is positively related to the stability of party identification;
3. that frequency of political talk is positively related to the consistency of political attitudes;
4. that discussant homogeneity is positively related to stability of political attitudes;
5. that discussant homogeneity is positively related to stability of party identification.
METHOD
Two surveys were conducted of respondents who had originally taken part in the 1983 British Election Study (Heath, Jowell and Curtice, 1985). The first set of interviews were conducted by trained personnel from Social and Community Planning Research in Spring 1985. The second set were conducted one year later. The response details are as follows :
Study 1 - issued n=419, total contacted 367, interviews=283 (77.1% of those contacted)
Study 2 - issued n=283, total contacted 259, interviews=216 (83.4% of those contacted)
Preliminary analyses suggested little difference between those that dropped out and those that were reinterviewed. Hence the data reported will focus on those respondents on whom data are available at both points in time. However, since data is occasionally missing from a few respondents, the full 216 respondents are not always included in all the analyses. (Further information on sample characteristics is given in Evans et al. (1996) and Heath (1986).)
Three of the sets of questions asked are of particular interest here.
1. Political talk. During the first wave, respondents were asked the following questions specifically about talk. 'We are interested in finding out when people talk about politics, not just about elections, but about all kinds of political matters. I would like you to think of the two people you talk to most often about politics.' For each person mentioned (and some respondents did say they did not talk about politics, while others only mentioned one person), respondents were then asked, 'About how often do you talk to this person about politics... very often, say at least once a week ... fairly often, at least once a month.... not very often, several times a year... rarely, once a year or less?' Respondents were also asked about their perception of the other person's party identification.
2. Party identification. At both points in time respondents were asked: 'Do you think of yourself as Conservative, Labour, Liberal, Social Democrat or something else?' In the case of the Ôsomething else' response they were asked with which party they identified.
3. Political attitudes. At both points in time, respondents were asked to respond to a number of items that were concerned with left-right and with libertarian-authoritarian issues. There were 5 items relevant to the left-right dimension. (Examples: Big business benefits owners at the expense of workers; Government should redistribute income from the better off to those who are less well off). In the first wave Cronbach's alpha, a measure of the extent to which the five items form a single scale, was .82, in the second it rose slightly to .84. The libertarian - authoritarian scale consisted of 10 items. (Examples: Young people today don't have enough respect for traditional British values; People who break the law should be given stiffer sentences.) For each item, respondents were asked to indicate the extent of their agreement or disagreement on a 5-point scale. Cronbach's alpha in the first wave was .77, increasing slightly to .79 in the scond wave. (For full details of the scales and their development see Evans et al, 1996.)
RESULTS
Frequency of Talk.
Respondents were divided into two evenly sized groups of high talk and low talk as a function of their answers to the questions on frequency of talk about politics, following Evans et al (1996). The High Talk group consisted of respondents who discuss politics with more than one person often or occasionally; and the Low Talk group those who report not talking about politics or only talking to one person occasionally. This division led to 107 respondents being assigned to the High Talk group and105 to the Low Talk group. (Four cases had missing data.)
Frequency and Stability. The stability of political attitudes was evaluated by examining the correlation between the scores of respondents in the two talk groups at the two points in time. For the Left-Right scale, the High Talk group have a higher correlation than the Low Talk group (High Talk: r =.82; Low Talk: r = .77). However the difference between the two correlations is not significant (z =1.014). For the Libertarian - Authoritarian Scale, the correlation for the High Talk group is .89, while the correlation for the Low Talk group is .70. The difference between the correlations is highly significant (z = 3.99, p<.0001). Thus the relationship between frequency of talk and the stability of Libertarian - Authoritarian attitudes is clear, while the relationship between frequency of talk and stability of Left - Right attitudes, while in the right direction, is weak.
The data also show that there is a significant relationship between frequency of talk and the stability of party identification. Those who claimed the same party identification at the two points in time were classified as stable, whereas those who changed their party identification were put in the unstable category. In the High Talk group, 85.7% of respondents were stable in their party identification; while in the Low Talk group, 73.8% were stable, a difference which is statistically significant (Chi-square, corrected for continuity =3.88, df 1, p<.05).
Frequency and Consistency. Consistency was evaluated by examining the covariances between each item and every other item on the scale. The covariances for the High Talk group were expected to be higher than the covariances for the Low talk group. For Left-Right attitudes, the mean of the covariances between the items for the High Talk group was .91, while the mean for the Low Talk group was .72. The difference was evaluated though a Wilcoxon Matched-Pairs Signed Ranks Test. The difference was highly significant. The High Talk group showed higher covariances between items than the Low Talk group on 14 of the 15 comparisons (z = 3.24, p<.001). For the Libertarian - Authoritarian attitudes, the mean for the High Talk group was .50, while the mean for the Low Talk group was .27. The difference, evaluated through the Wilcoxon test, was highly significant. The High Talk group showed higher covariances between items than the Low Talk group on all 55 of the comparisons (z = 6.45, p<.001). Thus the consistency of both Left-Right and Libertarian - Authoritarian attitudes clearly are related to the frequency of talking about politics.
Homogeneity of Discussion Partners.
Respondents were divided into 3 groups as a function of whether they spoke to two other people with similar party identification, one similar other, or none similar. The numbers in the 3 groups were: None similar, 60; One similar, 79; and Two similar, 60 (17 cases were missing a party identity or responses on the relevant talk questions).
Table 1
Discussant Homogeneity and the Stability of Political Orientation
DISCUSSANT HOMOGENEITY
None One Two
Similar Similar Similar
Left-Right Stability 1 .82 .84 .81
Lib-Auth Stability 1 .78 .83 .77
Stability of Party Id 2 76% 82% 95%
1 Measured through correlations over time for each group.
2 Presented as the percentage of respondents in each group
who claim the same party identification at both time points.
[Note: Lib-Auth = Libertarian-Authoritarian; Party Id = Party Identification. For statistical details see text]
Table 1 presents the data on the stability of left-right and libertarian - authoritarian political attitudes as a function of discussant homogeneity. Again stability here is measured by the correlation over the two time periods of respondents attitudes within each of the 3 groups. Table 3 shows no effects of discussant homogeneity on the stability of attitudes along either dimensions. (For both dimensions, the interaction term of the linear trend x homogeneity resulted in F values of less than 1.00.)
Table 1 also shows the stability of party identification as a function of discussant homogeneity. The table shows clearly that stability increases with increasing homogeneity (Chi-square=8.54, df 2, p=.014; Mantel-Haenszel test for linear association =8.14, df 1, p=.004).
Further Analyses.
Given the non-experimental nature of the research design it is important to rule out alternative plausible explanations of the association between Frequency and Stability and between Discussant Homogeneity and Stability. Stability was measured for this analysis by creating a score for the extent of change in each respondent's position on the Left - Right and Libertarian - Authoritarian scales between the two waves of the survey. 1 Respondent's Stability scores were regressed onto several control variables that might also be expected to be related to political attitude formation: educational qualifications (a scale running from Ôno qualifications' to Ôdegree level'), strength of party identity (strong, medium, weak), homogeneity of discussion partners, age (in years) and sex (being a woman). For analyses of the stability of left-right and libertarian-attitudes we further controlled for the stability of party identity; and for analyses of the stability of party identity we controlled for the stability of left-right and libertarian-authoritarian attitudes.
Table 2
Regressions of the Stability of Left-Right and Libertarian-Authoritarian Attitudes on Frequency of Talk and Control Variables.
Left-Right Lib-Auth Two
Variables
Stability of Party ID .11 -.03
Discussant Homogenity -.09 .01 .77
Frequency of Talk .05 .19**
Educational .09 .01
Qualifications
Age -.12 .03
Women .03 .05
Strength of Party ID .03 .11
R2 .05 .06
(** p<.01)
[Note: Lib-Auth = Libertarian-Authoritarian; Party ID = Party Identification. Coefficients are standardised betas.]
Table 2 presents an OLS regression of the stability of left - right and libertarian - authoritarian attitudes on to the Frequency of Talk and Discussant Homogeneity measures and the controls. There are no significant coefficients for the Left - Right variable. The only significant coefficient for the Libertarian - Authoritarian variable is the frequency of talk. These results support those presented earlier. Frequency of talk is related to stability of Libertarian attitudes, even when the other variables are included. Also supporting the earlier analyses, though contrary to the hypotheses advanced, Discussant Homogeneity was unrelated to either set of attitudes.
Table 3
Logistic Regression of Stability of Party Identity on Discussant Homogeneity and Control Variables.
Variables Model 1 Model
Stability of Left-Right .36 (.25) .32 (.26)
Attitudes
Stability of Lib-Auth .18 (.33) .16 (.33)
Attitudes
Discussant Homogenity 1.21 (.31)** .66 (.34)*
Frequency of Talk -.15 (.17) -.22 (.20)
Educational .11 (.11) .07 (.11)
Qualifications
Age .0 (.01) -.02 (.02)
Women -.16 (.40) .05 (.43)
Strength of Party ID 1.50 (.36)**
Chi Square 23.0 46.1
df 7 8
(* p<.05; ** p<.01)
[Note: Lib-Auth = Libertarian-Authoritarian; Party ID = Party Identification. Coefficients are log-odds coefficients with standard errors in parentheses.)
Table 3 presents the regression of stability of party identity on to the Frequency of Talk and Discussant Homogeneity measures and the controls. As the dependent variable is dichotomous, these analyses were conducted using logistic regression to obtain parameter coefficients. Logistic regression is one of a group of techniques which calculate the odds of being in one response category rather than another. It is therefore more suitable for dichotomous dependent variables than is linear regression. The first model shows that the only significant coefficient is for the effect of Discussant Homogeneity. In the second model, there is also a strong effect of strength of party identification on stability, but even then the coefficient for Discussant Homogeneity remains significant. These results too support the earlier analyses where there was a strong relationship between Discussant Homogeneity and Stability of Party Identity, but does not show a relationship between Stability of Party Identity and Frequency of Talk.
DISCUSSION
In the Introduction it was argued that talk was one of the ways in which a person's political identity was expressed and maintained. We suggested that making certain political statements leads to expectations on the part of the listener that the speaker will make similar statements in the future, and expectations that the speaker will make other statements that are consistent with it. The more frequently such statements are made, the firmer the expectations they give rise to. The speaker too understands the social demands for consistency and lives up to them. These are some of the processes through which attitudinal consistency and stability are socially constructed. This argument led to the hypotheses that the frequency with which a person talks about political matters would be related to the consistency and stability of their political attitudes. The data broadly support these hypotheses. The clearest evidence concerns the relationship between frequency of talk and consistency. For both the left - right and the libertarian - authoritarian dimensions, the high talk group showed more consistency than the low talk group. However, the results concerning the relationship between frequency of talk and the stability of political orientation is more complex. There is a clear relationship with the stability of libertarian - authoritarian attitudes, but the relationship with left - right attitudes and the stability of party identification though in the right direction are less clear. Perhaps the traditional left - right dimension is so clearly established in society, that interpersonal expectations with regard to a person's position on that dimension are quickly established, and thus less in need of frequent reiteration to become stable. Taken as a whole these results provide clear support for the hypotheses and for the orientation from which they were elaborated.
However, talk is clearly not the only way in which people's political attitudes are engaged, and talk itself may be a reflection of political interest and involvement. The current study cannot disentangle causality and has no data which will enable a comparison of different measures of political involvement. It would be surprising if frequency of talk were not also linked to other activities such as exposure to newspapers and television programmes dealing with political matters, knowledge about politics and participating in politics in other ways, like displaying political stickers. An elaboration of the different forms of political involvement will enable us to be clearer about the processes involved in the formation and maintainance of political orientation and about the functions these different activities fulfil.
The results concerning discussant homogeneity offer clear support for one hypothesis but not for the other. It had been hypothesised that believing that other people around you share your political orientation should provide social validation for these beliefs and be related to their stability. There was a clear relationship between the stability of party identification and discussant homogeneity. However, there was no relationship between discussant homogeneity and the stability of left - right or libertarian - authoritarian attitudes. One possible explanation for this could be that the initial level of attitudinal stability is itself very high. Another possibility is methodological. In this survey, respondents were asked if they thought that their discussants were supporters of the same party as they were. Respondents were not asked if they thought the other persons agreed with them on specific issues (such as those on the attitude scales). Perhaps if they had been, different results would have been obtained. However, this argument does not seem very plausible since it is surely reasonable to expect that those respondents who thought that the discussant(s) had the same party identification as they did also held broadly the same attitudes.
A further possibility revolves around the important ambiguity in the notion of 'talking about political matters'. We have assumed that talking politics at least sometimes involves talking about issues like social inequality and law and order. This assumption receives implicit support from the fact that there is a relationship between frequency of talking politics and the consistency and stability of left - right and liberal - authoritarian attitudes. But we do not know what else talking politics consists of. Much of it may involve general evaluative statements about political events or parties, or focus on the personalities and activities of leading political figures. These too are ways in which party identification can be expressed and identified. Such statements could express a person's party identification but since such talk does not involve the elaboration of political attitudes along the dimensions that are elicited in conventional attitude scales, the stability on such measures is unaffected.
While we have often used the term 'discussant homogeneity', it should be stressed that the measure we have is one of 'perceived' homogeneity. Huckfeldt and Sprague (1987,1995) have shown that the accuracy of the perception of the others' attitudes is partly a function of the respondent's own attitudes as well as features of the social environment. The difference between 'perceived' homogeneity and 'actual' homogeneity is of course important for understanding the processes involved. People may have differential tendencies to seek out others with similar views. It could be that people who seek stability are more likely to seek out others who agree with them, and also to over-estimate the degree of agreement. Homogeneity would be the vehicle through which stability is achieved, rather than the cause of the stability. The processes may well be mutually reinforcing. The consequences of individual differences in tendencies towards seeking stability could readily be explored through laboratory studies.
In a recent textbook on the psychology of attitudes, Eagly and Chaiken (1993) point to the lack of attention to social context as one of the biggest single limitations of psychological research on attitudes. This paper has attempted to present some data which disentangles some of the processes that are relevant for understanding attitude stability and consistency by locating these in the social contexts in which attitudinal expression takes place. The general ideas are readily extensible to other domains - for example to our attitudes towards religion, to work and leisure, to health and illness.
A fuller understanding of attitudes may well be achieved by examing the social context in which these and other attitudes are expressed.
Footnote
1 The analyses presented use raw change scores. The use of change scores with standardised scores does not affect the findings. We are unable to engage in a similar multivariate analysis of consistency because there is no suitable proxy measure of consistency at the individual level.
REFERENCES
Baumeister, R. F. (1982). A self-presentational view of social phenomena. Psychological Bulletin, 91, 3-26.
Berelson, B. R., Lazarsfield, P. F. and McPhee, W. N. (1954). Voting: a study of opinion formation in a presidential campaign. Chicago: Univ. of Chicago Press.
Campbell, A., Converse, P. E., Miller, W. E. and Stokes, D. E. (1960). The American voter. New York: Wiley.
Converse, P. E. (1964). The nature of belief systems in mass publics. In D. E. Apter (ed.), Ideology and discontent. New York: Free Press. pp. 206-261.
Eagly, A. H. and Chaiken, S. (1993). The Psychology of Attitudes. New York: Harcourt Brace Jovanovich College Publishers.
Evans, G., Heath, A and Lalljee, M. (1996). Measuring left-right and libertarian-authoritarian values in the British electorate. British Journal of Sociology, 47, 1, 93-112.
Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117-140.
Festinger, L. (1950). Informal social communication. Psychological Review, 57, 271-282.
Heath, A.F. (1986). Methodological aspects of attitude research: end of award report to the ESRC.
Heath, A., Jowell, R. and Curtice, J. 1985 How Britain Votes , Oxford: Pergamon Press.
Huckfeldt, R. and Sprague, J. (1987). Networks in Context: The Social Flow of Political Information. American Political Science Review, 81, 1197-1216.
Huckfeldt, R. and Sprague, J. (1995). Citizens, Politics and Social Communication: Information and Influence in an Election Campaign. Cambridge Studies in Political Psychology and Public Opinion. Cambridge University Press.
Jennings, M. K. (1992). Ideological Thinking Among Mass Publics and Political Elites. Public Opinion Quarterly, 56, 419-441.
Kinder, D. R. and Sears, D. O. (1985). Public Opinion and Political Action. In G. Lindzey and E. Aronson (eds.) The Handbook of Social Psychology, 3rd edition, vol. 2: Special Fields and Applications. New York: Random House.
Lalljee, M., Brown, L. B. and Ginsburg, G. P. (1984). Attitudes: Disposition, behaviour or evaluation? British Journal of Social Psychology, 23, 233-244.
Lazarsfield, P. F., Berelson, B. and Gaudet, H. (1948). The people's choice (2nd edition). New York: Columbia Univ. Press.
Petty, R. E. and Cacioppo, J. T. (1982). Attitudes and Persuasion: Classic and Contemporary Approaches. Dubuque. IA: W. C. Brown.
Schlenker, B. R. (1982). Translating actions into attitudes: An identity-analytic approach to the explanation of social conduct. In L. Berkowitz (ed.), Advances in Experimental Social Psychology, vol. 15. New York: Academic Press.
Return to CREST Page