Does Ignorance Matter ? The Relative Importance of Civic Knowledge and the Human Tendency to Engage in Motivated Reasoning

It has long been understood that political knowledge in the U.S. is very low. For those who care about the quality of American democracy, this is a big problem. In attempting to find a solution, many people often blame education. While increasing civic knowledge is a laudatory goal, increased political sophistication does not necessarily turn individuals into good democratic citizens. Research in cognitive and social psychology paints a picture of people as motivated reasoners. Instead of having an open-minded engagement with issues, individuals typically only seek, see, and understand information in a manner that reinforces what they already believe. Here, we examine motivated reasoning and argue that the strongest partisans and the most committed ideologues will be the most susceptible to holding contradictory policy positions with regard to same-sex marriage and religious freedom.

The importance of civic literacy is, and has long been, an axiom of democratic theory; a "generally-accepted belief that civic knowledge is an important foundation of democratic selfgovernment" (McCabe and Kennedy 2014).Americans generally agree that a basic understanding of the structure and philosophy of government is a necessary precondition for productive political engagement or policy debate; a mutual understanding of the constitutional framework, and that agreement on the meaning of basic legal, economic and scientific terminology is necessary if there is to be common ground for discussion.
Former U.S. Representative Lee Hamilton summarized this consensus, writing in 2003 The truth is, for our democracy to work, it needs not just an engaged citizenry, but an informed one.We've known this since the nation's earliest days.The creators of the Massachusetts Constitution of 1780 thought the notion important enough to enshrine it in the state's founding document: "Wisdom and knowledge, as well as virtue, diffused generally among the body of the people," they wrote, are "necessary for the preservation of rights and liberties.
Those who are concerned about America's toxic political environment contend that the lack of accurate, basic civic information is the major culprit, and civic ignorance is at the root of American's current polarization.That belief may be misplaced, or at the least, oversimplified.
The widespread concern over the polity's admittedly low levels of civic knowledge fails to take into account recent research on motivated reasoning, which is the often unconscious selective perception and interpretation of information so as to reinforce beliefs already held.The troubling findings in this research are that such reasoning is, if anything, more likely to occur among more sophisticated voters than with their less politically knowledgeable peers.Political scientists and pundits alike are left with the question: Which matters most?Which is more likely to predict increased political polarization, civic ignorance or ideological rigidity, and how do these factors interact?
A copious literature confirms the existence of a civic deficit: Only 36 percent of American citizens can correctly name the three branches of government (Annenberg Public Policy Center Judicial Survey 2007); 36 percent of 12th grade students fail to achieve a basic level of civic knowledge (National Center for Education Statistics 2011); and, only 35.5 percent of American teenagers can correctly identify "We the People" as the first three words of the U.S. Constitution (National Constitution Center Survey 1998).The National Assessment of Education Progress (NAEP) 2010 report on civic competencies found that barely a quarter of the country's 4 th , 8 th and 12 th graders could be considered proficient in civics (National Center for Education Statistics 2011).Numerous other studies confirm the extent of our civic deficit (Bennett 1995;Caplan 2008;Converse 2000;Delli Carpini andKeeter 1991, 1996;Shaker 2012).
Given the strength of the evidence demonstrating a broad public deficit of accurate civic information, a growing number of researchers and educators are working to identify best practices and to improve civic education in the schools.These initiatives to raise awareness of the issue and to identify measures that may ameliorate it are important.At the very least, a shared understanding of basic social and political institutions is necessary for communication to occur-a common reality, after all, is much like a common language.If we are looking to improvements in civic knowledge to reverse the political polarization that has paralyzed so much of our political system, however, the emergence of new lines of research in political psychology suggests we may well be disappointed.
In what follows, first we show that education or increased political interest and awareness are not enough to defeat the pernicious effects of a highly-polarized political system.Extant research (our own research included), routinely finds that the politically sophisticated are also the most partisan (Federico and Hunt 2013).Second, we argue that when these highly-polarized politically-sophisticated individuals engage in politics, motivated reasoning (Kunda 1987(Kunda , 1990) ) produces a rather paradoxical result.Those with the strongest partisan ties and the most extreme ideological views are also more likely to hold inconstant political beliefs.In other words, the strongest partisans and most committed ideologues will also be the most susceptible to policy contradictions because they never look critically at the totality of what they believe.When they think about politics, they are motivated to only "see" that which confirms what they already believe.Third, we test this idea with the case of individuals' beliefs about same-sex marriage and religious freedom.This produces two important findings.First, the most ideologicallyconservative individuals are those most likely to espouse a belief that government should not tell 61 religious organizations who they can and cannot marry, while also believing that same-sex marriage should be banned.Second, this effect of motivated reasoning is more pronounced among Republican-identifying conservatives than Democrat-identifying conservatives.

Political Polarization
Theoretically, in order to have a productive argument, the participants need to have at least a basic agreement on the definitions of the terms being employed and the facts involved.The recent debates about the Affordable Care Act-aka "Obamacare"-is a case in point.Citizens debating that legislation may have very different opinions about the wisdom of the policy choices involved, but decisions to repeal, implement, or amend the Act should be based upon agreement about what it actually says and does.If opposition to the policy is based upon "death panels" that don't exist, or its defense is based upon an insistence that the individual mandate isn't government coercion, the likelihood of reasoned discussion-let alone agreement on policy changes-disappears.
A similar example would be the ongoing battles over religion in the nation's schools.There are genuine arguments to be made about the proper application of the Establishment Clause in the context of public education, but reasoned disputes require people who recognize that the First Amendment's religion clauses require government neutrality in matters of religious exercise.
We certainly agree with those who advocate for the importance of a shared vocabulary and a conceptual common ground to facilitate legitimate and productive political debate and discourse.
A common civic language-an agreement on the basic nature of our shared political reality-is necessary, but it may not be sufficient.An examination of two robust literatures, political science research on political polarization and partisan sorting, and political psychology research on motivated reasoning, strongly suggests that efforts to calm the political waters by supplying accurate information, while necessary, may be inadequate for the task.
A comprehensive review of the literature on political polarization was conducted by Morris P. Fiorina and Samuel J. Abrams in 2008.Fiorina and Abrams surveyed the extant research, testing the "polarization narrative" that began in the early 1990s when Pat Buchanan famously "declared a culture war for the soul of America in his speech at the 1992 Republican convention" (Fiorina and Abrams 2008).They noted the emergence of the "notorious red-blue map" after the 2000 election and the acceptance of the polarization narrative by commentators and pundits (one of whom went so far as to compare Republicans and Democrats to Sunnis and Shias).
Although the authors noted their agreement with the scholarly consensus that elites and Congress had indeed, polarized their review of the then available research, this convinced them that the situation for the public at large reflected partisan sorting rather than polarization.
The political positions of Americans had not become more polarized between the early 1970s and the early 2000s.Importantly, however, within the larger population the parties in the electorate had become more distinct.This change was a product of two other senses of polarization that the DiMaggio group identified: constraint ("the more closely associated different social attitudes become.")and consolidation (" . . . the greater the extent to which social attitudes become correlated with salient individual characteristics or identities.")(DiMaggio et al. 1996, p.693)In the last few decades of the twentieth century, inter-issue correlations were increasing, and partisans were becoming more closely associated with one or the other of the increasingly interconnected clusters" (Fiorina and Abrams 2008).

62
Whether this differentiation is called sorting or polarization, the authors agreed with other observers that it had occurred and appeared to be continuing.And they conceded that other scholars, notably Abramowitz and Saunders (2008) "believe that the process of partisan sorting has proceeded so far that it is accurate to speak of a polarized America." In the wake of Fiorina and Abrams' influential and much-cited review, a number of other researchers have studied the phenomenon, with most agreeing that the partisan divide is increasing, especially among elites and in Congress (see, for example, Krasa and Polborn 2012).
To further complicate the search for common ground and collaborative policymaking, several political psychologists have found that partisans who are often quite well informed will reject "negatively valenced" information if that information is in conflict with their preferred worldview (Redlawsk, Civettini and Emmerson 2010), and still others have concluded that personality traits can predict a "considerable array of human behavioral patterns" (Ha, Kim and Jo 2013), including political preferences and behaviors.Indeed, Alford, Funk and Hibbing (2005), and Fowler, Baker and Dawes ( 2008), among others, have concluded that certain political behaviors and attitudes are genetically influenced and/or heritable.Personality traitsextraversion, agreeableness, conscientiousness, emotional stability and openness to experience, sometimes referred to as the "Big Five," have demonstrable effects upon political behavior (Ha, Kim and Jo 2013).

Polarization, Political Sophistication, and Motivated Reasoning
Motivated reasoning research poses challenges to widely-held beliefs about the way in which individuals search for information.Theoretically, when people are engaged in learning about the world around them, the primary goal is accuracy.The good citizen watches the news and reads newspapers or blogs as part of an effort to gain an accurate understanding of the particular topic under investigation.Unfortunately, as emerging research underscores, this is not how people actually go about gathering information, if they choose to gather it at all.The reality is that people are motivated reasoners (Kunda 1987(Kunda , 1990)).
The concept of motivated reasoning is built on decades of research documenting the biased cognitive processes by which individuals gather and understand new information.It is an unconscious process that occurs through selective perceptions of reality.Once people have developed a worldview--an idea about how something works, or what they like and do not like-they are extremely resistant to information that would require them to change that worldview.
Research confirms that most people do not engage in a wide search for information in order to understand a subject from various perspectives.Instead, they engage in selective exposure (Lodge, Taber, and Galonsky 1999;Mutz and Martin 2001;Sweeney and Gruber 1984), which means that they seek out information that will confirm what they already know (or think they know) and avoid information sources that might challenge their beliefs.While it may be difficult to avoid all contrary information, encountering contradictory facts will not usually require the individual to change or adapt a preexisting framework.When people are faced with a variety of information, some that is confirmatory and some not, they simply ignore or actively argue against the evidence they don't like while uncritically accepting the data seen as confirmatory (Ditto and Lopez 1992;Lavine, Borgida, and Sullivan 2000;Taber and Lodge 2006).Furthermore, when faced with ambiguous information, people do not spend time learning more about the topic; instead, they interpret the ambiguous information so that it is consistent with their current beliefs (Fazio and Williams 1986;Lord, Ross and Lepper 1979;Vidmar and Rokeach 1974).

63
The consequences of motivated reasoning for politics can be quite troubling.This is especially the case when we consider its effects on a political system in which political polarization is increasing, both within elites and the general public (Abramowitz 2010;Bishop 2008;Theriault 2008).As we have seen, the combination of motivated reasoning and increased partisanship leads to more deeply-entrenched beliefs and a corresponding increase in unwillingness to compromise as partisans build self-serving, motivated realities.Barker and Carman's (2012) recent work documents how different the realities are for citizens in Red versus Blue states, and Levendusky (2009) demonstrates how the increased sorting of average citizens into partisan camps has produced more polarized emotional responses to the parties.Politics thus becomes an "us versus them" competition.Taber, Cann, and Kucsova (2009) also find strong support for the polarizing effect of biased information processing (see also Slothuus and de Vreese 2010).
Indeed, the simple act of counting ballots can be affected by motivated reasoning when the counting instructions are vague or ambiguous.In such situations, individuals fill in the gaps in a self-serving, highly-partisan way (Kopko et al. 2011).Finally, Cohen (2003) shows through a series of experiments that "even under conditions of effortful processing, attitudes toward a social policy depended almost exclusively upon the stated positions of one's party."And to top it off, " . . .participants denied having been influenced by their political group, although they believed that other individuals, especially their ideological adversaries, would be so influenced" (p.808).Arguably, this kind of myopic adoption of the positions of one's party is not what the Founding Fathers had in mind.
Education is often proposed as the solution to the problem of political ignorance (Putnam 2000); it is thus reasonable to consider whether it can also solve the problem of motivated reasoning.Unfortunately, rather than moderating partisanship, political knowledge is often connected to an increase in polarization.Education has certainly been shown to be a good predictor of political knowledge (Delli Carpini and Keeter 1996; Price and Zaller 1993), and it is equally demonstrable that politically knowledgeable citizens are those most likely to acquire political information and most able to incorporate it into their existing knowledge framework (Zaller 1992).A recent study by Gillion, Ladd, and Meredith (2013) showed that the gender gap in voting occurred first among the highly educated because they were the first to be aware of elite polarization.Claassen and Highton (2009) find the same dynamic.Polarization among party elites leads those individuals who are the most politically aware to follow suit.In a similar vein, Federico and Hunt (2013) show that individuals who are highly knowledgeable and heavily invested in politics are more likely to approach politics in an ideological fashion, and more likely to exhibit a polarized response to politics (see also Abramowitz 2010;Federico 2007;Judd and Krosnick 1989;Sidanius and Lau 1989;Zaller 2004).
The dilemma posed by what we now know about motivated reasoning is that it occurs no matter how educated or sophisticated the individual.In his examination of partisan sorting, Levendusky (2009) shows that the highly knowledgeable are just as likely to change their ideology to match their partisanship as the politically unsophisticated.Political theorists might hope that the politically sophisticated would privilege policy positions over their devotion to the correct "team," but this is apparently not the case.In other somewhat disheartening research, Hartman and Newmark (2012) examined the motivated reasoning behind the belief that President Obama is a Muslim.It is not a shock to learn that this belief is stronger among Republicans than Democrats, but the fact that political sophistication does not appear to attenuate it is distressing.
To summarize, there is strong evidence indicating that political sophistication, rather than moderating partisan commitments, actually contributes to partisan polarization.We argue that this strengthening of the connection of the most highly engaged to a particular party is then reinforced through selective perception.That is, individuals "see" only that which confirms their existing beliefs.Thus, selective perception does not cause their beliefs but helps support them.
As a result, we theorize that the strongest partisans and most committed ideologues will also be the most susceptible to policy contradictions because they never look critically at the totality of what they believe, but simply see small independent confirmations of one belief or another at a time.
Connecting Political Sophistication to Polarization-Data and Methods In order to examine the effect of political sophistication and motivated reasoning on polarization and holding contradictory policy beliefs, we conducted a national survey measuring partisanship, ideology, and political knowledge.In addition, we asked specific questions about same-sex marriage.The survey was an online survey designed and hosted using Qualtrics online survey software.Survey Sampling International (SSI) was then contracted to provide over 2,300 respondents.SSI maintains national online panels of respondents and recruits participants from across the web using numerous methods, which gives them the ability to reach nearly anyone who uses the Internet.
For our survey, a quota method was used in order to match the survey's demographic and gender distribution to that of the overall United States.Thus, this is not a representative sample of the U.S. population.However, since we were not attempting to estimate characteristics of the U.S. population (for example, the percentage of the population who voted in the 2012 presidential election), this does not affect our results.We are interested in understanding the connection between cognitive processes and political beliefs.We are unaware of any variable that is correlated with participation in any of SSI's panels and these cognitive processes that, if present, would bias these results.Furthermore, we control for numerous demographic and political variables, which further alleviates any concern about bias in the sample.
We produce two models.The first is designed to explain the source of the strength of respondents' party identification.As discussed above, it is not controversial to claim that the most politically sophisticated are also the most partisan members of the electorate.We want to document this relationship in our data for two more compelling reasons.First, it provides yet another piece of research documenting this relationship and, second, because the second analysis focuses on showing preference contradiction, it will demonstrate that those who are the most highly partisan and ideological are the ones who are the most likely to suffer from this failure, and (as shown in this first model) they also happen to be the most politically sophisticated.
The survey contained standard questions designed to identify respondents' partisanship following the method used in each American National Election Studies (ANES) survey. 2 This survey produces a seven-point scale ranging from strong Democrats on the left to strong Republicans on the right.Since we are interested in understanding the causes of polarization, we are more concerned with what drives individuals to the poles of these scales rather than what causes them to choose one side or the other.Therefore, we fold this scale so that it ranges from independent (coded 0), leaning/weak partisan (coded 1), partisan (coded 2), and strong partisan (coded 3), resulting in a scale measuring respondents' strength of partisan attachment (e.g., Dolan and Holbrook 2001).

65
We are interested in capturing the effects of political sophistication, which is not just about political knowledge or education.Therefore, we include measures of political knowledge, education, and, most importantly, their interaction as key predictors of partisan strength.Political knowledge is measured on a nine-point scale and based on respondents' ability to answer eight factual questions (Zaller 1992).These questions are copied directly from ANES and ask respondents to identify (1) Joe Biden, (2) John Boehner, (3) John Roberts, and (4) David Cameron; and, whether they know which party controls the (5) House and the (6) Senate, ( 7) which party is more conservative, and ( 8) what the current unemployment rate is.Education is measured on a seven-point scale that indicates the highest level of educational attainment by the respondent.
Finally, the model also includes several standard control variables: Gender (coded 1 if male, 0 if female), race/ethnicity (1 if white, 0 otherwise), age (coded in years), household income (coded on a 10 point scale), and religious fundamentalism.This last variable is based on a General Social Survey question tapping respondents' beliefs about the Bible as follows: Which of these statements comes closest to describing your feelings about the Bible?
 The Bible is the actual word of God and is to be taken literally, word for word. The Bible is the inspired word of God but not everything in it should be taken literally, word for word. The Bible is an ancient book of fables, legends, history, and moral precepts recorded by men.
This produces a three-point scale ranging from most (coded 0) to least (coded 2) religious fundamentalist.
The second model predicts a contradiction in the policy beliefs among respondents.Our survey asks two questions about religious freedom and same-sex marriage: 1) should same-sex couples be allowed to marry, or do you think they should not be allowed to marry?And 2) Do you think the federal or state governments should make laws regarding who religious organizations can and cannot marry?If one answers "no" to the second question, one should then not answer that same-sex couples should not be allowed to marry because that would be making a law restricting a religious organization's marriage policies.This is an instance of individuals contradicting themselves on policy grounds.However, it may not necessarily be a contradiction for conservative Republicans who hold strong beliefs about smaller government and the legality of same-sex marriage.An individual engaged in motivated reasoning should be expected to make such a contradiction, with the strongest partisans and ideologues being the most susceptible.Thus, the comparison of respondents' answers to these two questions produces our measure of policy contradiction.
The key independent variables in this case are partisanship and ideology.While our general theory does not point specifically to one party or the other, when we operationalize a specific case of policy contraction, there arises the need to identify a specific party and ideological leaning.This is because not all partisanships or ideological leanings respond to specific issues in the same way.In this case, the issue is religious freedom and same-sex marriage.The more conservative and the more Republican one is, the more likely s/he is to answer these questions in a manner consistent with being a conservative Republican rather than in a consistent manner.
The key to motivated reasoning is that partisanship and ideology interact to produce an effect that moves people away from a purely rational thought process by which one either does not accept government intervention in religious choices, and thus accepts religious organization's decision to marry same-sex couples, or one is happy with government intervention in religious choices, and thus is okay with banning same-sex marriage.Simultaneously, disliking government restrictions on religion and liking government banning of same-sex marriage is the product of current partisan and ideological thinking.Those who engage in it are motivated to do so in order to protect their preferred set of beliefs through a form of self-affirmation (Steele 1988).Kahan (2013) finds strong evidence for this.His research shows that individuals are motivated to engage in information processing that reinforces their connection to important ideological groups, in this case a conservative self-identity.Following this line of reasoning, our  -912.18Notes: Strength of Party ID estimated using OLS.(Since this is a short scale running form 0-3, an ordered logit model was also estimated.The substantive results were identical, and therefore, the OLS results are presented here for easy of interpretation.)The Preference Contradiction model is estimated using logit.*= p < .05,** = p < .01,*** = p < .001,two-tailed.
supposition is that this effect should be most pronounced among the most extreme partisans and ideologues.
We measure partisanship as a set of dummy variables: (1) Republican, coded 1 if one identifies as Republican and 0 otherwise; (2) Democrat, coded 1 if one identifies as Democrat and 0 otherwise; and, (3) Independent, coded 2 if one is independent and 0 otherwise.The Democrat variable is then dropped and becomes the comparison group for Republican and Independent when interpreting the results. 3Ideology is measured on a seven-point scale ranging from 0 (strong liberal) to 6 (strong conservative), with a score of 3 indicating independent or nonideological.We then interact ideology with the Republican dummy variable in order to test for the conditional effect of each on the dependent variable.Finally, the model also includes the same control variables as the model predicting partisan strength above.
Table 1 presents results after estimation of the two models.The results of an Ordinary Least Squares (OLS) 4 regression model predicting strength of party identification are presented in the first column, while the results of a logit model predicting the likelihood of holding contradictory beliefs about religious freedom and same-sex marriage are presented in column two.Focusing first on the model predicting partisan strength, one can see that political knowledge, education, and their interaction are all highly statistically significant.This means that the effect of political knowledge on strength of partisanship is conditional on the value of education, and vice versa.
The negative sign indicates that the effect of one variable diminishes as the value of the other goes up.2012) for an explanation as to why one of these variables needs to be dropped. 4Strength of Party ID was estimated using OLS.but since this is a short scale running form 0-3, an ordered logit model was also estimated.The substantive results were identical, and therefore, the OLS results are presented here for easy of interpretation.Interaction effects are notoriously difficult to substantively interpret from an output table, thus, Figure 1 presents a graphical look at how the effect of political knowledge changes as the highest level of educational attainment moves from having a high school diploma, to a college degree, to a graduate degree.Figure 1 makes it clear that the effect of increasing political knowledge is strongest for those having only finished high school.Moving from a political knowledge score of 2 to a score of 7 is associated with an increase in partisan strength of about 3/4 of a point.That is an 18 percent jump in one's partisan leanings.The same type of gain in political knowledge for someone with a graduate degree shows virtually no movement in the strength of their partisanship.Interestingly, at low levels of political knowledge, people with a high school diploma are the least partisan, while at the highest levels of political knowledge, they are the most partisan.Therefore, it would seem that gaining political knowledge without education has a polarizing effect on people.This is a disturbing result for those advocating civic learning outside of formal education.It would appear that this may simply exacerbate polarization.
Returning to Table 1, the logit model estimating policy preference contradiction includes an interaction between ideology and Republican identification.The interaction and ideology variables reach significance, but the Republican variable does not.This is not particularly surprising since the substantive meaning of that variable is the effect of being a Republican when compared to a Democrat when ideology equals zero, which represents strong liberalism.There are likely very few strongly-liberal Republicans, and the data cannot tell the difference between them and Democrats.The significant interaction indicates, once again, that the effect of partisanship is dependent on ideology, and vice versa.Before looking at Figure 2, which helps clarify the interactive effect of ideology and partisanship, it should be noted that political knowledge and education fail to reach significance.Thus, there is no evidence in this data that political sophistication, as measured by either education or political knowledge, helps individuals avoid contradicting themselves.This is what one would expect when the driving force behind policy preferences is not rational calculation, but motivated reasoning to answer questions in a way that is consistent with one's partisan and ideological positions.
Figure 2 presents the predictive margins of the conditional effect of partisanship and ideology on the probability of contradicting oneself.As can be seen, on the left-hand side of the scale (scores 0-2), which corresponds to liberal ideological leanings, there appears to be no discernable difference between Republicans and Democrats.But when one moves to the righthand side (scores 4-6), which corresponds to increasingly conservative leanings, Republicans and Democrats respond much differently.To be sure, both see an increase in the probability of contradicting themselves, but the slope for Republicans is much steeper.For a strong conservative (score of 6), the probability of contradiction is about 23 percent greater if one is a Republican than a Democrat.Indeed, simply moving from a strong liberal to a strong conservative increases the probability of contradiction for Republicans by nearly 60 percent and about 35 percent for Democrats.This is the effect of motivated reasoning and it strongly supports our contention that it is the highly partisan ideologues that are most susceptible to holding contradictory policy beliefs.

Is Civic Literacy Irrelevant?
If motivated reasoning "trumps" accurate information, will our politics inevitably be ideological and polarized?Is the effort to provide credible, evidence-based information a fools' errand?Not necessarily.Redlawsk, Civettini and Emmerson (2010) explored this very question.The authors did not challenge the research on motivated reasoning, but they did test the thesis that a "tipping point" could be reached-that despite the tendency of motivated reasoners to ignore evidence inconsistent with their preferred beliefs (here, a candidate they viewed positively)given enough negative information, attitude change would occur.They find some support for this notion.Although the research is encouraging, it remains to be seen whether a tipping point exists to the extent that ideology concedes to evidence.Learning that a favored candidate is not as admirable as once thought is one thing; accepting evolution or climate change and adjusting one's literalist approach to religion accordingly is quite another.
Finally, if research tells us anything, it is that good information and civic literacy are necessary, but insufficient at ridding us of polarization and a highly selective approach to evidence.The alarming result presented here is that individuals who are supposed to be the best examples of democratic citizens this country has to offer, i.e., the politically sophisticated are also the ones that are the most partisan and the most susceptible to holding contradictory beliefs (at least in the context of same-sex marriage and religious freedom).Unfortunately, it would seem that ignorance isn't the only enemy of reason and political compromise.Predictive Margins with 95% CIs

Figure 1 :
Figure 1: Marginal Effect of Political Knowledge and Education on Strength of Party Identification

Figure 2 :
Figure 2: Marginal Effect of Ideology and Partisanship on the Probability of Holding Contradictory Policy Preferences

Table 1 :
Predicting Strength of Party Identification and Policy Preference