(Earlier versions of this research were presented as papers at the 2015 Southern Political Science Association Conference in New Orleans, LA, January 15–17; the 2016 Southwestern Political Science Association conference in Las Vegas, March 23–26; and the University of Kentucky Department of Political Science Colloquium (November 2015). We thank Kevin McGuire, Kenneth Fernandez, and colloquium participants at Kentucky for providing helpful comments.)

The fundamental role that citizen knowledge plays in American democracy is one of the most important and oft studied topics in the last 70 years. This is likely because conceptualizations of a healthy democracy fundamentally depend on some notion of an informed citizenry. More simply, democracy is only possible if citizens possess enough knowledge to hold government leaders accountable. While the importance of knowledge for voter turnout, vote choice, and public opinion has been well documented (e.g., Delli Carpini and Keeter 1996), legal scholars have only begun to scratch the surface in thinking about what citizen knowledge means for the courts.^[1]

In a rare exception, Gibson and Caldeira (2009b) argue that the public’s knowledge has been systematically underestimated for decades due to challenges in measuring it. They find the question format (multiple choice versus open-ended) matters for how knowledgeable the public appears as well as how strict (or lenient) scholars are when interpreting open-ended answers. At the same time, other research indicates there are different components to political knowledge that may provide alternative roads to democratic-civic competence. Specifically, Prior (2014) finds that some people store political knowledge visually and that visual knowledge can be as important as verbal knowledge. Together, these findings open the door for novel and interesting questions about the role of citizens in a democracy. In this chapter, we ask, What are the implications of political knowledge for the Supreme Court? To address this question, our chapter answers a smaller question. How does visual knowledge influence perceptions of the Supreme Court?

We begin by making an argument for why citizen knowledge matters in the context of the US Supreme Court. It is relatively easy to see why political knowledge is important when it comes to deciding which candidate to vote for. If a person is deciding which candidate will provide better representation on a range of issues, a more knowledgeable person will, all else equal, make a better decision (Lau and Redlawsk 1997). However, it is less obvious why we should care about citizen knowledge as it relates to the judiciary, especially the federal judiciary, since those judges serve appointments for life. In fact, the “third branch” was, at least partly, designed to be less responsive to (or insulated from) public opinion. Yet, we argue, citizen knowledge is important for the Supreme Court for at least two important reasons.

Perhaps the most important reason starts with the simple recognition that the Supreme Court plays a fundamental role in shaping our daily lives. Year after year, the Court hands down decisions on a range of issues, with most of them gaining little publicity. Yet these decisions shape and constrain our behaviors, curtail or expand our rights and liberties, are responsible for significant economic gains and losses, and control the balance of power between the state and federal government. This recognizes the Supreme Court as a key branch in a larger government framework that is responsible for the laws and policies of the country. Knowing and understanding how the system works and knowing the Court’s role and powers within that system are a crucial and necessary first step to understanding how citizens can hold political leaders accountable.

Second, citizen knowledge is important because of its direct links to government processes that directly connect to the judiciary. The most obvious examples are confirmation hearings of federal judges and justices. Confirmations are two-step processes, and both steps are relevant for citizen knowledge and attentiveness. The first step occurs when there is a vacancy on the federal bench and the president nominates someone to fill that position. There are many examples of nominations intersecting with presidents making campaign promises about either the Court or elections where the Supreme Court became a major issue. For example, in 1980, Ronald Reagan made a campaign promise to nominate the first woman to the Supreme Court, and he followed through with his nomination of Sandra Day O’Connor in 1981. In 2020 Joe Biden campaigned on a promise to nominate the first Black woman to the Supreme Court, and he followed through by nominating Ketanji Brown Jackson in 2022. And, of course, the Supreme Court was a centerpiece of the 2016 presidential election. When Justice Antonin Scalia died in February 2016, the Senate did not act on Merrick Garland, Obama’s nominee to replace Scalia. Instead, the Republican-controlled Senate announced that the winner of the 2016 presidential election (Donald Trump or Hillary Clinton) would get to fill the vacancy. After winning, Trump nominated Neil Gorsuch. Knowledge of these events would mean the voter would understand the “raised stakes,” so to speak, that with their vote, they would be selecting more than just a president.

The second step in the process is that the Senate must confirm the nominee. As part of this process, the Senate holds hearings to review the nominee’s record and ask questions of the nominee. These questions range from constitutional interpretation, the nominee’s knowledge, and their views of different areas of the law as well as their past behavior on or off the bench (Farganis and Wedeking 2011; Collins and Ringhand 2013). They may even ask whether the nominee supports vampires or werewolves!^[2] This questioning is important because research shows that public opinion of the nominee influences how senators vote on the nominee (Overby et al. 1992; Kastellec, Lax, and Philips 2010). Thus, having an informed citizenry is essential to following these hearings, making evaluations, and then communicating these opinions to senators. In fact, because confirmation hearings are often seen as a national dialogue on the current state of the Constitution, this is where citizens can play a small role in shaping the Constitution.

Our goal in this chapter is to explore the connection between knowledge, with a focus on visual knowledge, and perceptions of the Supreme Court. The first section starts by measuring knowledge of an important population—college students—and establishes a baseline of their knowledge, ascertaining whether a visual image enhances an individual’s knowledge. We find that receiving a picture alone does not increase knowledge, but we do find that subjects get higher knowledge scores when asked via multiple-choice format.

In the second section, we then wonder whether all visual images are the same. Perhaps our failure to find evidence of a visual knowledge effect was simply due to the pictures we selected. We argue that some visual images may contain clues that help identify the relevant knowledge domain for an individual. We draw on research that shows the importance and power of symbols for citizens when evaluating the Supreme Court (Gibson, Lodge, and Woodson 2014). Using an experiment, we test whether the presence of a legal symbol in a picture (a judge’s robe in this case), compared to a picture with no symbol, increases citizens’ knowledge. We find that it does.

Although these results are important in their own right, we agree with Luskin (2002), who notes that it is important to show how and why knowledge matters. Thus, in the last section we demonstrate why knowledge holds important implications for court scholars. We show that citizens who possess different and contrasting knowledge types (e.g., high in visual knowledge but low in verbal knowledge) structure their perceptions of the then current nine justices in a different manner compared to other knowledge types. We conclude by discussing the implications of citizen knowledge and the courts.

Political and Visual Knowledge

One question about the capacity of citizens to function and thrive in a democracy has concerned scholars for many decades: Do citizens possess the capacity to govern themselves effectively? Answers from early empirical research painted a grim portrait. Citizens seemed to possess little political knowledge and exhibited weak organization of the little knowledge that they did possess, where an opinion in one policy area was not likely to be consistent with opinions in other areas (Berelson, Lazarsfeld, and McPhee 1954; Campbell et al. 1960; Converse 1964). These findings carried enormous normative implications for traditional standards of democracy that emphasized the necessity of high levels of participation and knowledge. Because of these seminal findings, scholars began focusing on finding explanations that would account for the apparent low levels of knowledge in the citizenry. As Luskin (1987) and Delli Carpini and Keeter (1996) note, these research agendas typically focused on one or more of these three areas: ability, opportunity, and motivation. Gradually, scholars shifted their efforts in a slightly new direction to answer the original question and focused on the use of information shortcuts (heuristics) and informational processing (e.g., Lupia 1994; Zaller 1992). Other scholars focused on the measurement of knowledge and measurement error in survey questions and respondents’ answers (e.g., Achen 1975; Erikson 1979; Judd and Milburn 1980; Converse 1980; Judd, Krosnick, and Milburn 1981; Zaller and Feldman 1992).

Understanding the importance of visual political knowledge is relatively new in political science. Visual knowledge is like someone saying, “I can’t remember someone’s name, but I can clearly see their face and…” While this may seem obvious to many people, it was not until recently that political scientists studied how visual knowledge might play a role in civic competence. Prior (2014) argues that traditional-style questions used to measure knowledge rely on a person having a strong verbal link to their cognitions. These questions result in our knowledge measures being heavily favored toward individuals with what he calls strong verbal knowledge. In contrast, Prior argues that some individuals have a more visual cognitive style, where the key linkages to their cognitions are visual.

To test these arguments, Prior performed several experiments and found support for several propositions. First, he found that adding a visual image to an otherwise identical knowledge question increased the likelihood of a correct answer. Second, the size of the effect varied depending on the subgroup. For example, for those in groups that have been shown to have a knowledge gap—women, older people, the less formally educated, and people with a visual cognitive style—a visual image either significantly reduced or eliminated the knowledge gap of the subgroups. Third, Prior demonstrated that visual knowledge was as indicative of civic competence as verbal knowledge. Taken together, these findings suggest that to understand the link between citizen knowledge and the Court, we also need to incorporate visual knowledge. This is important because, as Prior found, some individuals prefer a visual cognitive style.

Exactly why visual images might be particularly helpful in measuring legal knowledge is something we will elaborate upon soon, but the gist is that we believe the presence of legal symbols in pictures will play a role in increasing the likelihood of correctly answering knowledge questions. For now, however, we state our initial hypothesis: Visual images (as opposed to no images) will increase the likelihood of a correct answer.

Research Design and Methods

To begin, we conducted an experiment in June 2014 with 213 undergraduate students as subjects.^[3] We asked subjects four questions in which they were given the name of a prominent political or legal figure, and then we asked them if they could identify the job the person held. These are standard political knowledge questions that are regularly asked on the American National Election Studies surveys and have an “objectively” correct answer.^[4] We asked subjects to identify the jobs, as of June 2014, of Joe Biden (vice president), John Boehner (Speaker of the House), David Cameron (prime minister of the United Kingdom), and John Roberts (chief justice of the United States).^[5] Each of these four individuals was, for several years prior to the survey, in the news often and should be relatively easy for most attentive followers of politics.

For our research design, we experimentally manipulate two factors that we think will impact citizen knowledge. Subjects received (1) a visual image of the person or not and (2) an open-ended answer format or a multiple-choice format that had five possible answer choices. For the multiple-choice questions, four of the answer choices were plausible answers and a fifth category was listed as “none of the above.”^[6] Because we are manipulating two factors and each factor has two options, this is commonly referred to as a 2 × 2 experiment, which has a total of four experimental conditions. Importantly, subjects were randomly assigned for each question. This means that the assignment to each of our conditions was independent across the four questions. That is, a subject could see a picture and be given a multiple-choice question for, say, Biden but then have no picture and an open-ended question for Roberts and so on. Lastly, because we are interested in comparing knowledge levels as a function of the four experimental conditions, this is commonly referred to as a between-subjects design.^[7]

Subjects came from four classes at a large Midwestern university, three of which were political science classes. Of the four classes, two were large, general introductory political science courses; another was an upper-division judicial politics course; and the fourth was an upper-division interdisciplinary course on research methods.

Scoring the answers on the multiple-choice response formats was straightforward: a response was correct if the subject identified the correct job out of those listed. Scoring the responses for the open-ended questions presented a bit more difficulty, as we sought to avoid the problems that have plagued previous efforts identified by Gibson and Caldeira (2009b). To that end, we were generous and awarded credit even if an answer was not completely correct. For example, we scored as correct respondents who identified Boehner as simply “speaker,” but we did not count as correct if someone only said “judge” for John Roberts (“justice,” however, would have earned credit).

Our dependent variable is whether the subject answered each question correctly. Because the dependent variable is binary (0 or 1), the appropriate statistical technique is to use a probit regression. This method allows us to estimate the likelihood (or probability) of whether the subject answered correctly, with the four treatment conditions serving as the main explanatory variables of interest. It also allows us to include statistical controls, which we did for both the specific individual a respondent was asked about and the class from which the respondent was drawn.^[8] More generally, the benefit of our empirical approach is that it allows an easy comparison across conditions. For example, we will be able to tell which of our four conditions produces higher or lower levels of knowledge. Our approach also provides us with an intuitive quantity of interest to measure knowledge—in this case, the probability a person gets a correct answer.

Results

Because our independent variables of interest are mutually exclusive dummy variables, we focus on predicted probabilities to describe our results. (A full table of parameter estimates is in the appendix.) We portray these results graphically in figure 1. The y-axis denotes the probability a respondent correctly identifies the queried individual’s job. The x-axis identifies the four possible combinations of question format and picture presence for each question.

Broadly speaking, these results provide strong support for the importance of question formatting but fail to demonstrate that providing an individual’s picture increases their likelihood of providing a correct response. When a respondent was presented with a name only and asked to provide a textual (i.e., open-ended) response, she had a 0.69 probability of giving a correct one. In addition, another way to evaluate these estimates is to examine their 95% confidence intervals. For example, the 95% confidence interval ranges from 0.63 to 0.76. This generally means that if we were to repeat this study a very large number of times, 95% of the results from this very large number of studies would live between values of 0.63 and 0.76. It is analogous to the /− that you (hopefully) see reported when reading about polls. Hereafter, we will put the 95% confidence intervals in brackets immediately after the point estimate.^[9]

Exchanging the open-ended question for a multiple-choice format increases the likelihood of a correct response to 0.83 (0.78, 0.88)—about a 20% relative increase that is statistically significant. We observe a nearly identical increase when switching between open-ended and multiple-choice questions for individuals who were also presented with the individual’s picture (0.71 [0.65, 0.77] vs. 0.83 [0.78, 0.88]).^[10]

As the figure makes clear, however, when we hold the question type constant and examine the impact of the presence of a picture (i.e., 1 vs. 3 and 2 vs. 4), we see essentially zero difference in terms of a respondent’s likelihood of correctly identifying the individual’s job. For the open-ended condition, we observe a small increase in the likelihood of correctly answering a question when the respondent is presented with a picture (0.71) versus when they are not (0.69), but the difference between the two does not approach conventional levels of significance (p = 0.71, two-tailed test). For the multiple-choice condition, the point estimates are identical (i.e., 0.83).^[11]

In summary, first, our results validate previous work on measuring knowledge that multiple-choice answer formats increase knowledge scores. Second, visual images did not significantly contribute to increased knowledge, and this was the case across all four individuals that we asked about.^[12] This creates an interesting puzzle. If respondents were not helped by images yet we know that some people prefer a visual style, how might visual knowledge be relevant for legal scholars?

Symbols and Visual Knowledge

We argue that domain-relevant cues might be important for enhancing knowledge. It is like having a hint when answering a trivia question. Having a cue that helps a person locate the relevant stored cognitions, we argue, is important in the legal domain because of the unique symbols that populate the legal world. In their research on understanding legitimacy evaluations of the Supreme Court, Gibson and Caldeira (2009a) argue that symbols play a powerful role. Specifically, they argue that whenever people pay attention to the Supreme Court or information about it, they are usually exposed to powerful visual symbols of judicial authority and legitimacy that can increase support for the Court. These symbols include such things as the robe, the gavel, and the cathedral-like court building (see also Gibson, Lodge, and Woodson 2014). This suggests that adding a legal symbol to a visual image should help a subject establish a linkage to the legal knowledge domain and access the relevant stored cognition. Given this, we hypothesize that adding a powerful legal symbol to a visual image will increase the likelihood of a correct answer.

In a preliminary examination of this symbol hypothesis, in the same experiment described above, we randomly assigned subjects to receive either a picture of Justice Clarence Thomas with his robe on (where he is posing for a picture) or a picture of him in professional clothes without his robe (where he is facing the camera with an office background). We note that this appears similar to common pictures for many politically elected officials and that there is even a symbol in its background, the American flag, that is commonly used by elected politicians.

The two pictures we used can be seen at the links listed in the footnote.^[13] What is different about this stimulus is that since all subjects were given a visual image of Clarence Thomas, we did not give them his name and instead asked, “Who is this person, and do you know what job or political office they currently hold?” Answers were scored such that anything close to being accurate was coded as correct. Overall, 50% of the subjects were able to correctly list his job as a Supreme Court justice (or some variant of that), but only 33% were able to correctly give his name (this percentage includes those individuals who only gave a last name or first name).

Starting with Thomas’s name identification, 22% of subjects assigned to the no robes condition provided a correct response. By contrast, fully 45% of those assigned to the with robes condition were correct—a relative difference of more than double. This difference is both statistically and substantively meaningful. Results for his job identification were even more staggering. Among those who saw Thomas without judicial robes, just 28% were able to correctly identify his job. For those who had the added cue of his robes, the percentage jumped to 70%, which is an increase by a factor of 2.5 (and is statistically significant).

While this test is supportive of the idea that domain-relevant visual cues might enhance knowledge, it is only one test, and it featured a relatively well-known justice who, in 2014, was the only Black justice on the Supreme Court. It raises concerns about whether we would find similar results if we were to do the same experiment on the other justices. To that end, we conducted a second experiment in June 2015 with approximately 137 undergraduate students as subjects. Subjects came from two classes at a large Midwestern university. One class was a general introductory political science course that is taken by mostly political science majors, while the other was an upper-division interdisciplinary social science course. We did the same experiment on seven of the then current justices: Samuel Alito, Ruth Bader Ginsburg, Elena Kagan, Anthony Kennedy, Antonin Scalia, Sonia Sotomayor, and Clarence Thomas.^[14] Subjects were randomly assigned to receive a picture of each justice dressed in either judicial robes or professional clothes (like a business suit). When scoring the answers, we again rewarded subjects with a correct answer if they were close, even counting “judge” as a correct answer for job.

Figures 2 and 3 visualize our results for the justice name and job tasks, respectively. Each x-axis identifies the justice that a respondent saw a picture of, and the y-axis shows the proportion of individuals who correctly identified the justice’s name (figure 3) or job (figure 4). Respondents who saw a photograph without robes were represented by gray bars, while respondents who saw a photograph with robes were represented by black bars. The symbols appended to a justice’s name in the figures denote whether the difference between the no robes and with robes conditions was statistically significant. The caret symbol (i.e., ^) corresponds to a significance level of 0.10, and the asterisk (i.e., *) indicates a significance level of 0.05.

Starting with figure 2, among the seven justices we presented photographs of to our respondents and asked them to name, we found a statistically significant robe effect for four: Justices Ginsburg, Kagan, Kennedy, and Sotomayor. This is to say that respondents who saw each of these justices photographed in their robes were significantly more likely to correctly name that justice as compared to respondents who saw them photographed without their robes. The substantive magnitude of these effects is sizable as well. For example, just 7% of those who saw Kennedy in a suit were able to correctly identify him. The analogous value among those who saw Kennedy in his robes was roughly 19%—a sizable increase by a factor of 2.7. We observe similar robe effects for Ginsburg, Kagan, and Sotomayor. We also see evidence of Ginsburg’s “Notorious RBG” status, which started in 2013. Our respondent’s ability to identify her even without her robes matched (or even exceeded) their ability to identify other justices with their robes. Even still, it is also worth noting that fully 60% of respondents could not identify Ginsburg with robes on.^[15]

Figure 3 showcases the results of our job identification question. Perhaps unsurprisingly, we find plentiful evidence that showing a justice in their robes ultimately enhances a respondent’s ability to identify said individual as a Supreme Court justice. Indeed, for all seven justices examined, respondents were significantly more likely to successfully identify a justice as a justice when presented in robes versus not. Unlike the effect of robes on name identification, which was largely consistent across justices, we see some differences across justices in both the magnitude of the effect and how successful respondents were in the with robes condition. In this regard, compare the robe effect for Elena Kagan, where there is more than a sixfold increase in job identification (and a majority who get her job correct), with that of Samuel Alito, who, despite doubling his job recognition, still had less than 30% right.

Viewed together, these results suggest that different images likely matter in terms of evaluating knowledge. Given this, we believe it provides evidence that visual knowledge likely holds implications for how people think about the Supreme Court.

Why Knowledge Matters for the Judiciary: Perceptions of the Justices

In this section, we directly examine the differences between how high- and low-knowledge subjects, among both visual and verbal knowledge types, view Supreme Court justices. In other words, do individuals with high-verbal but low-visual knowledge structure or organize their perceptions differently than individuals who are high on visual but low on verbal knowledge? This is important because of how citizens learn about the Court and make judgments about legitimacy when they receive information about the Court’s outputs. If they receive most of their information in the form of news stories, presuming most news stories contain judicial symbols like a robe, and if symbols are more likely to make a cognitive connection, then understanding how these individuals may organize their perceptions differently is an important task.

To answer this question, we use an analysis technique known as pathfinder associative networks to examine people’s conceptual view of Supreme Court justices. Pathfinder is a technique that sorts data into scales originally developed for the analysis of proximity or distance data in psychology (Schvaneveldt, Durso, and Dearholt 1989). By using a pathfinder network, we provide a different analytical approach to examine whether views of the Supreme Court differ by knowledge type and level. If we find evidence of a difference, then this supports previous research findings (e.g., Gibson 2007; Gibson and Caldeira 2009b) and also adds to the field by providing a novel perspective of how knowledge and perceptions of the Supreme Court intersect.

Pathfinder takes the data and constructs a network. Construction of a pathfinder network begins with the researcher providing a proximity matrix, which generally represents how similar subjects deem a number of pairwise comparisons. For example, Schvaneveldt, Durso, and Dearholt (1989) asked a sample of students to rate how similar they thought 25 natural concepts were to one another on a nine-point scale. These concepts included words such as color, green, animal, and deer. Each student saw all unique pairwise combinations (e.g., color-green, color-animal, color-deer, and animal-deer) and rated how distinct/similar they were on a nine-point scale. As one would anticipate, respondents found more similarity between the color-green pairing than the green-deer pairing. For us, our proximity matrix comes from all pairwise comparisons of the nine justices (that is 36 total comparisons).

Specifically, we asked our subjects in our first experiment to assess, using a five-point scale, the relatedness of each of the 36 unique pairwise combinations of the then current members of the Court. That is, a respondent was asked to compare Alito and Scalia, Alito and Ginsburg, Alito and Breyer, and so on. Pairings were presented to all subjects in the same order, but this order was randomly generated (as opposed to presented in alphabetical order). The ratings given by a particular participant for a particular pair of justices were then used to calculate dissimilarity or distance between pairs of justices.^[16]

One nice attribute of pathfinder is that it takes the pairwise distance data and generates a visual network of the results, where each justice corresponds to a node (e.g., there will be nine nodes), and network links are added based on the patterns of similarities. Since we ask subjects to rate the similarities/differences between justices, pathfinder will take this data and, where it finds “close” relationships between two justices, connect those justices in the network. For the justice pairs that pathfinder finds as “distant,” it will not connect those pairs in the network.^[17]

We are substantively interested in examining whether different types of knowledge (e.g., visual versus verbal) and different levels of knowledge (e.g., high vs. low) have any implications for how citizens perceive the Court. Thus, we need to create four separate networks that correspond to those different knowledge types, which are (1) high visual, high verbal; (2) low visual, low verbal; (3) high visual, low verbal; and (4) low visual, high verbal. To generate these four networks, we simply use the two knowledge scales created above (i.e., verbal and visual) to classify each of our subjects as being either high or low.^[18]

To compare networks, we start with an objective baseline that we can use for understanding perceptions of the Supreme Court, which we call our “gold standard.” We derived this network using data taken from the Supreme Court Database (Spaeth et al. 2022) for the 2010–14 terms that indicated how often justices voted together (i.e., both voted for the majority or both voted for the dissent). To do this, we constructed a nine-by-nine similarity matrix where each entry indicated the proportion for which two pairs of justices voted together. We then converted this matrix into a distance matrix using the same city block metric as above. The pathfinder algorithm then used this distance matrix to create a network representation of the justices. Figure 4 contains the gold standard network derived from justice voting data (4a) as well as the four ratings-derived networks (4b–e), one each for the high- and low-knowledge groups for each knowledge type. Nodes represent the justices on which subjects were asked to provide similarity ratings, and links between justices indicate participants viewed these justices as similar.

Before taking a closer look at each of the four knowledge networks, it should be pointed out that physical location in the networks does not represent a precise distance between two nodes. Instead, how “closely” related two concepts are is determined by the number of links between them. For example, in figure 4b (high-visual, high-verbal knowledge), there are three links between Breyer and Alito but only one link between Breyer and Kennedy. Thus, Breyer and Kennedy are perceived to be more closely related than Breyer and Alito. Furthermore, if we use the same assumptions made in the literature, then we can make inferences about the perceived relative ideological space. For instance, using the same example (4b), we can measure how moderately perceived Alito and Kennedy are relative to each other by counting the number of links away each is from Breyer. Alito is perceived as less moderate than Kennedy because Alito is three links away from Breyer, but Kennedy is one link away.

When we visually scan the networks in figure 4b–e, it appears that the type and amount of knowledge an individual has affects their perception of the Supreme Court justices. Furthermore, by comparing the networks in figure 4b–e to the gold standard network in figure 3a, we can see how far from justices’ actual behaviors these perceptions are. The high-visual, high-verbal knowledge network portrayed in panel (b) has a clear, distinct separation of justices along ideological lines that matches those found in the gold standard network in panel (a). For both (a and b) networks, there is a conservative group containing Roberts, Alito, Thomas, and Scalia and a liberal group with Kagan, Breyer, Ginsburg, and Sotomayor. Both networks also place Kennedy in between the two ideological clusters, consistent with numerous media portrayals of him as the “swing” or “median” justice. The main differences in the networks have to do with the relationships of the justices to one another within their ideological cluster. For example, in the gold standard network (4a), Ginsburg and Breyer are two links apart. However, in the high-visual, high-verbal network (4b), they are more closely associated with only one link between them.

More starkly, we see a complete breakdown of ideological clusters that occurs in the low-verbal, low-visual knowledge network shown in panel (e). This network is characterized by “odd” groupings of nodes that are not found in the gold standard network. By “odd,” we mean that two justices are perceived to be similar (as evidenced by a connecting link) that would not typically be thought of by people with a basic knowledge of the Court’s decisions. For example, Breyer, who marks the start of the liberal cluster in the gold standard network (4a), is linked with two conservatives—Roberts and Alito—in figure 4e, the low-visual, low-verbal knowledge network. Even more puzzling, although the conservative Scalia is linked with the conservative Alito, he is also linked with the liberal-leaning Kagan. Furthermore, Kennedy appears at the end of the network rather than in the middle, where we would typically find the more moderate justices. In short, the knowledge group in figure 4e seems to have a somewhat random organization.

In networks that have some component of high knowledge (e.g., panels [c] and [d]), we see two distinct clusters of justices among ideological lines. However, both networks have some peculiar aspects that suggest important differences from the gold standard network. In the interest of space, we have moved the remaining discussion of figure 4 to the appendix. Suffice it to say, we can conclude that each network represents a different structure of knowledge. Readers may find it difficult, however, to assess by visual inspection alone whether, say, the high-visual, low-verbal network or the low-visual, high-verbal network was more similar to the gold standard network. Without getting into the details here, it is possible to make formal comparisons of the network structures. We did this, and the results are shown and discussed in the appendix. The results support our earlier discussion.

Conclusion

Political knowledge and its direct connection to the functioning of government has been a concern and source of debate for a long time. Moreover, political knowledge is an important concept that is vital for answering many normative and empirical questions. Understanding that there are different types of knowledge and that those different types hold important implications for how we perceive the political and legal world suggests that we have yet to fully understand and flesh out the broad implications of knowledge and politics (see also Barabas et al. 2014).

The main goal of this chapter was to demonstrate that political knowledge has implications for the Supreme Court. To accomplish that, we demonstrated that people with different knowledge types will structure their perceptions of the justices differently. It is clear that more knowledgeable subjects clearly viewed the justices in terms of an ideological relationship, clustering the liberal and conservative justices together, respectively. Others differed in their visual and verbal knowledge.

We believe that this last finding holds important promise for future research in understanding some of the long-standing debates in the legal field about the relationship of legitimacy to the Court and its decisions. Specifically, if some people are more visually oriented and tend to inform themselves through that channel, then when they receive information about a Court decision, that new information may not interact with their cognitive structure in the same way that a person who is more reliant on their phonological memory for doing so. While this is purely speculative at this point, we note it seems to be a fruitful area for future research.