In a tradition dating back to the American founding, many Americans use religious language to describe their esteem for the Constitution. When the constitutional convention finished its work in 1787, Dr. Benjamin Rush (as cited in Levinson Reference Levinson1988, 13) described the resulting document: “[it] is as much the work of a Divine Providence as any of the miracles recorded in the Old and New Testaments were the effects of a divine power.” George Washington, in his Farewell Address (Reference Washington2008 [1796]), implored Congress to “sacredly maintain” the Constitution. In the Federalist Papers, James Madison hoped “veneration” of the Constitution would promote governmental stability (Publius 2008b [1788]). Thomas Jefferson, who like Madison was not a particularly orthodox Christian, nonetheless used religious rhetoric to mock Madison's desire for constitutional durability. He observed (as cited in Levinson Reference Levinson1988, 9): “Some men look at constitutions with sanctimonious reverence and deem them like the ark of the covenant, too sacred to be touched.”
Because so many Americans put their faith in the Constitution as a part of a national civil religion, hermeneutical debates have significant political consequences. In today's cultural politics, controversies often become constitutionalized, both in courtrooms and popular discourse. For example, after Justice Antonin Scalia died in early 2016, Baptist Press, the news agency for the Southern Baptist Convention, published an article on the occasion of his death linking his “original meaning” judicial interpretation with the proper interpretation of the Bible. The article opened by stating that critics of Scalia's jurisprudence “tend to employ a method of interpreting written documents that not only undermines American democracy, but historic Christianity as well” (Roach Reference Roach2016). The message from various leaders within the evangelical denomination was unwavering: interpreting the Bible is directly relevant to interpreting the Constitution.
As this paper suggests, religious leaders—like those in the Southern Baptist Convention—hope to teach their religious rank-and-file how to interpret both the Bible and the Constitution. The ability of the mass public to link religious interpretation and constitutional interpretation seems tenuous, especially since constitutional knowledge is quite low (Cillizza Reference Cillizza2017). Nonetheless, some progressives in the legal academy have argued for a new wave of popular constitutionalism to counteract what they see as a growing conservative judiciary. On the right, the Tea Party championed popular constitutionalism as a way to return America to its founding roots of freedom and limited government.
If citizens do engage in constitutional interpretation, we suggest that they are likely to learn a hermeneutical approach from their social contexts, and religious congregations might be most conducive for learning these skills. Though religious participation has declined somewhat in recent years, religious institutions remain the most widespread social organizations in America (Putnam and Campbell Reference Putnam and Campbell2010; Wald and Calhoun-Brown Reference Wald and Calhoun-Brown2011). In this paper, we bring together three datasets to test whether ordinary Americans interpret the Constitution using methods similar to how they interpret religious texts. We first use cross-sectional data from two national surveys from 2014 and 2018 to understand the relationship between religion and constitutional interpretive styles. We then use a survey experiment to investigate whether hermeneutical approaches are transferable from religious contexts to political ones. Our findings suggest that political and hermeneutical factors inform popular approaches to constitutional interpretation.
CONSTITUTIONAL FAITH AND HERMENEUTICS IN THE ELECTORATE
As Kramer (Reference Kramer2004) has noted, from the founding through the early twentieth century, the people often played a more prominent role in constitutional interpretation—often over and against the judiciary's supreme role. In Federalist 46, Madison argues that “ultimate authority … resides in the people alone,” and they will and, by implication, must protect against the federal government misusing its power (Publius 2008a [1788]; 2008b[1788]). Abraham Lincoln also forcefully expressed popular constitutionalist sentiments in his First Inaugural, castigating the Taney Court and the Dred Scott decision. While popular constitutionalism was more common in the nineteenth century, for most of the twentieth century, however, there was little emphasis on it. Judicial supremacy became the means by which constitutional disputes were resolved, and the people's understanding of the Constitution became largely irrelevant. With the Court firmly in charge, the Warren Court's willingness to expand constitutional rights to meet changing times gradually gave way to an originalist renaissance.
Popular constitutionalism is currently undergoing a renaissance on both sides of the aisle. Kramer and other legal academics revitalized the idea constitutionalism in the early twenty-first century after the Court had become more conservative (Tushnet Reference Tushnet2000; Reference Tushnet2006; Kramer Reference Kramer2004; see also Levinson Reference Levinson1988). If constitutionalism were more democratic, these scholars reasoned, popular liberal ideals would triumph (see Schmidt Reference Schmidt2010; related, see Keck Reference Keck2010). Popular constitutionalism has also taken hold in the Tea Party movement, which advocates that the people have the independent authority to interpret the Constitution more strictly. Yet, the ideology of the Tea Party's advocacy is in direct contrast to many of the supporters of popular constitutionalism in the academy.
In order for popular constitutionalism to work, the public needs to understand constitutional concepts, and recent scholarship suggests constitutional knowledge presents a substantial barrier (Caldeira Reference Caldeira, Gates and Johnson1991; Mondak Reference Mondak1991;Reference Mondak1992). That said, one reason why Americans might be up to the task of making constitutional decisions is that the most basic debates in constitutionalism mirror similar disagreements within Christianity. In his book, Constitutional Faith, Sanford Levinson (Reference Levinson1988) explores how the Constitution integrates Americans into a sense of nationhood. To analyze American civil religion, he draws on the fundamental theological distinctions between the Catholic and Protestant traditions of Christianity. Levinson develops his Catholic/Protestant dichotomy and applies it to two basic questions in constitutional theory: what the Constitution is and who has authority to determine its meaning.
Constitutional Catholics embrace all the constitutive elements of the American political order, written and unwritten, which is analogous to the teaching of the First Vatican Council of 1870. For example, Walter Murphy (Reference Murphy1991, 113) draws upon Aristotle and extends the concept of a constitution beyond the text to include “the state's most basic ordering.” A Catholic conception of the Constitution would include “super-statutes,” such as the 1964 Civil Rights Act, because they serve constitutional functions, despite their status as ordinary law (Eskridge and Ferejohn Reference Eskridge and Ferejohn2001). Other features of the American constitutional order include important judicial decisions; informal institutions, like the two-party system; and extra-textual constitutional principles, such as judicial independence.
On the other hand, constitutional Protestants look to the text of the Constitution as their sola scriptura. As Levinson (Reference Levinson1988, 30) describes this viewpoint, “The Constitution is a discrete document that can be taken off the shelf, like the Bible, and perused. Its writtenness supplies a tangibility, an ability quite literally to take the Constitution into one's own hands and read it with one's own eye.” Levinson casts Justice Hugo Black, a Primitive Baptist, as his archetypical constitutional Protestant on this dimension. Justice Black's judicial philosophy, which has been described as an “absolutist” (Hockett Reference Hockett1996, 15) or “liberal strict constructionist” (Rosen Reference Rosen1994) approach, stems from his zealous commitment to the text and the intent of its authors. Justice Black always carried a copy of the Constitution in his jacket pocket because, as he put it (as cited in Levinson Reference Levinson1988, 32), “When I say something about it, I want to quote it precisely.”
On Levinson's second dimension, constitutional Catholics view the U.S. Supreme Court as supreme in a manner that mirrors papal infallibility. The justices actively cultivate this authority in numerous ways. The current home of the U.S. Supreme Court is often described as “the Marble Palace,” and the justices sit high above the audience in the imposing courtroom in which they hear cases. The motivations underlining these architectural choices are similar to those in houses of worship. The justices also assert their infallibility in some of their most famous decisions. When resistance to integration in Little Rock, Arkansas threatened to undermine the Court's position as articulated in Brown v. Board of Education, the Court issued a stirring rebuke in Cooper v. Aaron (1958, 18). Drawing upon language establishing judicial review in Marbury v. Madison, the Court said, “It is emphatically the province and duty of the judicial department to say what the law is. This decision declared the basic principle that the federal judiciary is supreme in the exposition of the law of the Constitution…”
The Protestant alternative to judicial supremacy is not anarchy or even civil disobedience. Instead, the legislative and executive branches can incorporate constitutional reasoning into their decisions, as when President Jackson famously vetoed the renewal of the Second Bank of the United States. President Lincoln viewed the Dred Scott decision as binding on the litigants involved but refused to recognize the precedential value of the case. According to Lincoln, if our policies are “irrevocably fixed by decisions of the supreme court…the people will have ceased to be their own rulers, having to that extent practically resigned their government into the hands of that eminent tribunal” (Lincoln Reference Lincoln2008 [1861]). Even Justice Robert Jackson (Brown v. Allen 1953, 540), the last member of the Supreme Court not to attend law school, humbly acknowledged the distinction between judicial and papal infallibility: “We are not final because we are infallible, but we are infallible only because we are final.”
While Levinson's taxonomy of constitutional faith seems insightful, a common criticism of this book is that the theoretical developments raise more questions than answers and that “more determinacy” is required (Amar Reference Amar1989, 1153). In particular, if these connections between theology and constitutional interpretation are to have any practical import, there must be a process that leads people to view their governing document similarly to their sacred text. In the following section, we outline a potential causal mechanism.
LEARNING TO DO CONSTITUTIONALISM
Based on prior scholarship, we theorize that there are primarily two potential paths to learning to interpret the Constitution: politics and social context. There is a grand literature regarding how individuals form political opinions. Some of the frequently cited mechanisms are party identification (Campbell et al. Reference Campbell, Converse, Miller and Stokes1960; Green, Palmquist, and Schickler Reference Green, Palmquist and Schickler2002; Carsey and Layman Reference Carsey and Layman2006; Goren, Federico, and Kittilson Reference Goren, Federico and Caul Kittlison2009; Lenz Reference Lenz2012), individual biases (Calvert Reference Calvert1985), observed history (Sobel Reference Sobel1985; Sniderman, Brody, and Tetlock Reference Sniderman, Brody and Tetlock1991), elite cues (Zaller Reference Zaller1992; Bullock Reference Bullock2011), informational shortcuts (Carmines and Kuklinski Reference Carmines, Kuklinski, Ferejohn and Kuklinski1990; Popkin Reference Popkin1994), and persuasion (Lupia and McCubbins Reference Lupia and McCubbins1998), and often these forces overlap.
There is scant theorizing on how individuals choose to interpret the Constitution, but it is likely that they gather information about the Constitution from trusted sources, including their political party, interest groups, elites, and persuasive confidants. As such, the political model suggests that individuals, to the extent that they are knowledgeable about the Constitution, will interpret it consistently with their partisan identity. From this perspective, the people are merely legal realists, engaging in motivated reasoning (see, e.g., Kunda Reference Kunda1990; Taber and Lodge Reference Taber and Lodge2006; Leeper and Slothuus Reference Leeper and Slothuus2014). Constitutionalism is politics by another name.
Beyond being exclusively political, many scholars also suggest that learning about politics has much to do with one's social context (see, e.g., Brooks and Prysby Reference Brooks and Prysby1991; Huckfeldt and Sprague Reference Huckfeldt and Sprague1995; Baybeck and McClurg Reference Baybeck and McClurg2005). Within these contexts, people engage with issues, they interact with those who agree and disagree, and they meet persuasive individuals. This socialization produces political learning. The social context may be particularly important if people lack a strong ideology or issue awareness, as other stories have found (see, e.g., Converse Reference Converse and Apter1964; Zaller Reference Zaller1992). While it seems unlikely that in most contexts people learn about how best to interpret the Constitution, sociolegal scholars have argued conceptions about fairness and respect for others that guide social interactions are frequently constructed in terms of legal discourse (Ewick and Silbey Reference Ewick and Silbey1998).
One social context in particular, churches, may provide the transferable skills to perform constitutional interpretation. In the pews, parishioners learn how to interpret sacred religious texts. Different religions, denominations, and congregations promote diverse approaches to interpretation. In fact, one of the primary “culture war” divisions among religious people is how literally to interpret the Bible (see Wuthnow Reference Wuthnow1989; Hunter Reference Hunter1992). As such, people attending religious services learn from their religious leaders how to engage with and read the Scriptures.
In related work, prior studies indicate that churches can serve as environments that communicate useful political skills and reasoning patterns (Wald, Owen, and Hill Reference Wald, Owen and Hill1988; Djupe and Gilbert Reference Djupe and Gilbert2009; Djupe and Lewis Reference Djupe and Lewis2015), and clergy are sometimes able to provide important political information to their congregants. There is evidence that both Protestant pastors (Guth et al. Reference Guth, Green, Kellstedt, Smidt and Poloma1997; Smidt Reference Smidt2003; Crawford and Olson Reference Crawford and Olson2001; Fetzer Reference Fetzer, Crawford and Olson2001; Djupe and Gilbert Reference Djupe and Gilbert2003;) and Catholic priests (Jelen Reference Jelen2003; Smith Reference Smith2008) can influence their members. There are, however, a number of features about the clergy's position that may condition their influence, including their political difference from parishioners, their personal discretion about how stridently to engage with political matters, denominational demands to engage (or not), as well as their frequency of engagement and congregants' exposure to it through attendance (see McDaniel Reference McDaniel2008; Calfano Reference Calfano2009; Djupe and Gilbert Reference Djupe and Gilbert2009; Calfano, Oldmixon, and Gray Reference Calfano, Oldmixon and Gray2013; Calfano, Michelson, and Oldmixon Reference Calfano, Michelson and Oldmixon2017; Holman and Shockley Reference Holman and Shockley2017). Nonetheless, everyone agrees that clergy are in the position to help shape political attitudes.
Importantly, research on how clergy's interpretative style might be carried over into the political arena is limited. One strand of work has suggested that elite reasonning about the connection of religion and politics is important. Seeing a trusted decision-making process helps religious identifiers adopt their arguments (Calfano and Djupe Reference Calfano and Djupe2009; Djupe and Gwiasda Reference Djupe and Gwiasda2010). But, we want to take a step back to suggest that members may adopt interpretive styles from religious elites and apply them to political matters. While previous scholars have suggested this hermeneutical connection between religion and constitutionalism (Pelikan Reference Pelikan2004; Kalscheur Reference Kalscheur2005; Lund Reference Lund2009), there has been very little empirical inquiry into the relationship between religious interpretation and constitutional interpretation.
Levinson (Reference Levinson1988) suggests that Catholics, because of their organizational and theological structure, will adhere to the whole of the constitutional order, including the text and judicial decisions. Protestants (especially evangelical Protestants), because of their Biblicist theology and limited hierarchy, however, are primarily focused on the text. Of course, politics also matter. The American Catholic Church contains both conservative and progressive elements, particularly among Hispanic Catholics, which sometimes produces tension within (and between) the public and the Church hierarchy (Leege and Welch Reference Leege and Welch1989; Smith Reference Smith2008; McKenzie and Rouse Reference McKenzie and Rouse2012; Calfano, Michelson, and Oldmixon Reference Calfano, Michelson and Oldmixon2017; Holman and Shockley Reference Holman and Shockley2017; Kraybill Reference Kraybill2019). Protestants, too, have competing elements, though white evangelical Protestants dominate political discussions of the Protestant landscape. White evangelicals are more uniform in their conservative beliefs and their integration into Republican politics (Guth et al. Reference Guth, Green, Kellstedt, Smidt and Poloma1997; Layman Reference Layman2001; Claassen Reference Claassen2015).
In what follows, we analyze three separate data sets to test the relationship between different facets of religion and constitutional interpretation. First, using a national survey, we examine whether religious Catholics and Protestants relate to the Constitution similarly based on how they approach the Bible. We then analyze whether views of the Bible and the Constitution covary with respect to concepts like sacredness and unchangeability. Finally, we leverage a survey experiment to uncover the processes by which citizens come to these constitutional attitudes, seeking to ascertain the political ramifications of mass-level hermeneutics.
EXPLORING HERMENEUTICAL CONNECTIONS BETWEEN THE CONSTITUTION AND SCRIPTURE
In each segment of our analysis, we are focused on similar dependent variables: support for originalism versus a living Constitution approach; or support for a literal versus a metaphorical form of textual interpretation. Recall from the previous section that Justice Hugo Black combined a literal form of textualism with a concern for originalism, and it is possible that he adopted this jurisprudence because of the hermeneutical approaches he witnessed at his Primitive Baptist church.Footnote 1 The only empirical study of originalism attitudes to date (Greene, Persily, and Ansolabehere Reference Greene, Persily and Ansolabehere2011) provides some evidence of a connection between religious beliefs and constitutional interpretive preferences. The study demonstrated low, positive correlations between originalism and self-reports of church attendance frequency (r = 0.32) as well as evangelicalism (r = 0.31), and Americans who believe in the literal truth of the Bible were significantly more likely to identify as originalists.
In the portion of our analysis pertaining to the first cross-sectional survey, our hypotheses derive from Levinson's taxonomy of constitutional faith. We predict religious Catholics are more likely to identify as constitutional Catholics, and evangelical Protestants are more likely to identify as constitutional Protestants. If religious Catholics take a wider view of the meaning of the Constitution and are more supportive of the Supreme Court's authority to interpret it, it stands to reason they will be more likely to favor a living Constitution approach over originalism. Conversely, if evangelicals believe the text of the Constitution is their sola scriptura, and Supreme Court decisions are not the only legitimate source of constitutional meaning, they should feel a greater obligation to interpret it according to its original understanding.
The analysis of our second cross-sectional survey builds upon the earlier analysis in examining whether the links between religious and constitutional interpretive preferences are direct or merely products of underlying religious and political affiliations. However, in this design, we do not impute theological beliefs on the basis of religious affiliation. Instead, we ask respondents an identical battery of questions measuring their views of the Bible and the Constitution. We predict that political factors will influence their biblical and constitutional interpretive preferences, but we also predict that religious interpretation will have an independent effect on constitutional attitudes.
In the experimental portion of our analysis, we hypothesize that the messages provided by religious leaders will influence individuals' interpretive approaches to biblical passages. However, the effectiveness of this treatment should vary based on a respondent's pre-existing knowledge of a particular Bible verse. In addition, we hypothesize hermeneutical guidance from religious leaders will also influence how individuals interpret clauses in the Constitution.
STUDY 1: A SURVEY OF AMERICAN CONSTITUTIONAL FAITH
Data and Variables
We test our predictions on a survey administered by the Pew Research Center, made available by the Roper Center for Public Opinion Research.Footnote 2 The survey, conducted February 12–26, 2014, consists of 3,337 landline and cellular telephone interviews from a national adult sample weighted to the 2012 Census via gender, age, education, and race.Footnote 3 The dichotomous measure of a respondent's interpretive philosophy serves as the dependent variable. The survey asked, “Should the U.S. Supreme Court base its rulings on its understanding of what the U.S. Constitution meant as it was originally written, or should the court base its rulings on its understanding of what the U.S. Constitution means in current times?”Footnote 4 The living Constitution approach enjoyed slightly more support than originalism (51.1–49.9%).
The first independent variable of interest is religious tradition. Combining racial self-identification, religious self-identification, and whether the respondent considered herself a born again or evangelical Christian, we constructed a variable with the following categories: white evangelical, mainline Protestant, black Protestant, Catholic, Other Christian, Jewish, Other Faith, and Unaffiliated for religious typologies (see Kellstedt et al. Reference Kellstedt, Green, Guth, Smidt, Green, Guth, Smidt and Kellstedt1996; Steensland et al. Reference Steensland, Robinson, Wilcox, Park, Regnerus and Woodberry2000; for using born again or evangelical self-identification, see Burge and Lewis Reference Burge and Lewis2018; Smith et al. Reference Smith, Claire Gecewicz and Hackett2018). Religiously unaffiliated respondents serve as the reference group in the analysis. The second independent variable of interest is scriptural literalism. Using the holy book that corresponds to the respondent's religious tradition, the survey inquired, “Would you say that the Bible/Torah/Koran/Holy Scripture is to be taken literally, word for word, or, not everything in the Bible/Torah/Koran/Holy Scripture should be taken literally, word for word?” Thirty-six percent of respondents endorsed a literalist position.
Prior scholarship (Greene, Persily, and Ansolabehere Reference Greene, Persily and Ansolabehere2011) found older Americans, conservatives, Republicans, and supporters of the Tea Party were significantly more likely to identify as originalists. Similarly, we control for other demographic characteristics that could structure constitutional attitudes. The survey measured ideology on a 5-point scale with higher values indicating conservatism, and we also control for whether the respondent agrees with the Tea Party movement, which was still quite prominent when the survey was administered. The survey constructed party identification along a 7-point scale with higher values corresponding to the Republican Party. Respondent race and ethnicity were captured in a combined, categorical variable, with whites serving as the reference group. We also control for frequency of religious attendance (6-point scale), age, gender, and how closely the respondent follows politics (4-point scale). Because originalism often (but not always)Footnote 5 leads to conservative judicial decisions, it is possible that evangelicals only support originalism instrumentally in the hopes the Supreme Court will decide cases consistent with their policy preferences. As a result, we control for support for same-sex marriage (4-point scale).Footnote 6
Results of Cross-Sectional Analysis
Table 1 reports the results of three logistic regression models of support for originalism. Model 1 measures the effect of demographic characteristics, while Model 2 adds same-sex marriage attitudes, and Model 3 includes respondent's approach to interpreting holy scripture. The results of Model 1 highlight how politics structure the debates over constitutional interpretation. Partisanship, ideology, and Tea Party attitudes are all significantly related to originalism attitudes. A one-unit change in partisanship towards the Republican Party increases the predicted probability of originalism support by 0.06. Supporters of the Tea Party are 35% more likely to identify as originalists than opponents of that movement. Respondents who claim to follow politics more closely are significantly more likely to endorse originalism, while higher levels of education significantly reduce the likelihood of preferring this interpretive philosophy.
Logistic regression models of support for originalism
Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.
At the same time, religious affiliation and religiosity are associated with how individuals approach the task of constitutional interpretation. The results of Model 1 provide support for our Constitutional Catholicism and Constitutional Protestantism hypotheses, and Figure 1 displays the predicted probability of supporting originalism across religious denominations. Moreover, the frequency of church attendance is also related to views of originalism (p = 0.07).
Predicted probability of supporting originalism from Model 1, with 95% CIs.
In regard to religious affiliation, Catholic respondents are 7.5% less likely to favor originalism than unaffiliated respondents (p = 0.08). Interestingly, a t-test reveals no significant difference in mean ideology between Catholic and non-Catholic respondents (t = 0.51), which suggests that greater Catholic preference for a living Constitution may be grounded in theological attitudes, not ideological differences. Other analyses presented in the Appendix also show that there are no statistically significant differences between white and non-white Catholics, though there is some indication that ideologically conservative Catholics may be slightly more likely to prefer originalism.Footnote 7
White evangelical affiliation is more strongly related to conservative constitutional interpretation than Catholic affiliation is. In fact, white evangelicals appear to be the religious group most clearly connected to originalism, though there are not statistically significant differences between a few religious traditions (see Figure 1). Compared to the religiously unaffiliated, white evangelicals are 12% more likely to identify as originalists (p = 0.01). As theoretically expected, there is an important, statistically significant contrast between evangelical Protestants and Catholics in their support for originalism, shown by the non-overlapping confidence intervals in Figure 1.
One year following the publication of this survey, the three self-proclaimed originalist justices on the Supreme Court voted dissented from the majority decision and voted (along with Chief Justice Roberts) to allow bans on same-sex marriage to stand. The results of Model 2 demonstrate the connection between same-sex marriage attitudes and constitutional interpretation extends from the bench to public opinion. A one-unit increase in support for same-sex marriage decreases the predicted probability of preferring originalism by 0.11 (p < 0.001). This addition depresses the effect of Catholic affiliation (p = 0.10), and the coefficient for white evangelicals is no longer a significant predictor, which may indicate these interpretive preferences are more a result of politics, not theology.
Model 2 also highlights an interesting set of findings concerning race and constitutional attitudes. Compared to whites, African American respondents are 23% less likely to identify as originalists (p = 0.03). Black Protestants, on the other hand, are 19% more likely than unaffiliated respondents to prefer originalism (p = 0.08). In addition, Black Protestants appear to be more likely to favor originalism than other African Americans (p = 0.05). A difference of means t-test reveals no significant difference in same-sex marriage attitudes between Black Protestants and other African Americans (t = 0.28), which again suggests the difference in levels of support for originalism may be a result of religious differences in the black community.
Finally, Model 3 indicates preferences for scriptural interpretation are related to preferences for constitutional interpretation. Respondents who believe their holy book should be interpreted literally are 8% more likely to favor an originalist interpretation of the Constitution (p = 0.03). Overall, the survey analysis presents some evidence of religious affiliation influencing constitutional attitudes. On the whole, however, partisanship seems to be more important. We proceed to the second phase of the analysis, which uses a paired survey question design to more explicitly link individuals' views about the sacredness of the Bible and the Constitution and their approach to interpreting each.
STUDY 2: A SURVEY OF RELIGIOUS AND CONSTITUTIONAL HERMENEUTICS
Data and Variables
Our second cross-sectional survey builds upon the previous analysis, but it more explicitly engages the ideas of constitutional faith and comparative hermeneutics by leveraging a unique survey design. This approach gives us the ability to analyze the relationship between hermeneutical styles for the Bible and the Constitution. The data come from a May 2018 national sample of 1,428 individuals using Qualtrics panels. While technically a convenience sample, the Qualtrics data was balanced by census region and gender, and the participants are nearly identical to nationally representative benchmarks. Table A3 in the Appendix contains demographic information of the survey respondents.
In the survey, respondents were presented with a series of four side-by-side questions, asking them about their views of the Bible and the U.S. Constitution. The four questions were: (1) “Is a Sacred Document;” (2) “Needs to change with the times;” (3) “Should be interpreted word for word;” and (4) “Should be interpreted metaphorically.” Individuals were then asked to select one of the following response options for both the Bible and the Constitution: (1) “strongly agree,” (2) “agree,” (3) “disagree,” and (4) “strongly disagree.” In the analysis below, the answers have been recoded, so that all responses run from the liberal (1) to the conservative (4) direction.
Figure 2 displays the responses for each question for both the Bible and the Constitution. The descriptive results suggest that people evaluate both texts similarly and with a high level of sacredness. Among the full sample, the correlations between biblical and constitutional views on each question range from 0.28 (metaphoric) to 0.44 (change with the times). While both texts score high on sacredness, more people agree or strongly agree that the Bible is a sacred document (87%), compared to the Constitution (67%). Evangelical Christians, as a subgroup, score higher on the sacredness of both documents (95% for the Bible; 78% for the Constitution).
Comparing Constitutional and Biblical Interpretive Indices.
The interpretive questions provide some interesting nuance. More people disagree that the Bible needs to change with the times (67%), compared to the Constitution (45%). That said, respondents were more likely to support reading the Constitution (67%) word-for-word than the Bible (41%). The response was nearly identical to the percentage who opposed the metaphoric reading (66% for the Constitution and 42% for the Bible), suggesting consistency in responses, at least when the items are side-by-side. Though evangelicals are more likely to prefer a literal interpretation of the Bible than the general population, they too have greater support for reading the Constitution word-for-word (66% for the Bible; 77% for the Constitution).
To further understand people's attitudes about these questions, we specified a series of multivariate regression models. Separate additive indices of the Constitution and Bible questions serve as dependent variables with higher values indicating more conservative interpretive approaches. The indices hold together well according to statistical tests.Footnote 8 Recoding to a zero baseline, the indices range from 0 to 12, making linear regression an appropriate choice. In what follows, we first predict support for the Biblical Interpretive Index, followed by support for the Constitutional Interpretive Index.
In these first two models, the independent variables are similar to the models in Table 1 above. We include separate variables for various religious traditions (based on denominational affiliation), with the religiously unaffiliated serving as the comparison group, and a measure of religious commitment (attendance). For political variables, we include party identification, ideology, and demographic controls for race, sex, age, and education. The final model is the same as the second—predicting support for the Constitutional Interpretive Index—but we also add the Biblical Interpretive Index as a separate independent variable. This model allows us to test one of the centerpieces of our theory—that respondents' biblical hermeneutics directly affect their constitutional hermeneutics.
Results
To begin, we analyze how religious characteristics and political views are related to support for the Biblical Interpretive Index. Figure 3 presents the coefficient points and 95% confidence intervals for all religious and political independent variables. The full results are in the leftmost column of Table 2. Almost every denomination of Christianity or being Jewish is related to taking more traditional, conservative positions on the Bible, compared to being religiously unaffiliated. Evangelicals express the most conservative support, though their point estimate is not statistically distinguishable from Mainline Protestants or Other Christians. Those who attend religious services more frequently are also more likely to interpret the Bible more traditionally, as are Republicans and conservatives.
Linear Regression Model of Support for Biblical Interpretive Index from Model 1 with 95% CIs.
Linear regression models of Biblical Interpretive Index and Constitutional Interpretive Index
Standard errors in parentheses. *** p < .01, ** p < .05, * p < .1
Religious affiliation exerts less influence in determining constitutional attitudes. The left-hand panel of Figure 4 and the middle column of Table 2 present the results. In this model, only evangelical Christians have significantly more conservative views of the Constitution compared to the religiously unaffiliated. Those who attend religious services more often are also more likely to score higher on the Constitutional Interpretive Index. Politically, as expected, both Republicans and conservatives again support the conservative approach to the Constitution.
Linear regression models of support for Constitutional Interpretive Index from Model 2 (left) and Model 3 (right), with 95% CIs.
In the right panel of Figure 4, we display the results of Model 3, which is the same as Model 2 except it includes the Biblical Interpretive Index. What we find is that not only does the R 2 increase, but the effects of the religious coefficients also change. The Biblical Interpretive Index is strongly predictive of the Constitutional Interpretive Index, with a magnitude that is similar to both partisanship and ideology. Moreover, adding the Biblical Interpretive Index into the model negates the statistically significant effect of being an evangelical. The evangelical variable transitions from positive and statistically significant to nonsignificant, as does the religious attendance variable. The effect of being Catholic changes from being negative and nonsignificant to significant and negatively related to support for the conservative Constitutional interpretation style.
The findings from this last model suggest that biblical and constitutional interpretive approaches are directly related, as well as that the biblical hermeneutical approach's influence outweighs that of traditional measures of religious affiliation.Footnote 9 However, the cross-sectional analyses fall short of specifying causation. We proceed to the final phase of the analysis to try to untangle partisan and religious effects further. The experimental manipulation allows us to evaluate whether citizens can learn approaches to constitutional interpretation in religious settings.
STUDY 3: HOW RELIGIOUS HERMENEUTICS BECOME CONSTITUTIONAL HERMENEUTICS—A SURVEY EXPERIMENT
Data and Variables
To test our hypotheses about whether individuals can learn hermeneutical skills in a religious setting and then apply them in a political context, we developed a survey experiment (see Appendix for question wording). Our experiment randomly presented respondents with one of two religious passages from the Hebrew Bible: (1) Genesis 2—the creation narrative where God forms man “from the dust of the ground” and “made a woman from the rib he had taken out of the man;” and (2) Joshua 10—where Joshua asked God to stop the sun from setting in order for the Israelites to defeat the Amorites, and “the sun stopped in the middle of the sky and delayed going down about a full day.” Respondents were then asked to rate the accuracy of the passage. From there, respondents were randomly assigned to one of four conditions: (1) a religious leader giving a literal interpretation of the passage at a local house of worship; (2) a religious leader giving a metaphorical interpretation of the passage at a local house of worship; (3) two religious leaders giving competing interpretations (literal and metaphorical) at a local house of worship; and (4) a control group with no religious leaders giving interpretations.
We then asked how much the respondents agreed with the religious leaders and how they would describe the leaders. We ended this section by asking the respondents to choose whether they would interpret the passage literally or non-literally. We do not expect that a religious leader's influence will be uniform across all religious texts, but rather the effect should be greater when a passage is less familiar to the reader. In other words, if a person comes to a particular passage with an already well-formed view of how the text should be interpreted, then a religious elite's message is less likely to influence the person's hermeneutical approach.
Following the religious passage, we transitioned to asking respondents about two constitutional passages. Each respondent was shown text from the Free Speech Clause of the First Amendment and the Search and Seizure Clause of the Fourth Amendment. Respondents were randomly assigned to receiving either the First or Fourth Amendment section first. Respondents provided their opinion on the constitutional text on a 10-point slider scale.
For the First Amendment, respondents were asked, “Do you think there are circumstances in which the First Amendment's protections of free speech should be limited, or do you think there should be no limits on free speech protections?” Higher values on the scale corresponded to fewer limits on free speech. The Fourth Amendment question is intended to draw upon the prefatory portion of the Search and Seizure Clause—“the right of the people to be secure in their persons, houses, papers, and effects.” A literal interpretation of that passage would limit the reach of the Search and Seizure Clause to the context specified in the text of the amendment, whereas a more metaphorical reading would encompass a broader right of privacy. Respondents were asked, “Do you think the right to privacy should be limited to the items listed in this passage from the Fourth Amendment, or do you think the right to privacy should include other items not mentioned in this passage?” Here, higher values on the scale indicated support for a more expansive right to privacy.
Following the experimental components, we asked a variety of political and religious questions, including one's religious affiliation, attendance, and commitment as well as views of the Bible and evolution. The political items include ideology, party identification, 2012 presidential vote, views on same-sex marriage, abortion, and the death penalty, and ratings of various political groups. Finally, respondents answered a battery of demographic questions covering race, gender, education, and age.
The data for this study were collected from Amazon's Mechanical Turk (MTurk). We paid $0.50 to 1,800 independent respondents during July 2014. In our request, we specified that MTurk workers be U.S. residents, have a 95% success rate, and have completed more than 500 surveys. We also limited responses to one per worker, and we completed an extra check by deleting multiple responses from the same IP address.
There are certainly some drawbacks to using MTurk workers to complete surveys, including the fact that they are essentially professional survey takers, more likely to be younger, and more technologically savvy, among others. These workers are also more liberal and slightly more Democratic than the respondents from the American National Election Study (Berinsky, Huber, and Lenz Reference Berinsky, Huber and Lenz2012). Of particular note for our study, the MTurk respondents were significantly more secular than the broader population, which is common among MTurk samples (Lewis et al. Reference Lewis, Djupe, Mockabee and Wu2015). All this said, results derived from samples of MTurk workers have been shown to be more representative of the national population than other types of convenience samples, including student samples, while maintaining a high level of internal validity (Buhrmester, Kwang, and Gosling Reference Buhrmester, Kwang and Gosling2011; Horton, Rand, and Zeckhauser Reference Horton, Rand and Zeckhauser2011; Berinsky, Huber, and Lenz Reference Berinsky, Huber and Lenz2012; Anderson and Lau Reference Anderson and Lau2018), even on religious variables (Lewis et al. Reference Lewis, Djupe, Mockabee and Wu2015). More importantly, because we are investigating an experimental manipulation, we are less concerned with external validity and more focused on internal validity, which has been shown to be similar to national samples (Coppock Reference Coppock2019).
Experimental Results
We first examine whether this treatment produced direct effects. If the experimental manipulation of clergy's messages influenced the way respondents interpreted the biblical passages, we should observe differences across conditions in the proportion of those who took a literal interpretation approach versus other types. Table 3 shows the percentage of respondents in each experimental condition who reported interpreting their assigned Bible passage literally, word for word. The differences across conditions are statistically significant (χ2 = 10.4, p < 0.02) for interpretation of the less familiar Joshua passage. The differences with regard to the interpretation of the Genesis passage are smaller and not statistically significant. Recall that our expectation was that priming effects were likely to be greater for unfamiliar texts than for passages that were already well known to the respondent. It is quite possible that many respondents came into the study having already formed firm views about the creation account in Genesis and therefore were not as susceptible to the influence of religious leaders.
Interpretation of bible passage by experimental condition
Entries are column percentages with cell counts in parentheses.
Next, we evaluate whether the exposure to different religious interpretations (i.e., treatments) affected respondents' views of the Constitution. This design isolates the relationship between the priming of a religious hermeneutic and the preference for a legal hermeneutic. We first consider the distribution of our dependent variables—the 10-point scale of support for fewer limits on free speech or a more expansive right to privacy. Given the high degree of veneration the Constitution enjoys, it should come as no surprise that the average respondent scored quite highly on these measures. The mean responses were 6.97 for the First Amendment question and 7.32 for the Fourth Amendment question, and the modal response for both was a 10. On each constitutional question, just over 40% of respondents chose either a 9 or a 10. The Appendix includes histograms of these distributions (Figures A2 and A3).
These distributions suggest the possibility of censoring, which occurs when cases above some threshold take on the value of that threshold (Siegelman and Zeng Reference Sigelman and Zeng2000). In that scenario, the true value of the variable might be higher than the scale allows. For some respondents, a 10 might perfectly represent their level of support for freedom of speech, while for others their views on civil liberties are artificially constrained by a scale that cannot accommodate their enthusiasm.Footnote 10 Consequently, we employ tobit regression with a correction for right-censoring.
Table 4 presents the results of three tobit regression models of the constitutional attitudes for respondents exposed to the Joshua passage. As Table 3 indicated, the religious leaders' interpretation of the Joshua passage affected the respondents' interpretation of that passage in the predicted directions, whereas respondents were not influenced by the treatment in those interpretation of the Genesis passage. By confining the analysis to respondents exposed to the Joshua passage, we examine whether a treatment that successfully shaped a respondent's religious hermeneutic could also influence their constitutional hermeneutic. The tobit models also include a control for African American, as this variable did not randomize well across treatment conditions (F = 2.36, p = 0.07).
Tobit models of scriptural interpretation treatments on constitutional interpretive attitudes
Standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.10.
The left-hand column of Table 4 displays the results of respondents' interpretive preferences on the First Amendment. Recall here that higher values on the 10-point scale indicate support for fewer limits on freedom of speech, which is a more literal interpretation of this constitutional right. The analysis provides some support for our hypothesis that hermeneutical skills gained in a religious context can subsequently influence attitudes on constitutional interpretation. Respondents exposed to a literal interpretation of the Joshua passage were significantly more likely to endorse a literal interpretation of the First Amendment than those in the control group. Those who encountered a metaphorical interpretation of the Joshua passage or competing interpretations were no different in their constitutional attitudes than members of the control group. Finally, African American respondents were significantly more likely to endorse restrictions on free speech than members of other racial groups.Footnote 11
The center and rightmost columns of Table 4 display the results of two models analyzing Fourth Amendment interpretive attitudes. On this dependent variable, lower values are associated with a more literal interpretation of the Fourth Amendment—that it only protects a right to privacy in one's person, home, and papers. As indicated in the middle column, there are no significant effects between treatment conditions and interpretive preferences regarding the right to privacy, although African Americans are somewhat more likely than others to favor a stricter interpretation of the Fourth Amendment. One possibility for this non-finding is that the right to privacy is so strongly connected to cultural issues like gay rights and abortion that these attitudes overwhelm the effects of the treatment. The final model includes controls for abortion and same-sex marriage attitudes, and support for these rights is significantly associated with support for a broader right to privacy. While none of the treatment groups have privacy attitudes that significantly differ from the control group, the effect of exposure to a literal interpretation of the Joshua passage trends in the predicted direction (p = 0.181).
While nuanced, the experimental results are suggestive that religious leaders can prime hermeneutical styles, and these effects carry over to the constitutional domain. This opinion leadership is more effective when the biblical passages are less well known (such as the Joshua passage), and when the relationship to a clause in the Constitution is not contaminated by intervening political attitudes (such as the disagreement concerning privacy). When combined with the cross-sectional data about religious and biblical hermeneutical styles and constitutional faith, the data provide some evidence of a direct connection.
DISCUSSION AND CONCLUSION
If citizens are to govern themselves in republics with written constitutions, then it is imperative that they can engage in some form of constitutional interpretation. But how do they learn to do hermeneutics? To date, we have more theories and speculation about how the public might engage in hermeneutics than scientific evidence. Although we expect that people can use partisan-motivated reasoning to determine their constitutional philosophies, we also develop a theory of constitutional hermeneutics, whereby citizens connect their views of sacred religious texts to sacred governing documents.
Using multiple datasets and approaches, we empirically tested these theoretical discussions, examining if ordinary citizens might acquire hermeneutic approaches needed to engage in constitutional interpretation. Our cross-sectional and experimental data allowed us to begin to develop answers to these questions. First, as we hypothesized, rank-and-file Catholics are more likely to support a living Constitution, a jurisprudence in which the Court has ultimate authority to give meaning to both the text and tradition of the Constitution. Likewise, evangelical Protestants tend to be attracted to originalism, which limits the scope of inquiry to the text and does not allow the Supreme Court's understanding of the Constitution to trump that of the founders.
While the originalism findings from Study 1 are interesting, one must carefully consider whether they are the result of interpretive approaches encountered in their respective churches or simply a reflection of underlying political differences between these groups. Politics certainly predicts support for originalism, but so do hermeneutical styles. In the Pew data, our findings amongst African Americans and Catholics suggest hermeneutics also plays a direct role. While there are no differences in same-sex marriage attitudes between Black Protestants and other African Americans, Black Protestants are significantly more likely to endorse originalism. Likewise, there were no ideological differences between Catholics and non-Catholics in our sample, yet Catholics were marginally more likely to favor a living Constitution hermeneutic. Moreover, in the Qualtrics data, we find a strong direct linkage between biblical interpretive styles and constitutional interpretive styles that overwhelm the effects of religious affiliation and behavior.
The survey experiment provides some support for a possible causal mechanism that explains our cross-sectional survey analysis. It appears that religious elites may be able to influence the hermeneutic approaches that ordinary people use to interpret sacred texts and the Constitution. In particular, literal religious interpretation can influence respondents into adopting a more literal constitutional interpretation of the First Amendment. Places of worship are likely venues for learning how to interpret sacred texts, and our experiment shows that manipulating the cues given by religious elites makes a difference in the interpretive approaches that people take. The findings of the survey experiment also offer support for the idea that hermeneutical styles are correlated across domains.
As with any empirical study, there are limitations. Some readers might feel that our experimental stimuli—the quotes from fictitious religious leaders—do not replicate the real-world elite cues that people encounter in their daily lives. However, to the extent that our stimuli missed the mark, we think it is likely that we have underestimated the effects of religious elites' discourse. Our excerpts were only short paragraphs from unknown religious figures that were read by online respondents who had little motivation other than a small payment.
In contrast, the messages people encounter in the real world come from religious figures that are often well known to parishioners. It seems likely that their messages would carry more weight than those of the fictitious leaders in our study. So, if anything, we expect that our findings of elite influence may be understated. We also recognize that MTurk respondents are, on average, younger, more politically liberal, and more secular than the general public as a whole. Again, these demographic differences actually work to provide a tougher test for our experimental study. While we found considerable evidence to support the translation of hermeneutical styles, especially conservative ones, more research is needed on different samples.
The downstream effects of allowing ordinary citizens to have a greater share in constitutional decisions, including participation in judicial elections and ballot initiatives, are central to modern American politics. An unrecognized consequence of the increasing secularization of the political left in America is that popular constitutionalism may gain followers more easily on the ideological right, contrary to the wishes of its academic proponents (see Schmidt Reference Schmidt2010). The politics of hermeneutics, therefore, is a largely under-analyzed component of American culture wars, being fashioned in the pews and capitalized on by movements on the political right. Scholars should pay more attention to the empirical roots of the public's views toward constitutional interpretation as more political controversies become constitutionalized.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/S175504831900021X
