Three aspects of political sophistication - which one can be blamed for generating bias?

Transcription

1 Three aspects of political sophistication - which one can be blamed for generating bias? Veronika PATKÓS patkos.veronika@tk.mta.hu Hungarian Academy of Sciences, Centre for Social Sciences Corvinus University of Budapest, Institute for Political Science The influential book of Achen and Bartels (2016) further reinforces the existing concerns in the literature about the link between the level of voters political sophistication and partisan bias in their political evaluations. The paper differentiates between three important elements of political sophistication, investigating the effect of 1) education 2) the interest in politics and 3) the time spent on gaining political information. Using the first five rounds of the European Social Survey it identifies political groups based on voters winner-loser positions, and it investigates the extent of bias in the political evaluations of these two groups. The main question of the paper is whether partisan bias in political evaluations is inherently connected either to political interest, awareness or levels of education. The results show that while being more interested in politics is clearly connected to having more biased political evaluations, the level of education and the time spent on gaining political information are not clearly related to more biased views. While there is a positive correlation between the level of bias and all the three analysed variables, it is relatively weak. Moreover, when controlling for political interest, the effect of the level of education on bias is actually negative, yet not significant. Therefore, results are mixed, but compared to former evidence, they offer some room for optimism regarding the link between sophistication and biased views. Paper to be presented at the ECPR General Conference, 9 September 2017, Oslo

2 INTRODUCTION The question whether politically sophisticated people make better political decisions is of particular interest when thinking about the functioning of democracy. Generally, both ordinary people and political scientist incline to assume a link between political sophistication and a capability to form better decisions regarding politics. Following this line of thought, many argue that raising the level of education and political awareness of citizens improves the democratic performance of a political community. While it is hard to operationalize the goodness of political decisions, one important element is the extent of partisan bias present in them. In this regard, the literature offers contradicting hypotheses. Political sophisticates are the people who have a motivation to seek and process political information, an ability to handle that information, and who are willing to make an effort to do so. On the one hand, theoretically, the more voters are educated and well-informed, the less they depend on such easily available heuristics as party labels. To put it with Dalton s (1984, p. 281) words, greater political skills enable citizens to be functionally independent of party cues. On the other hand, as many scholars note including Dalton as well political interest and the level of education is positively correlated to partisanship, and they imply a psychological involvement in politics. This resonates with what we know since the seminal study of Converse and his co-authors (1960), who showed that both political interest and political sophistication are linked to partisan attachments. Since then, voting behavior researchers have regularly found that politically wellinformed, sophisticated and interested people reported stronger partisan attachments and applied more partisan bias while processing political information (Abramowitz & Saunders, 2008; Achen & Bartels, 2016; Campbell et al., 1960; Lodge & Hamill, 1986; Lodge & Taber, 2005; Rose & Mishler, 1998, pp ; Tilley & Hobolt, 2011). Besides the association between age and partisanship, this proved to be the most stable result regarding the correlates of partisanship. Thus, the relationship between political sophistication and partisan bias is far from being clear: theoretically, political sophisticates should rely more on their own evaluations and less on partisan cues, but empirically they have resulted to be the most biased in a partisan sense. The question of the paper is that which elements of political sophistication might account for higher levels of bias: is partisan bias in political evaluations related to political interest, awareness or higher levels of education. I tested this question on the sample of 30 European countries, using the repeated cross-sectional data set of the European Social Survey between

3 2002 and I measured political bias based on the difference between the political satisfaction of political camps, meant as governing and opposition party voters. While results show that according to previous results the level of bias is correlated to all three key variables of the research, the main finding of the paper is that when controlled for political interest, education and awareness do not exacerbate bias. Education and awareness are not important predictors of bias on their own, the only important factor is political interest. The paper is organized as follows. The next section is an overview of the literature regarding the links of sophistication and bias. In the third part I describe the database and the variables of the research, with special attention to the method used to estimate bias on opinion formation. In the fourth part I present the regression models and describe the results. The fifth section summarizes the results and it highlights the implications of the paper from the viewpoint of democratic theory. THEORETICAL FRAMEWORK Are politically sophisticated people more autonomous opinion formers and thus better political decision makers? Using the definition of Luskin (1990) political sophisticates are those who have an elaborate political belief system, with numerous and highly organized political cognitions (see also Converse, 2006). To obtain a highly organized belief system with numerous elements one should possess the necessary motivation to seek and process political information, an ability to handle that information, and to make an effort to do so. Admitting that these aspects may be interrelated, I operationalize motivation with the level of political interest, ability with the years spent in the education system, and effort with the time spent on reading political news. While it is challenging to operationalize the quality of political opinions, an important aspect to explore is how much they are biased in a partisan sense. The cognitive capacities of the political sophisticates enable them to articulate a political opinion and to make reflected choices, thus they should rely less on party heuristics (Dalton, 1984). A combination of higher cognitive capacities, more information and less partisan cues should lead to more grounded and rational decisions than that of fewer capacities and information and more partisan cues. To cite some example in line with this theory, experimental evidence of Anduiza, Gallego and Muñoz (2013) shows that politically more informed people apply less bias when evaluating corruption cases, and the results of Taber et al. (2009) are similar, yet inconclusive. Interestingly however,

4 the majority of empirical evidence regarding the level of partisan bias in political evaluations does not really support the above hypothesis. Descriptive results of Abramowitz and Saunders (2008) about the American electorate show that interest, awareness and partisanship go hand by hand, as Campbell and his co-authors (1960) showed it half a century ago. They found that neutral or centrist political attitudes are still strongly correlated to uninformedness and disinterest, while those who hold consistently liberal or conservative views are generally the most informed politically. Therefore, the most interested and politically aware are strongly polarized in an ideological sense. Reviewing many former results, Achen and Bartels also (2016, 294) argue that political rationalization is most powerful among people who are well- informed and politically engaged, since their fundamental political commitments tend to be most consistent and strongly held. As they conclude, the politically attentive and interested are less likely to be responsive to the actual political situation, as their political attitudes are mostly stable. Lodge and Hamill (1986) apply the framework of cognitive sciences and reach to similar results. They label people with a sound knowledge structure about politics (high interest and high knowledge) as partisan schematics and those who lack such elaborate knowledge structure as partisan aschematics. They find that partisan schematics recognize and recall political information much better than aschematics, as they are capable to accurately categorize policy positions as Republican or Democratic. But at the same time, partisan schematics show also a serious partisan bias while processing political information. They remember schema consistent information more, and they restructure their memories about political information in a strongly distorted way. (It is worth to note the alternative that partisan aschematics offer, which is to not to remember political information at all.) As the authors state, the knowledge of political sophisticates generates expectations that often-times help but at other times hinder their ability to deal with schema-relevant information (Lodge & Hamill, 1986, p. 517). On the other hand, Lodge and Taber (2005) show that the relationship between interest, information and bias has not only cognitive roots, but also affective ones. They found that the affective charge of previously evaluated political concepts was activated automatically, faster than conscious responses. Sophisticated subjects had longer reaction times when evaluating the affective charge of words (like sunshine, or toothache ) after an incongruent prime with political content (the name of leaders, parties or policies they disliked), while less informed and interested subjects were less influenced by those affective heuristics. As they conclude, while

5 bias is generally present in political evaluations, political sophisticates with strong political attitudes are the most biased information processors. All in all, the literature offers extensive evidence regarding a positive link between political sophistication and biased views, and far less empirical support to an opposite or no effect. Experimental evidence (mainly from the US) also shows that this bias has both cognitive and affective roots. But which element of political sophistication can be blamed for such a bias? The aim of the paper is to clarify the relationship between bias and the components of political sophistication: education level, political awareness and political interest. METHOD, DATA AND VARIABLES When investigating the links between partisan bias and the elements of political sophistication the most challenging methodological task is the measurement of partisan bias. The majority of empirical works about partisan bias has an experimental design, thus the methods used by them are hardly applicable in a comparative, cross-country research. Nevertheless, following the logic of such experimental researches, to estimate partisan bias it is essential to identify 1) partisan groups and 2) a politically relevant topic on what the views of these partisan groups evidently differ, in a way that one important source of such difference be partisan attachment. Identifying relevant political groups or political camps in different party systems which are composed of various number of parties of various size and ideological position is not an easy task. Here I used a strongly simplifying yet straightforward method to differentiate between political camps: the government-opposition status of parties. The two poles fundamentally reflect who are with who on the political scene of the given country. Another option to make a distinction between political camps could be to use left-right scales, but the sense of left-right classification of parties across countries and political contexts is always questionable. Comparing the political evaluation of voters based on their winner-loser status is not a novelty, there is a growing body of literature that approaches the so called winner-loser gap (Anderson & LoTempio, 2002; Anderson & Tverdova, 2001; Blais & Gélineau, 2007; Brunell & Buchler, 2012; Craig, Martinez, Gainous, & Kane, 2006; Curini, Jou, & Memoli, 2012; Howell & Justwan, 2013; Singh, Karakoç, & Blais, 2012; Singh, Lago, & Blais, 2011). Nevertheless, while the literature is dynamically growing, the winner-loser gap has been rarely linked to partisan bias in an explicit way.

6 After identifying relevant political groups, the other important task is to choose a topic with clear partisan relevance which can be adequate to estimate the extent of bias. A general, extensively surveyed and easily interpretable survey item with clear partisan relevance (if we identify partisan groups as governing and opposition party voters) is satisfaction with the performance of the national government. While it may contain evaluations about single policy decisions, the evaluation of the government is strongly related to previously established beliefs. Government supporters are supposed to have positively biased evaluations, while the supporters of opposition parties are likely to have negatively biased evaluations on how the government performs. Across the difference between the two groups evaluations it can be shown how strong the partisan perceptual screen is: how much rosier the evaluations of governing parties voters are compared to the evaluations of opposition parties voters. The analysis relies on the first five rounds of the European Social Survey s data set, excluding countries which are not members of the European Economic Area. In order to make a clear distinction between governing and opposition party voters, I also excluded the cases when fieldwork period overlapped with parliamentary elections and the cases when surveys were conducted when a new government after the parliamentary elections has not yet entered in office. I also excluded the voters of two smaller parties that left their government during the fieldwork period. Neither do I include cases when the incumbent government was a technocratic one. Data about government composition and date of elections and investitures came from the Comparative Political Data Set project (Armingeon et al. 2013; 2015). A dummy variable indicates whether voters are government or opposition supporters. The process finally resulted a pooled data set of more than individual subjects from 30 countries, and 100 countrycases (see Appendix 1). However, the statistical models rely on a significantly reduced number of observations compared to the original data base, due to the high rate of subjects who did not participate at the previous elections or refused to answer which party did they vote for. Variables According to the above mentioned, the dependent variable of the research is satisfaction with the national government 1, measured on a 0 to 10 scale, ranging from extremely dissatisfied to 1 The exact wording of the question was Now thinking about the [country] government, how satisfied are you with the way it is doing its job?

7 extremely satisfied. A key independent variable of the research is being a governing or an opposition party voter. Other key variables are the main elements of political sophistication. The components that I analyse here are political awareness, level of education and political interest. I measure political awareness with the time spent on reading political news on an average weekday. Answers range from 0 (not at all) to 7 (more than 3 hours), where 1 point means additional 30 minutes on an average. Education is measured by the number of years spent in the education system, assuming values between 0 and 56 (with the 99.9 per cent of observations below 30 years). Political interest is measured on a four-point scale which ranges from (1) not at all to (4) very interested. Descriptive statistics of the dependent and independent variables are reported on Table 1. The three aspects of sophistication are indeed correlated, as expected (see Appendix 2). Table 1. Descriptive statistics of the dependent and independent variables. N Minimum Maximum Mean Std. Deviation Satisfaction with the government Government voter Newspaper reading, half an hours Years spent in the education system Political interest Valid N (listwise) Anyhow, the main questions of the research do not focus merely on the effect of government/opposition status, but rather on the interaction of such status and the components of political satisfaction. Thus, the effects of primary interest are interaction effects between voter s government/opposition status and the elements of political sophistication. - government/opposition status x education - government/opposition status x political interest - government/opposition status x political awareness Additional individual level control variables are gender and age, both in a simple form, both in interaction with voter status. In the models, to reduce multicollinearity caused by multiple interaction terms, ordinal and continuous independent/control variables are standardized. ANALYSIS

8 The dependent variable (measured on a 0 to 10 scale) is not the best neither to a linear model, nor to an ordinal one. Although it might be transformed to make it adequate to an ordinal model, this would imply a severe loss of information. Therefore, to not to lose information, I decided to use a linear model, even if the dependent variable is not exactly a continuous one. As the data set is a repeated cross sectional one, I considered country effects and general time trends in political satisfaction across Europe by controlling for ESS rounds and clustering robust standard errors by countries, and (in some models) by controlling for countries. The results of OLS regressions are reported on Table 2. As the question of the paper is which element of political sophistication influences partisan bias, to have a fuller picture, models 2 to 4 include either interest, education or interest in interaction with voter status. Table 2. OLS regressions with clustered robust standard errors (30 clusters by countries). Dependent variable: satisfaction with the government. Entries are regression coefficients with standard errors in parentheses below. Population and design weights applied. Model 1. Model 2. Model 3. Model 4. Government voter 1.44*** 1.691*** 1.443*** 1.683*** (0.167) (0.168) (0.17) (0.169) Government voter x age 0.233*** 0.304*** 0.242*** 0.247*** (0.067) (0.06) (0.053) (0.053) Government voter x gender *** *** (0.051) (0.045) (0.054) (0.044) Government voter x reading political news *** (0.04) (0.027) Government voter x interest 0.342*** 0.354*** (0.08) (0.083) Government voter x education (0.056) (0.066) Control variables Age (0.045) (0.044) (0.05) (0.05) Gender (0.063) (0.061) (0.065) (0.061) Reading political news (0.034) (0.032) (0.032) (0.032) Political interest * * 0.08* (0.054) (0.044) (0.051) (0.044) Education 0.074** ** 0.07** (0.029) (0.033) (0.026) (0.026) Rounds (0.194) (0.193) (0.194) (0.193) * * * -0.33* (0.184) (0.183) (0.184) (0.182) (0.177) (0.177) (0.178) (0.177) ** ** ** ** (0.189) (0.19) (0.189) (0.191) Constant 3.773*** 3.651*** 3.77*** 3.658***

9 Linear Prediction Linear Prediction (0.154) (0.154) (0.156) (0.154) R *p<0.1 **p<0.05 ***p<0.01. N= As for the research question, we have to consider the interaction effects in the first part of the table. Significant effects show that there is a significant difference between the satisfaction of governing/opposition party voters with the government between the cathegories of the other analysed variable. The results reported here are in line with the oldest results of the partisanship literature, saying that age and political interest are important predictors of partisanship. The results of these models show that these two factors are important also in predicting partisan bias. Linear predictions for these two variables are reported on Figure 1. On the left side of the figure it can be seen that those governing party voters who are absolutely not interested in politics (but still cast their vote on the elections) are significantly closer in their evaluations to the not-at-all-interested opposition party voters than the very interested opposition- and governing party voters are. Between the average satisfaction scores of the two not at all interested groups there is a difference of 0.84 point, which raises to a difference of almost two entire points (1.98) when speaking about the most interested supporters of the government and of the opposition. The evaluations change considerably also by the number of ages, which is reported on the right side of Figure 1. The evaluations about the performance of the government are more similar in the two groups for the youngest voters even if the initial 1 point difference is also notable and the difference between the two groups grows to more than 2 points for the oldest. Predictive Margins of voter status with level(95)% CIs Predictive Margins of voter status with level(95)% CIs Standardized values of (political interest) Opposition voter Government voter Standardized values of (age) Opposition voter Government voter Figure 1. Predictive margins of voters government/opposition status with a 95% confidence interval, with regard to political interest and age. Dependent variable is satisfaction with the government, ranging from 0 to 10. When political interest is included, the other elements of political sophistication have no significant effect on the level of partisan bias. Omitting interest, the effect of political awareness

10 the time spent on reading political news becomes significant, and it points to the expected direction. Education instead has clearly no effect. These results are relatively stable across model specifications. The main results (government voter x interest; government voter x age) hold their significance and effect sizes are similar to what is reported here when controlling (or not) to ESS rounds and country dummies. It is important to note that the research includes voters from all over Europe who participated at the last national elections, and were willing to report their vote choice. This way, it excludes the politically less aware and less interested part of the electorate. One additional finding of the research is that the well-known correlations between partisan bias, age and interest hold their significance even when excluding the politically unsophisticated. CONCLUSION The paper investigated the effect of three elements of political sophistication on the level of bias in voter s evaluations. The results show that political interest and political bias are strongly linked. The other two elements, education and political awareness, do not have a significant effect when interest is also included. The contribution of the paper is twofold. First, it used an innovative method to estimate the level of bias in political evaluations in a comparative context. The majority of results which explicitly focus on partisan bias (and not on the wider concept of partisanship) rely on experimental data. This research is experimenting to measure partisan bias with using survey data, proposing a method that can be used for international comparisons. Second, it tried to clarify the relationship between three aspects of political sophistication and partisan bias, which is of key importance when thinking about the prerequisites, scopes and potential outcomes of democratic functioning. The topic has long been characterized by a contradiction between theory (or simply, intuition) and empirical evidence these results may bring theory and empirics somewhat closer to each other. The question whether political sophistication decrease bias and improve the quality of political opinions and decisions have important theoretical and empirical consequences. Theoretically, raising the levels of education and awareness should help people to have more grounded and more rational decisions, thus it has the potential to improve the quality of democratic functioning. Political sophisticates might control more effectively their irresponsible or lying representatives. If their preferred party or politician makes a mistake, they might recognize such failures earlier than the less aware and well-informed. Moreover, they might perceive more

11 correctly the real costs of appealing but on a long run deleterious policies, which are attractive to naïve voters. If the real costs of harmful policies are more obvious for a more educated and politically aware electorate, raising the levels of awareness may be a medicine against populism, which is considered one of the major threats to contemporary democracies. If awareness and education lead to more grounded and rational decisions, an improvement in democratic functioning is possible, and projects aiming to raise the levels of education and civic activity are likely to be effective tools for this purpose. All these hypotheses are normatively appealing and widely shared, anyhow, this research offers no support to them. To put it in a normative frame, the good news of the research is that the educated and politically aware are not significantly more biased opinion formers than the less educated and aware. This result opens some room for optimism compared to what Shani (2006, 31; cited by Achen & Bartels, 2016, 279) found by testing factual knowledge on a sample of US citizens, where political knowledge was in important predictor of elevated bias. The bad news is that neither they are significantly less biased. Education and awareness are not important predictors of bias on their own. The only component of political sophistication that has a robust significant (positive) effect on the level of bias across many model specifications is political interest. Thus, the paper stresses the importance of political interest, and it brings attention to the importance of further research on the role of political interest on the formation of political attitudes and opinions.

12 REFERENCES Abramowitz, A. I., & Saunders, K. L. (2008). Is polarization a myth? The Journal of Politics, 70(02), Achen, C. H., & Bartels, L. M. (2016). Democracy for Realists: Why Elections Do Not Produce Responsive Government. Princeton University Press. Anderson, C. J., & LoTempio, A. J. (2002). Winning, losing and political trust in America. British Journal of Political Science, 32(02), Anderson, C. J., & Tverdova, Y. V. (2001). Winners, losers, and attitudes about government in contemporary democracies. International Political Science Review, 22(4), Anduiza, E., Gallego, A., & Muñoz, J. (2013). Turning a Blind Eye Experimental Evidence of Partisan Bias in Attitudes Toward Corruption. Comparative Political Studies, 46(12), Blais, A., & Gélineau, F. (2007). Winning, losing and satisfaction with democracy. Political Studies, 55(2), Brunell, T. L., & Buchler, J. (2012). How Does Winning, Losing, and Electoral Competitiveness Affect Voters Attitudes Toward Government? Evidence from Three Western Democracies. Evidence from Three Western Democracies. Retrieved from Campbell, A., Converse, P. E., Miller, W. E., & Stokes, D. E. (1960). The American Voter. New York and London: John Wiley & Sons, Inc. Converse, P. E. (2006). The nature of belief systems in mass publics (1964). Critical Review, 18(1 3), Craig, S. C., Martinez, M. D., Gainous, J., & Kane, J. G. (2006). Winners, losers, and election context: Voter responses to the 2000 presidential election. Political Research Quarterly, 59(4), Curini, L., Jou, W., & Memoli, V. (2012). Satisfaction with democracy and the winner/loser debate: The role of policy preferences and past experience. British Journal of Political Science, 42(02), Dalton, R. J. (1984). Cognitive mobilization and partisan dealignment in advanced industrial democracies. The Journal of Politics, 46(01), Howell, P., & Justwan, F. (2013). Nail-biters and no-contests: The effect of electoral margins on satisfaction with democracy in winners and losers. Electoral Studies, 32(2), Lodge, M., & Hamill, R. (1986). A partisan schema for political information processing. The American Political Science Review, Lodge, M., & Taber, C. S. (2005). The automaticity of affect for political leaders, groups, and issues: An experimental test of the hot cognition hypothesis. Political Psychology, 26(3), Luskin, R. C. (1990). Explaining political sophistication. Political Behavior, 12(4), Rose, R., & Mishler, W. (1998). Negative and positive party identification in post-communist countries. Electoral Studies, 17(2),

13 Singh, S., Karakoç, E., & Blais, A. (2012). Differentiating winners: How elections affect satisfaction with democracy. Electoral Studies, 31(1), Singh, S., Lago, I., & Blais, A. (2011). Winning and Competitiveness as Determinants of Political Support*. Social Science Quarterly, 92(3), Taber, C. S., Cann, D., & Kucsova, S. (2009). The Motivated Processing of Political Arguments. Political Behavior, 31(2), Tilley, J., & Hobolt, S. B. (2011). Is the government to blame? An experimental test of how partisanship shapes perceptions of performance and responsibility. The Journal of Politics, 73(02),

14 APPENDIX 1 Countries included in the analysis by ESS round Austria * * * Belgium * * * * * Bulgaria * * * Czech Republic * * * Croatia * Cyprus * * Denmark * * * * * Estonia * * Finland * * * * * France * * * * * Germany * * * * Greece * * * Hungary * * * * * Italy * Iceland * Ireland * * * * Latvia * Lithuania * Luxembourg * * Netherlands * * * Norway * * * * * Poland * * * * * Portugal * * * * Romania * Slovakia * * * * Slovenia * * * * Spain * * * * * Sweden * * * * Switzerland * * * * * United Kingdom * * * * *

15 APPENDIX 2 Correlation table of the variables. Entries are Pearson correlation coefficients. Government/ opposition voter (1/0) Age Gender Time spent on reading political newspapers Years spent in education Political interest Satisfaction with the government Government/opposition voter (1/0) 1.036**.007** ** ** Age.036** 1.042**.202** -.281**.101**.009** Gender.007**.042** ** -.049** -.141** -.033** Time spent on reading political newspapers ** -.090** 1.095**.323**.058** Years spent in education -.033** -.281** -.049**.095** 1.278**.046** Political interest ** -.141**.323**.278** 1.120** Satisfaction with the government.284**.009** -.033**.058**.046**.120** 1 ** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed).