Assessing the Credibility of Constitutional Experts

Abstract

This study investigates how citizens assess the credibility of constitutional experts on matters of government authority. Analyses of data from two similarly designed experiments, conducted with national samples, reveal that partisanship, race, and level of education are significant predictors of survey respondents’ willingness to extend credibility to constitutional experts. The compatibility of the views expressed by experts with respondents’ own policy views on issues that are the subject of proposed government action is also important. Evidence shows that this consistency is more important in the decision that experts are credible than in decisions that they are not credible, suggesting that esteem motives are relevant in the decision to credit experts who express views congenial to our own that are distinct from social-identity motives scholars have theorized to be important in partisan resistance to expertise. The implications of findings for holding government officials accountable to constitutional limits on government authority are considered.

Introduction

We live in a highly specialized society where citizens cannot be expected to have the knowledge or inclination to comprehend the implications of every socially relevant behavior. To deal with this, journalists commonly employ the opinions of experts on matters of public concern (Merkley 2020a). Expert commentary is often used by news media to aid their consumers in understanding complicated causal relationships, historical events, and factors that can be relevant in policy decisions of government officials. Where those who are knowledgeable in some recognized field tend to agree about the effects of some phenomenon of interest, journalists might even refer to expert consensus to alert readers to the weight of evidence regarding a particular claim (Johnston and Ballard 2016; Merkley 2020a; 2020b).

One domain where reporters commonly employ expert commentary is on matters related to the novel assertion of authority by government actors, such as Congress and the President. Conceptions of procedural justice (Tyler 2021) in our constitutional democracy suggest citizens want to be confident that appropriate government actors are working to address matters of public concern and that they are doing so within the bounds of their rightful authority. Questions about which branch or level of government is empowered to act and address specific problems concerning immigration, mask mandates, or the debt ceiling can be challenging, because constitutional provisions regarding institutional powers are unclear. Many rules constraining the behavior of government actors involve esoteric constitutional provisions and opaque bodies of doctrine from past historical eras. Thus, journalists often cite law professors (or even political scientists) as constitutional experts to help citizens understand how legal constraints may relate to the actions of government officials. As such, these authorities can play an important role in our democratic system; they are often the first to signal a problem if office holders are taking actions that exceed the bounds of their legitimate authority.

According to longstanding research on persuasion, cited experts are high-credibility sources that can yield substantial influence in moving people’s opinions toward the evidence and arguments they express (Petty and Cacioppo 1986). Hovland et al. (1953) originally theorized that the power of persuasion for such authorities depends on the extent to which citizens believe their assertions are valid, ¹ coupled with their perceived trustworthiness (Whitehead 1968; Pornpitakpan 2004).

In our current polarized environment, constitutional experts may not be uniformly considered the high credibility sources they once were. Research on the acceptance of scientific findings and expertise indicates growing trends in anti-intellectualism prompted by populism (Nichols 2017) and skepticism about the motives and methods of researchers that produce results contrary to our own worldview (Hornsey and Fielding 2017). Recent studies reveal that journalistic allusions to experts or even expert consensus does not necessarily sway citizens’ thinking in a direction consistent with the views of the expert sources who are cited. This is the case with respect to social-science authorities, such as economists (Johnston and Ballard 2016), as well as hard-science experts with regard to issues such as global warming (e.g., Kahan et al. 2011; Bolsen and Druckman 2018), vaccine effectiveness (Jolley and Douglas 2017), and the safety of nuclear power and genetically modified organisms (Merkley 2020b). ²

In this age of academic skepticism, where political figures on both sides of the political aisle are constantly touting alternative versions of what the constitution forbids and requires, it is not at all clear citizens will see cited legal experts as credible on issues of national authority. Research on affective partisanship shows that citizens think of copartisans as team members, and often demonstrate antipathy toward members of partisan outgroups (Iyengar et al. 2019). Do we see experts on constitutional authority in a similar light depending on whether they express opinions compatible with our predispositions?

Legal interpretation, even at the highest levels, is not hard science. There are no objective truths in the same sense. Although constitutional rules are designed to guide and constrain the behavior of government officials and institutions, they are entirely socially constructed and interpreted by judicial officials with differing worldviews. As Bybee (2010) effectively observes, citizens have an increasingly sophisticated understanding of the role of personal factors in the interpretation of judges. Under such circumstances, what proprietary knowledge do constitutional scholars possess in the minds of citizens? What determines whether individuals find constitutional scholars credible experts regarding questions of government authority?

I delve into these questions using evidence from two similarly constructed experiments administered to nationally representative samples: Timed Shared Experiments in the Social Sciences (TESS) and the Cooperative Congressional Elections Study (CCES). I conduct the inquiry to discover the factors that are important when people credit and discredit cited authorities. I also ask what groups are most likely to express uncertainty about constitutional experts. Exploring assessments in this way helps shed light on different reasons individuals engage in motived reasoning about constitutional expertise based on their agreement with the content of views expressed by experts that is distinct from other factors that have been found to be important in assessments of expertise, such as partisanship and degree of educational attainment.

Thinking about credibility of constitutional experts

There have been many studies looking at individual cognition with respect to expertise in the last decade across several disciplines, including psychology, political science, and sociology. A good proportion of these studies are specifically interested in citizens’ thinking with respect to climate change (Hornesy et al. 2016), but there are also studies that consider how citizens respond to social scientists’ input about policy matters (Johnston and Ballard 2016) and scientific consensus with regard to other policy-relevant issues (e.g., Merkley 2020b; see also Milosh et al. 2020 for example on emerging work on coronavirus disease 2019).

Druckman and McGrath (2019) offer a process-based argument as to why people may resist authority. Specifically, the authors argue that, as people’s ability to directly assess information decreases, the complexity of processing (and, thus, the points for sincere questioning) increases. As such, parsing directionally motivated skepticism from the differing requirements of evidence that could be necessary to sway people’s opinion is difficult. Druckman and McGrath (2019) point out that when evidence is presented by an intermediary, people must first determine whether they find the expert credible before they assess the weight that the expert’s argument and evidence should have in their judgments. Their point is well taken, but perhaps contemplates cognitive processes that are more independent in theory than in practice. There is experimental evidence, for instance, from one of the earliest studies on motivated reasoning (Lord et al. 1979) that people tend to find studies that produce evidence contrary to their opinions about the death penalty as less effectively executed than individuals given identical studies where they are led to believe the results are consistent with their opinions. This suggests that decisions about expert credibility and the soundness of scientific methods may be inexorably tied to whether the proffered evidence is compatible with one’s prior beliefs.

Although there are studies in law and psychology exploring the persuasiveness of expert witnesses in the context of civil and criminal litigation, there are scant studies on the role of legal experts may have in shaping public opinion. One exception is a study by Simon and Scurich (2013), who looked at the influence legal experts can have in shaping individual judgments about the factors that are relevant in judges’ decisions in particular cases. The authors also investigated what shapes people’s confidence in legal experts, and found that legal commentators do not shape individuals’ evaluations of jurists’ decisional processes as much as people’s agreement with the outcome of those decisions themselves. Moreover, people’s attitudes about the competence and reliability of legal commentators are determined by whether experts express a view that is consistent with their preferred outcome. ³ These findings suggest that compatibility with the views legal analysts express can have a substantial influence on citizens’ assessments of expert credibility.

Another factor that has been found to be important in the evaluation of experts is political orientation. Studies have demonstrated that Democrats and liberals tend to defer to expertise more than Republicans and conservatives (Blank and Shaw 2015 [role of conservatism on dampening influence of expertise]; Shen and Gromet 2015; Johnston and Ballard 2016 [using a combined measure of partisanship and ideology]; Suhay and Druckman 2015; Bolsen and Druckman 2018). One might suspect that the rhetoric of former President Donald Trump plays an important role in this observed disparity. Although he may certainly have exaggerated partisan differences in citizens’ reliance on experts, evidence of partisan (and/or ideological) bias in the interpretation of expertise predate the former president becoming a major focal point in American politics (Gauchet 2012; Suhay 2017).

There are also studies demonstrating that, where findings threaten the values of those who identify with the Democratic party, Democrats seem to be less likely to defer to expertise (Kraft, Lodge, Taber 2015; Nesbitt, Cooper, and Garret 2015). These studies exploit a problem with some of the extant research on the acceptance of scientific evidence. Often, the agreement with findings of a particular study (or scientific consensus on a policy matter) is conflated with partisanship. For instance, Republicans who have been found to be more resistant to scientific evidence also tend to be against the specific policy measures involved with reducing global warming or requiring citizens to wear masks. In the current study, by randomly assigning participants to conditions where experts express views consistent (or inconsistent) with their own, as well as controlling for partisan identity in my analyses of credibility assessments, I can distinguish the relative influence of agreement and partisanship on evaluations of constitutional experts. This is important to separate the causal influence of these distinct factors, but also because different goals may be relevant for individuals engaged in motivated reasoning based on their agreement with the views authorities express versus motivated reasoning that is attributable to partisan identity.

The studies by Ross, Lord, and Lepper (1979) and Simon and Scurich (2011; 2013) mentioned previously, suggest that people tend to find research with findings they agree with more credible than research with which they disagree. This sort of motivated reasoning serves self-esteem (Kunda 1990) by helping individuals believe that their personal attitudes are correct or based on sound reasoning. These cognitions can bolster self-perception by helping people feel more intelligent. Concluding that scientists who conduct research that supports our attiitudes are competent, skilled, and effective also serves individuals’ ego by attributing positive traits to others who have expressed similar viewpoints (Brewer 2007).

Other researchers, including Kahan et al. (2011), have suggested that people engage in motivated reasoning about expertise for entirely different reasons: to feel close to relevant social groups in an effort to solidify valued social identities. Although bolstering social identity by attempting to forge close bonds with similar others can also enhance esteem, the mechanism for doing so is more indirect. Hornsey and Fielding (2017) elaborate on this idea with respect to partisanship:

[T]o the extent that people identify with a certain group – and to the extent this group prescribes antiscientific beliefs – internalization of antiscientific views is likely. …In the last three decades… scientific debates have been caught up in political debates. …Republicans sought to discredit the moral integrity of the scientific community. …For a self-identified Republican, the motivation would be to look favorably on their party’s views and to dismiss or dispute contradictory evidence. Given this it is perhaps not surprising that trust in science has declined since the 1970s and that this decline is attributed exclusively to conservatives [citations omitted].

As such, parsing the influences that are a result of agreement with the views authorities express from those that are consequence of partisanship is theoretically important. ⁴

Another factor that could influence individuals’ assessments of experts is their education, particularly where, as in this study, the constitutional experts cited are identified as academics. This may be true for different reasons. First, people who have attained high levels of education might recognize the value of knowledge gained through degree programs (or there may be an element of sunk costs, where they are more likely to value such knowledge because they have gone through the effort of attaining higher levels of education themselves). A second reason education could be important is that scholars have identified anti-intellectualism as a factor that is relevant in the evaluation of expertise (e.g., Merkley 2020b). These twin mechanisms suggest two distinct ways education might matter. If highly educated people value expert knowledge more, they should be more likely to say that experts are credible; alternatively, if those with less education act against elitism, they should be more likely to dismiss experts as not credible. To be sure, this is a fine distinction, we might be able to explore the analyses by looking at factors that influence each type of assessment.

There are several other demographic factors that could be relevant in the evaluation of expertise. People from traditionally privileged groups (i.e., whites and/or men) could be more likely to recognize expert authority than those from nontraditional backgrounds (ex. non-whites, women). This could be the case with constitutional authorities, because certain demographics remain underrepresented in the higher echelons of the legal hierarchy where constitutional experts reside (Kay 1991; Hurwitz and Lanier 2008). Hornsey and Fielding (2017) suggest that this type of acceptance (or rejection) of expertise could be rooted in a kind of system justification motivation. Alternatively, some might expect women to be particularly deferential to expert authority based on traditional gender stereotypes (Anderson et al. 2012). A final characteristic that could be important in the assessment of expertise is age. Younger individuals may be more likely to defer to experts than older ones because of their relative lack of knowledge and experience.

Design

Data for this inquiry come from two similarly designed experiments looking at citizen assessments of national legislative and unilateral executive authority. Both experiments involve a hypothetical article about proposed government action. Participants were not given a source for the article, but each scenario began with a date and “AP” to simulate the style used by the Associated Press news service. The TESS study of congressional authority took place in September 2016, and the experiments involving executive authority were included on the 2017 CCES. The Appendix includes treatments and sample characteristics for participants in each study. ⁵

Both experiments were originally deigned to answer a somewhat different question about how rules and political context interact in the minds of citizens to shape views of the appropriate use of government authority (Braman 2021). Each experiment was conducted to discover how participants’ policy views and their feelings about relevant political actors combined with the majority support for the proposed action and the expert commentary about compliance with rules to influence assessments of the appropriateness of state action.

Experimental scenarios employed a consensus of experts about the exercise of national authority to address a matter of public concern. Right after participants were exposed to treatments where experts expressed a view that was divided, consistent, or inconsistent with their expressed policy preferences, survey respondents were asked whether they thought such experts were credible regarding the question of government authority. As detailed below, there is some very interesting variance to explain in how study participants answered this question. Moreover, participants were randomly assigned to conditions that can be slightly retooled to fit this inquiry. These answers about expert credibility are appropriate for exploration as a dependent variable to shed light on the question of how citizens think about constitutional authorities.

The experiment on assessments of legislative authority involved a 2 × 3 × 2 design where the issue that was the subject of congressional action, expert consensus, and the level of public support for proposed measures were manipulated. Participants in the TESS administration read one of two articles that stated Congress was considering an action that would either impose limits on the shipment of guns in interstate commerce or cut off federal funds to states that allowed undocumented students to receive in-state tuition benefits. ⁶ Prior to reading the article about congressional authority, to gage participants’ policy views on the issues that were the subject of the scenarios, survey respondents were asked questions related to their support for limiting gun access and the restriction of benefits to U.S. citizens and those in the country legally.

Each article stated that national legislative action could interfere with state prerogatives in our federal system of government, specifically pitting congressional authority against states’ rights. What participants were told about expert consensus concerning congressional authority to take each action was manipulated in the experiment. One third of participants were told that the consensus among constitutional experts was that Congress was acting within its authority under Article I, one third were told that the expert opinion on the issue was divided, and one third were told that legal experts agreed that the action would infringe on states’ reserved powers in our federal system. This consensus was exemplified with a quote from an authority identified as a constitutional law professor at George Washington University Law School. ⁷ The level of public support for such measures was also manipulated in the article. Half of participants read an article stating that 85% of the population supported such measures, and the other half were told only 15% of the population supported the proposed action. ⁸

After reading the hypothetical article referencing expert consensus, participants were asked whether they thought constitutional law scholars were credible on the issue of government authority related the proposed measure, and whether they thought the action was an appropriate exercise of legislative authority. The question about expert credibility is the dependent variable analyzed in this inquiry. Respondents were given three options to indicate whether they thought experts were credible: Yes, no, and do not know.

The design of an experiment on executive authority deployed on the CCES in the autumn of 2017 was purposefully quite similar. That administration involved a 3 × 2 design where participants read an article stating that President Trump was considering issuing an executive order that would take away federal funding from sanctuary cities that declined to report nondocumented citizens to federal authorities. Before participants read the article, they were asked about their support for sanctuary cities. Again, one third of participants read an article stating the consensus among experts was that the President had authority to do so, one third were told experts agreed he did not have such authority, and one third were told expert opinion was divided on the question of authority. The quote exemplifying consensus was attributed to a professor who was “a noted authority on government powers.” Once more, half of participants were told 85% percent of the population supported such measures and the other half that only 15% of the population supported the action. As in the study of legislative authority, after reading the article, participants were asked whether they thought constitutional experts were credible on this matter of government authority using the same three response options.

For the purposes of this study, I am interested in how individuals respond to experts who express views that are either consistent or inconsistent with their policy views on proposed government action controlling for other relevant demographic variables. Participants in both administrations were randomly assigned to one of three expert consensus conditions (clear authority consensus, divided consensus, and clearly no authority consensus). I combine these categories with the variable reflecting whether survey respondents agreed or disagreed with the policy implications of the proposed measures described in the articles about gun control, in-state tuition benefits, and sanctuary cities. This results in three new categories across administrations reflecting participants who agreed with the consensus opinion cited in the article for the condition to which they were assigned (n = 267 TESS; n = 156 CESS), participants where whether they agreed or disagreed was unclear because they were in the divided consensus treatment (n = 254 TESS; n = 164 CCES) and, participants who disagreed with the consensus opinion expressed in the article (n = 273 TESS; 156 CCES).

I use dichotomous versions of assignment to each of these agreement conditions to see how participants in each group respond in relation to one another in the analyses presented. This should tell us if there are similar factors at play in deciding if an expert is credible or not. Moreover, a high proportion of respondents expressed uncertainty about constitutional experts; analyzing these responses will help us understand whether there are specific factors that contribute to that assessment.

Results

Table 1 sets forth the distribution of responses to the credibility question across administrations. In each sample, responses were quite similar. A plurality of respondents indicated that they agreed the cited expert was credible. Approximately one fifth of respondents said the expert was not credible, and approximately a third of respondents said they were not sure. Thus, here, as in previous studies (Blank and Shaw 2015), more people deemed the experts cited in the article credible than not credible. Appendix Tables A1 and A2 demonstrate this is also true across partisanship and agreement conditions in both samples.

Table 1. Expert Credibility Ratings Across Experimental Administrations

It is also worth noting that the proportion of respondents who express uncertainty about the credibility of experts is not negligible comprising more than a third of all participants in both samples. This suggests that individuals who find experts credible are not just a mirror image of those who say legal experts are not credible. Different factors may be relevant in each assessment. Moreover, there might be systemic factors that explain who is likely to express uncertainty about expertise. To explore this, I run probit models on each of these distinct responses (credible, not credible, and do not know) in addition to an ordered probit model on the three-point response variable reflecting growing skepticism about expertise.

As I conduct the multivariate analyses of judgments about expertise in several different ways, each model answers a different question about the cognitive processes involved in the judgments of the credibility of constitutional authorities. I aim to answer three distinct questions in the analyses: what variables explain skepticism about constitutional experts; are the same factors relevant in crediting experts and dismissing their views, and who is most likely to express uncertainty about constitutional expertise?

Question 1: What variables explain skepticism about constitutional experts?

First, I employ an ordered probit model to see what leads to skepticism about expertise. The dependent variable in the analyses was the three-point response to whether the constitutional authorities are credible experts regarding the question of government authority described in the article. Responses were coded to reflect increasing skepticism: 1 for yes, 2 for do not know, and 3 for no. I employ each agreement category as a dichotomous variable to see how individuals in different categories respond to questions about expertise in relation to one another. Participants who agreed with the expert consensus in the condition to which they were assigned are the excluded category in these models. ⁹

I also include measures of participant partisanship ¹⁰ (coded 1-7, with higher numbers representing strong Republican identity), race (1 = white, 0 = nonwhite), sex (1 = female, 0 = male), birth year (higher numbers reflect participants who are younger), and level of education (higher numbers represent higher levels of education). The public support manipulation in each experiment was coded dichotomously (0 for 85% support; 1 for 15%). In the models for legislative authority, I also included a control for whether the respondent was answering questions about gun control measures or limits on in-state tuition benefits for undocumented students (coded as 0 and 1, respectively). The results for the ordered model regarding skepticism are shown in Table 2.

Table 2. Ordered Probit Skepticism Across Administrations

Note: +p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001.

The results demonstrate that participants who are in conditions where they disagree with the expert consensus are significantly more skeptical of experts than the excluded category of those who agree with the consensus across administrations. Marginal effects indicate that differences are substantively important. In the executive authority administration, for instance, participants in conditions where they disagree with the expert opinion regarding sanctuary cities are 13% less likely to say experts are credible and 9% more likely they are not credible than participants in conditions where they agree with experts. Those in the divided consensus category also tend to be more skeptical of experts than those who agree. This difference is significant in the administration regarding executive but not legislative authority.

The models also reveal that partisanship is significant in the assessment of experts, with Republicans judging experts with more skepticism than Democrats. In the executive authority scenario, strong Republicans are 3% less likely to deem experts credible and 2% more likely to say they are not credible than Republicans who are not strong identifiers. Moreover, white participants are 11% more likely to say experts are credible and 7% less likely they are not credible than nonwhite participants. Education also plays a significant role in judgments, with participants who are more educated expressing less skepticism about expertise. There is also evidence that younger individuals defer more to experts (in the CCES sample, the birth year variable is marginally significant), and participants who evaluated gun control treatment were more skeptical about experts than those who read about the restriction of in-state benefits in the study about legislative authority.

Question 2: Are the same factors relevant in crediting expert opinion and dismissing their views?

A second question one may ask in the context of this inquiry is whether the same factors are relevant in deeming experts credible as in dismissing experts as not credible. To probe this question, I run identical probit models on each of these response categories in Tables 3 and 4. ¹¹ In Table 3, the dependent variable was coded 1 if the respondent deemed the expert credible and 0 otherwise. In Table 4, the dependent variable was 1 if the respondent said the expert was not credible, and 0 otherwise.

Table 3. Probit Credible Response Across Administrations

Note: +p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001.

Table 4. Probit Not Credible Responses Across Administrations

Note: +p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001.

Looking across the results for the two responses, similar demographic factors appear to come into play in both judgments. Republicans are significantly less likely than Democrats to say experts are credible, and also significantly more likely to say they are not credible. Those who are educated were more likely to say experts are credible and less likely to dismiss them as not credible. Race is also important; it is significant for both responses in the TESS administration and marginally significant in the CCES models. White participants tended to be more trusting of experts than nonwhite participants, and less likely to dismiss their views.

One particularly interesting finding looking across these models is that participants’ agreement with the expert consensus in the condition to which they were assigned is more important in judging experts to be credible than in saying they are not credible. ¹² This particular result is interesting and consistent with findings about self-esteem, suggesting that attributing positive attributes to people who are like you is more satisfying than attributing negative attributes to those who are not (Brewer 2007). Viewed in this light, attributing credibility to those who share our views makes us feel better about ourselves, but dismissing the opinions of those who disagree does not serve the same purpose as effectively (see, Clark and Evans 2014 on the role of consistent high-credibility messages in self validation).

To get a better idea of the substantive importance of variables in the analyses, Table 5 sets forth the predicted probabilities for credible and not-credible responses in the executive authority scenario. The results demonstrate that participants who agreed with experts act distinctly from those who disagreed with the expert consensus and participants in unclear conditions, and the predicted probabilities for the latter two categories were quite similar. We observe that participants in conditions where they agreed with experts had a 0.56 probability of saying they are credible and 0.13 probability of saying they are not credible, holding demographic variables at their means, compared with 0.43 and 0.21, respectively, for those in conditions where they disagreed with the opinions expressed by constitutional authorities.

Table 5. Predicted Probabilities for Executive Authority Scenario

Note: Predicted probabilities are with demographic variables at their means and condition variables set to agree (disagree = 0, unclear = 0), disagree (disagree = 1, unclear = 0), or unclear (disagree = 0, unclear = 1), as indicated.

When considering demographic factors, partisanship had a substantial effect on assessments of expertise. Strong Democrats who agreed with experts had a 0.64 probability of deeming experts credible versus 0.45 for strong Republicans, holding other demographic variables at their means. Thus, predictions indicate strong Republicans are approximately 20% less likely to say experts are credible and more than twice as likely to say they are not credible (0.20 vs. 0.08) than strong Democrats, even when they agreed with their views. White paticipants who agreed with expert views had a 0.60 probability of agreeing that experts are credible compared with 0.50 for nonwhite participants.

Education also has a powerful influence on assessments of constitutional expertise. Participants with postgraduate education had the highest probability (0.77) of finding experts credible when they agreed with the views that authorities were expressing. Moreover, even when experts were expressing views that highly educated individuals did not agree with, they were still substantially more likely than not (0.66) to deem them credible. Those without a high-school diploma had the lowest likelihood of deciding experts were credible, particularly when they disagreed with the views the experts were expressing (0.21). This group was also most likely to respond that experts were not credible when they disagreed with the views the authorities expressed (0.38).

Question 3: Who is most likely to express uncertainty about expertise?

Table 6 displays the probit models for do-not-know responses across administrations to explore if there were systematic factors that explained the expression of uncertainty about expertise. Responses were coded 1 for participants who said they did not know if experts were credible, and 0 otherwise.

Table 6. Probit “Do Not Know” Response Across Administrations

Note: +p < 0.10, *p < 0.05, **p < 0.01, ***p < 0.001.

The results demonstrate that, although those who disagreed with the experts were somewhat more likely to express uncertainty about expertise, there were no significant differences in terms of respondents’ agreement with the consensus conditions they were in. Moreover, those who identified as Democrats and Republicans seemed to respond with do not know similarly. However, we do observe that a pair of demographic features are relevant. Women in both samples are more likely to respond with uncertainty about the credibility of experts, as are those who are less educated. Women who agreed with experts have a 0.35 probability of responding that they did not know if the experts were credible in the executive authority scenario compared with 0.23 for men. This is interesting because women are also less likely than men to give definitive answers, responding that experts are either credible or not credible, although the differences were not significant. Taken together, this perhaps demonstrates that men tend to be more decisive in their thinking about the credibility of experts.

Participants who did not have a high school diploma and agreed with the views experts expressed are approximately twice as likely to respond that they did not know if experts were credible (0.41) compared with participants who had graduate education (0.21). Thus, the analyses demonstrate that participants who were less educated were both more skeptical and more likely to express uncertainty about the expertise of constitutional authorities.

Conclusions

This study examined how citizens assess the credibility of constitutional experts on matters of government authority. Findings were strikingly consistent across experiments conducted on 2016 TESS and 2017 CCES samples, using scenarios involving two governmental institutions and three different issues. This pattern of results contributes to the overall confidence in findings about the factors and processes relevant in the evaluation of constitutional authorities from this inquiry. The analyses showed that demographic factors, including partisanship, race, and level of education, were significant predictors of survey respondents’ willingness to extend credibility to constitutional experts. Republicans, nonwhite participants, and those who were less educated all expressed more skepticism about expertise. The compatibility of the views expressed by experts with respondents’ own policy views on the issues that were the subject of proposed government action was also important.

Looking at relevant factors in each response revealed that agreement with the views expressed by experts was more important in the decision that experts were credible than in decisions they were not credible. This comports with research demonstrating that such agreement serves our self-esteem more effectively by attributing positive characteristics to those who are like us than giving negative attributes to those who are not. This may be especially true when evaluating experts. By deeming experts who share our views credible, those views themselves take on greater validity. Deciding that experts who do not share our views are not credible does not have that same self-validating effect. This finding helps distinguish the purpose of motivated reasoning that is due to agreement with the views expressed by experts from social-identity benefits of motivated reasoning that have been theorized to be relevant in motivated judgements about experts driven by partisanship.

Although partisanship and agreement with the views expressed by experts both had considerable effects on assessments of expertise, differences between participants with the highest and lowest levels of education in our samples were the most pronounced. The reason education has such a substantial influence on such judgments could be because there is pressure from above and below, so to speak. Highly educated individuals are more likely to credit experts because of their own experience in academic environments, and those who are less educated are more likely to say they are not credible because of a growing degree of anti-intellectualism identified by scholars such Merkley (2020b). Evidence here supports the notion that both mechanisms are at play. Moreover, the analysis of do-not-know responses suggest a third dynamic, where those who are less educated are not only more skeptical, but also more likely to express uncertainty about expertise than those who have achieved higher levels of education.

Expressing uncertainty about expertise did not seem tied to any passive-aggressive disagreement with the views authorities were expressing. There was no significant pattern between agreement conditions and the likelihood of giving an uncertain response about expert credibility. We observed that women were significantly more likely to express uncertainty about the credibility of constitutional experts than men rather than defer to experts, as gender stereotypes might suggest, or question their credibility, similar to other groups (nonwhite participants) that have been traditionally underrepresented in legal academia where the constitutional experts that are cited by journalists commonly reside.

Considering the implications of these findings for the constraining power of constitutional rules in our current political environment is important. One may question how findings about the credibility of constitutional experts influence the persuasive weight of arguments and the evidence they provide. The findings here suggest the relationship is somewhat complex. Although the distribution of responses regarding expert credibility across the TESS and CCES samples was quite similar, with more than 50% of participants saying they did not think (or did not know if) the cited experts were credible, the analyses of participants’ ultimate judgments about the legitimacy of the government action indicated that expert consensus significantly influenced participants’ assessments in the legislative context, although its influence was not significant when evaluating unilateral executive authority (Braman 2021). ¹³ However, there was evidence in the assessments of the legitimacy of both congressional and executive authority that participants were heeding experts in some respect, because legitimacy ratings were the highest when constitutional experts agreed that Congress and the President had the authority to take the proposed action, and the lowest when the experts agreed they did not.

These results suggest a bit of a disconnect between participants’ expressed assessments of expert credibility and the persuasive influence of the views that those authorities profess. In their review of research on resistance to scientific authorities, Hornsey and Fielding (2017) suggest that such resistance could serve an independent value expressive function for individuals. The various reasons we have to assess constitutional authorities in a biased manner may slant our assessments, such that they do not accurately reflect the influence of expert information in our own judgments. The suggestion that this may cause some people to misjudge the influence of expert information in their own decision-making is an intriguing idea that seems worthy of further investigation.

Finally, the unique context in which the opinions of such authorities are relevant should be acknowledged. Citizens are not judges, but widespread acquiescence to official assertions of authority that may be of questionable constitutional validity can certainly add to the appearance of appropriateness. Under such circumstances, public opinion can have real implications for government authority. We rely on the media to inform us about official acts we cannot observe directly. The media, in turn, relies on constitutional experts to alert citizens about dubious assertions of power on the part of government actors.

Findings here suggest that our willingness to credit experts so that we may heed such warnings depends in no small part on our opinions, our partisanship, and the extent to which we share the academic training of authorities that are cited. Ultimately, if we want constitutional rules to constrain government institutions in the manner they were designed, there must be a way for individuals to recognize, even socially constructed, limits on state officials that are not so closely tied to our backgrounds and predispositions. In the absence of such limits, the only thing keeping us in check in the current political environment may be the fact that we are so evenly divided. If Congress or the President takes some questionable action, as many people are likely to see the action as legitimate as illegitimate. Taken to its extreme, official authority that depends so critically on political context may not have any real limits at all.