The contours of constitutional approval.
Scholars and judges agree on the importance of constitutional approval--that is, people's subjective support for their constitution. The Supreme Court has asserted that it owes its very legitimacy to popular backing for its decisions. Academic luminaries have concurred, while also connecting constitutional approval to constitutional compliance and durability, as well as the easing of the countermajoritarian difficulty.
Until now, though, no information has been available on either the levels or the causes of constitutional support. In this Article, we rectify this shortcoming by presenting the results of a nationally representative survey that we conducted in late 2014. The survey asked respondents about their approval of the federal Constitution and of their state constitution, and about several potential bases for support. We also supplemented the survey by coding dozens of features of state constitutions. This coding allows us to test hypotheses about the relationship between constitutional content and constitutional backing.
What we find is illuminating. First, people highly approve of their constitutions--the federal charter more so than its state counterparts. Second, approval is unrelated to what constitutions say; it does not budge as their provisions become more or less congruent with respondents' preferences. Third, approval is only weakly linked to respondents ' demographic attributes. And fourth, the most potent drivers of approval are constitutional familiarity and pride in one's state or country. To know it--and to be proud of it--is to love it.
These results unsettle several literatures. They mean that people form opinions about constitutions differently than about other institutions. They also mean that comparativists may be going down a blind alley as they focus ever more intently on constitutional design. But perhaps our study's clearest implication is for leaders who value popular support for their constitution. Our advice to them is to forget about constitutional change, and instead to try to build the public's knowledge and appreciation of the charter. Constitutional approval, like statecraft, is ultimately a project of soulcraft.
TABLE OF CONTENTS INTRODUCTION I. BACKGROUND A. Theory 1. Meaning 2. Consequences 3. Causes B. Empirics II. METHODOLOGY A. Survey Design B. Congruence Scores C. Non-Substantive Features III. DESCRIPTIVE EXPLORATION A. Overall Levels B. Geography and Demography C. Civic Knowledge D. Institutional Attitudes E. Constitutional Congruence F. Non-Substantive Features IV. EXPLANATORY ANALYSIS A. Demographic Attributes B. Civic Knowledge C. Institutional Attitudes D. Constitutional Congruence E. Non-Substantive Features V. DISCUSSION A. Institutional Comparisons B. Domestic Prescriptions C. Additional Literatures D. Future Research CONCLUSION
Justice Kennedy concluded his opinion with a flourish in a landmark 2005 case. "Over time, from one generation to the next," he declared, "the Constitution has come to earn the high respect and even, as Madison dared to hope, the veneration of the American people." (1) But has it? Do Americans actually feel "high respect" and "veneration" for their federal Constitution? And if so, what about their other constitutions--the charters that structure the governments of the fifty states? Do Americans prize them too?
Justice Kennedy also offered an intriguing explanation for the support (allegedly) enjoyed by the federal Constitution. Its "doctrines and guarantees"--federalism, the separation of powers, the Bill of Rights, and so on--are "essential to our present-day self-definition and national identity." (2) "Not the least of the reasons we honor the Constitution, then, is because we know it to be our own." (3) But is this, in fact, why we honor it? Does our allegiance stem from its congruence with our values, or from something else entirely? And even if Justice Kennedy is right about the cause of federal constitutional approval, does his claim hold for the states as well?
In this Article, we begin to answer these questions. We do so not just because they are prompted by Justice Kennedy's ruminations, but also because support for the constitution is a critical--and critically understudied--concept. Luminaries from the bench and academia have argued that it is a key driver of constitutional legitimacy: that is, the loyalty a charter commands from its constituents. These observers also have linked it to constitutional compliance, durability, and status as law, as well as the easing of the countermajoritarian difficulty. (4) But to date, no one has tried to measure it, to determine what the levels and causes of constitutional approval actually are. This empirical project is the linchpin of this Article.
Our methodology is straightforward. (5) To find out whether and why people support their constitutions, we simply asked them. In October 2014, we carried out a nationally representative survey with roughly 2,000 respondents, two questions of which were to what extent people approve of the federal Constitution and of their state constitution. The survey also included questions about an array of potential bases for support: demographic attributes (gender, age, race, education, and income), civic knowledge (about the constitution specifically and the news generally), and institutional attitudes (toward one's state, country, and party).
In isolation, though, the survey would have been unable to test some of the most salient hypotheses about constitutional approval--such as Justice Kennedy's claim that it follows from consistency with people's preferences. We therefore supplemented the survey by coding many of the features of the fifty state constitutions. We tracked whether or not they contain twenty-nine substantive provisions, as well as their age, length, and amendment frequency. The latter three variables slide directly into our analyses, while we pair the former with questions from our poll to create a measure of congruence. That is, we compare the provisions that each respondent wants in her state constitution with the provisions actually in the document, and so determine how closely it reflects her views.
We find, first, that Americans generally back their constitutions, though to different extents at the federal and state levels. The federal Constitution achieves an average approval score of 7.8 out of 10, while state constitutions earn a somewhat lower rating of 6.7. Constitutional support also does not vary much geographically. The federal Constitution is most popular in Idaho and least popular in Vermont, while Wyoming residents are happiest with their state constitution and Mississippians are least pleased with theirs. None of these state-level averages diverges very far from the national mean. (6)
But the existence of constitutional approval is less interesting than its explanations. To identify them, we build regression models in stages for both federal and state constitutional support. (7) We add, in turn, each of our five sets of hypotheses, involving (1) demographic attributes, (2) civic knowledge, (3) institutional attitudes, (4) constitutional congruence, and (5) non-substantive constitutional features. (The last two of these apply only at the state level since there is only one federal Constitution, and so no federal constitutional variation.)
Perhaps our most surprising finding, in light of Justice Kennedy's (and others') predictions, is that how closely a constitution corresponds to a respondent's preferences essentially has no impact on her approval of the document. Congruence fails to rise to statistical significance in any of our models, and varying it from its minimum to its maximum barely budges support. Our results for non-substantive constitutional features are equally unimpressive. Our respondents appear entirely unmoved by their charters' age, length, and amendment frequency.
The demographic story is somewhat more complicated. Gender, education, income, and most racial categories either fail to attain significance, or have small and inconsistent effects on approval. But older respondents reliably rate their constitutions more favorably than their younger peers. And unique among racial groups, African Americans consistently are less constitutionally satisfied, even controlling for other demographic and socioeconomic factors. America's familiar black-white cleavage thus extends to people's sentiments toward their charters.
A clearer picture emerges for civic knowledge and institutional attitudes. At both the federal and state levels, respondents who are more familiar with their constitutions, and who follow current events more closely, are more supportive of the documents. Similarly, at both levels, respondents who are prouder of where they live are stauncher constitutional advocates. In fact, the results for constitutional knowledge and jurisdictional pride are the most robust generated by our models. As these variables go from their minimums to their maximums, about a three-point spike in approval ensues (on a ten-point scale).
The most important implication of our findings is that constitutional support cannot be won through constitutional refinement. Since neither charters' substantive content nor their non-substantive features influence approval, constitutional design is effectively useless as a tool for increasing public backing for the document. This is a sobering truth for constitutional drafters, many of whom hope that their handiworks will reshape society in fundamental ways. Constitutions may have all kinds of consequences, but contra Justice Kennedy, earning the people's "high respect" and "veneration" is not one of them. (8)
Another insight from our analysis is more sanguine. Leaders who want their constituents to back their constitution are not powerless to bring about this outcome. But the right strategy is not to tweak the document to make it more attractive, but rather to boost people's familiarity with it and to swell their pride in their state or country. How can this be done? This is not the place for detailed prescriptions, but civic education, in the form of classes, marketing campaigns, and the like, is an intuitive way to inform the public. These techniques may also foster civic pride, but here the sounder option may be actually to compile a record worth being proud of. A well-run jurisdiction is its own reward--and if it results in higher constitutional approval, so much the better.
Our findings confirm in some respects, but challenge in others, several distinct literatures. (9) The first is normative constitutional theory, several of whose leading lights contend that popular support for the constitution is necessary for the document to achieve legitimacy, compliance, durability, and binding legal status. These scholars should celebrate our results, which show that constitutional backing is high and so imply that key constitutional values indeed are being realized. But these observers may be taken aback by our conclusion that constitutional approval is unrelated to constitutional content, which contradicts their widely held view that charters must be just in order to be popularly accepted.
The second literature, sounding more in political science than in law, examines the reasons for other institutions' approval (especially the Supreme Court and Congress). For the most part, it holds that knowledge and congruence are crucial factors while demography is not. We arrive at similar judgments as to knowledge and demography in the constitutional context. But to reiterate, however relevant it may be in other areas that policies correspond to people's preferences, it is immaterial to support in ours.
The third area is the study of comparative constitutional design. Historically, it has focused on the impact of different drafting choices on outcomes such as compliance, durability, growth, and yes, public backing. More recently, though, attention has shifted from constitutional substance to the process of constitutional ratification. Our results throw cold water on the notion that public opinion toward the constitution can be influenced by what the document says. But they dovetail nicely with the growing emphasis on ratification procedure. If people are more involved in the constitution's drafting and entry into law, they also may become more familiar with it and prouder of their own pivotal role. And these factors, again, are the primary drivers of constitutional approval.
The fourth and final literature is the sociological analysis of nationalism and its consequences. These consequences are usually thought to be mixed--positive when nationalism takes the form of patriotism, but negative when it transmutes into an assertion of national superiority. Our result that jurisdictional pride boosts support for the constitution reveals another favorable aspect of patriotism, one that has not been documented by any studies to date.
The Article unfolds as follows. In Part I, we discuss the existing work on the causes and consequences of constitutional approval. While there is ample theoretical scholarship, the available empirics deal almost exclusively with other institutions. Next, in Part II, we explain our methodology. We describe the nationwide survey we conducted as well as our coding of state constitutions. Parts III and IV then form the Article's analytical core. Part III offers a descriptive account of constitutional backing, while Part IV constructs our federal and state regression models. Lastly, in Part V, we comment on the implications of our findings. We address how they relate to other literatures, what policy reforms they entail, and how they could be bolstered by further research.
One more introductory point: Because this is the first study to assess constitutional support empirically, our analysis is necessarily provisional. We are sure there are ways to refine our measurements of both support and its potential causes. It also is possible our conclusions would change if we examined different time periods or countries. Nevertheless, we think there is substantial value to this project. A subject of great qualitative interest now--finally--has been opened to quantitative exploration.
Constitutional approval is not a self-explanatory concept. It is not obvious, at first glance, what it is, why it matters, or what produces it. So, in this Part, we comment briefly on the meaning, the consequences, and the causes of public support for the Constitution. We draw first from theoretical literatures in law and social science, and then from recent empirical work on the approval of other governmental institutions.
Half a century ago, David Easton distinguished between two kinds of political support that people may give. (10) The first, specific support, refers to "the satisfactions that members of a system feel they obtain from the ... outputs and performance of the political authorities." (11) It refers, that is, to people's approval of an institution's actual policies and operation. The second, diffuse support, "consists of a 'reservoir of favorable attitudes or good will that helps members to accept or tolerate outputs to which they are opposed.'" (12) It is people's willingness to comply with policies they dislike due to faith in the body promulgating the policies. It is essentially a synonym for institutional legitimacy.
Easton's framework is ubiquitous in the scholarship on public attitudes toward the branches of government. (13) And under it, there is no doubt that our variable of interest, constitutional approval, is closer to specific support than to diffuse support. (14) When people are asked how strongly they approve of their constitution, they are prompted to consider and then to rate their current views of the document. They are not induced to reflect on whether they still would adhere to its commands if they thought them unjust, or whether they would like to scrap it and start afresh. Constitutional approval, like equivalent questions about judicial, legislative, and executive branch approval, thus taps people's opinions on constitutional performance. It does not capture their feelings on constitutional legitimacy. (15)
So defined, as a measure of specific support for the constitution, why does constitutional approval matter? One possibility is that it is an intrinsic good, a value that is desirable for its own sake. Perhaps, in a constitutional democracy, we simply think that people should deem their charter worthy of respect and admiration. This is the position that Sandy Levinson takes in his classic work on "constitutional faith," a concept that straddles the line between specific and diffuse support. (16) "Is not the central question," Levinson asks, "whether, after reflection, we can genuinely ... declare and celebrate our status as Americans 'attached to the principles of the Constitution'?" (17)
Well, it may or may not be the central question. For one thing, as Levinson himself recognizes, too much constitutional approval may not be a good thing. (18) It may signify that people are ignorant of the document's shortcomings, unaware of its obsolete provisions, ethical compromises, and flawed notions of governance. For another, excessive approval may cause people to become too wedded to the constitution as it currently stands, too resistant to proposals to amend or replace it. Love that makes us blind is not love we should applaud.
Furthermore, support for the constitution may matter less for its own sake, and more because it gives rise to other deeply important values. Chief among these is constitutional legitimacy--or, in Easton's terms, diffuse support for the constitution. On at least two memorable occasions, Supreme Court Justices have argued that this is precisely why public approval of the Court's decisions is essential; without it, the Court's legitimacy would be severely undermined. In Baker v. Carr, the 1962 case that launched the reapportionment revolution, Justice Frankfurter asserted that "[t]he Court's authority--possessed of neither the purse nor the sword--ultimately rests on sustained public confidence in its moral sanction." (19) He worried (needlessly, it turns out) that the public would oppose the Court's foray into the "political thicket," (20) thus tarnishing the Court's reputation.
Likewise, in Planned Parenthood v. Casey, the 1992 case that entrenched Roe v. Wade as the law of the land, Justices O'Connor, Kennedy, and Souter declared that the Court's "legitimacy" is a "product of ... the people's acceptance of the Judiciary as fit to determine what the Nation's law means." (21) The Justices feared that if the Court reversed Roe "under fire" from Roe's critics, a "loss of confidence" would follow that the Court makes its decisions based on law rather than "political pressure." (22) This loss of confidence, in turn, "would subvert the Court's legitimacy beyond any serious question." (23) The Court's authority would fall in tandem with its public support.
The Justices, though, are not the only ones to have speculated about a link between approval and legitimacy. So too have scholars in constitutional law, legal philosophy, and political science. In constitutional law, Jack Balkin, Richard Fallon, Frederick Schauer, and David Strauss all have contended that the Constitution's legitimacy stems from its continuing endorsement by the public. (24) As Fallon has put it, "The Constitution is law not because it was lawfully ratified ... but because it is accepted as authoritative." (25)
In legal philosophy, similarly, positivists like H.L.A. Hart and Brian Leiter have asserted that a norm counts as law if there is "general acceptance of or acquiescence in" its legally binding status. (26) Under this approach, the social fact that a constitution is widely supported makes it more likely that its provisions are treated as compulsory by the public. And in political science, Gregory Caldeira, James Gibson, and others have theorized that courts' diffuse support is tied to their specific support. (27) The concepts are not perfectly correlated--indeed, Easton's contribution was to tease them apart--but "[o]ver the long-term, the two types of support should be related (and may converge)." (28)
A second value to which constitutional approval may be connected is constitutional compliance. When people mostly back their charters, they may be more prone to obey them, and to insist that their governments abide by them too. In previous work, one of us has documented the startling degree of noncompliance that characterizes many countries' constitutions. (29) Finding ways to improve enforcement is thus a priority--and an intuitive means to this end is persuading people to support their constitutions more strongly. Randy Barnett has made the point nicely: "[S]ome form of general acquiescence is necessary for any constitution to be implemented ...," (30) Analogously, in his work on "rights revolutions" that induce governments to respect constitutional rights, Charles Epp has suggested that they are most likely to succeed if there exists a "broad support structure in civil society." (31) This support structure, in turn, arises due to widespread public backing for the constitution. (32)
Third, constitutional approval could lessen the countermajoritarian difficulty: the worry that courts are behaving undemocratically when they strike down popularly enacted laws. (33) This worry is exacerbated when a constitution is disliked by the public; then courts not only nullify legislation that navigated the usual democratic channels, but do so in the name of a document whose standing is open to doubt. In contrast, if a constitution is broadly supported, then aggressive judicial activity on its behalf may be less troublesome. Then the activity may seem like the realization of the people's deepest public values, not their frustration. (34) A claim of this kind is at the heart of Bruce Ackerman's prominent theory of "constitutional moments." (35) The reason why these moments deserve to be judicially enforced is that they, not ordinary legislation, boast the closer connection to the people's true aspirations for their society.
Lastly, constitutional approval may promote constitutional durability. A popular constitution may be more likely to stand the test of time, to resist successfully efforts to replace it with another charter. This durability argument dates back to Madison, who remarked in The Federalist that, without "veneration" for the Constitution, "perhaps the wisest and freest governments would not possess the requisite stability." (36) It also has been advanced by contemporary scholars like Rosalind Dixon and Tom Ginsburg, who have speculated that "[t]he lower the level of popular support for a constitution, the more vulnerable a constitution will ... be to whole-scale replacement." (37) It is worth noting, though, that constitutional longevity is not an unalloyed good. Madison's great rival and friend, Jefferson, famously argued that the dead should not govern the living, and thus that no constitution should survive for more than a generation. (38)
To be clear, these are all hypotheses, not statements of fact. Constitutional approval may bring about greater legitimacy, compliance, longevity, and so on--or it may not. Later in this Part, (39) we survey the (quite limited) empirical evidence on its consequences. But for now, our point is only that many scholars believe, rightly or wrongly, that it yields significant benefits. This is enough, in our view, to justify our present examination. Next, we turn from the consequences of constitutional approval to the causes. They matter too because, depending on which are correct, very different prescriptions may follow for constitutional designers and governmental leaders.
First, people's demographic attributes may affect their support for the constitution. People who have prospered under its regime--whites, men, the wealthy, the well-educated--may back it more strongly than their less fortunate peers. Conversely, the disadvantaged may be more loyal to the constitution if they believe that it espouses a message of dignity and equality. In the words of Christine Kelleher and Jennifer Wolak, "It is ... possible that confidence in state institutions is driven more by individual demographic differences than contextual differences in state institutions." (40)
Second, constitutional approval may stem from constitutional knowledge. Those who are more informed about the constitution may be "exposed to a series of legitimizing messages focused on the symbols of justice," thus increasing their affection for it. (41) Maybe "to know it is to love it," in the pithy phrase of Caldeira and Gibson. (42) Or maybe not. Perhaps those who are more educated about the constitution also are more conscious of its deficiencies, of the ways it has failed to fulfill its promises. Then familiarity may breed contempt, not admiration.
Third, constitutional approval may reflect people's attitudes toward other institutions. For instance, those who are prouder of their state or country also may be more attached to the constitution that structures its public affairs. Analogously, if a particular party emphasizes devotion to the constitution, this party's members may profess higher levels of constitutional support. "[Citizens' basic affect toward governmental institutions will extend to their evaluations of the [constitution]," according to Robert Durr, Andrew Martin, and Christina Wolbrecht. (43) Fourth, people may support the constitution because its substantive content corresponds to their preferences. People may hold opinions as to which provisions should be included in (and excluded from) the constitution, and they may back it to the extent it is congruent with their views. Balkin has made this argument succinctly (if disapprovingly). "Many people may be reasonably comfortable with the status quo .... For such people, constitutional faith is not particularly difficult....', (44) Their approval flows from their constitutional contentment.
And fifth, people may favor the constitution because of its key non-substantive features--its age, length, amendment frequency, ratification process, and so on. In Madison's view, "veneration" is a property that "time bestows on every thing;" (45) constitutional approval, that is, arises from constitutional longevity. Similarly, Ginsburg and his coauthors have noted the common "claim that participatory design processes," in which people are involved in charters' drafting and approval, "generate constitutions with higher levels of ... popular support." (46) Furthermore, people may prefer a longer constitution because it is able to address more issues they care about in greater detail. (47) Or their taste may run to a more easily amendable charter because it can adapt more readily to changing societal circumstances. (48)
Once again, these are only hypotheses. Below, we summarize the existing empirical work on why people approve of constitutions and other governmental institutions. It also is true that many more explanations for constitutional support could be posited: judicial decisions, economic trends, elite opinions, mass polarization, etc. But we are confident we have identified the main causal claims in the theoretical literature (and the ones most relevant to constitutions specifically rather than institutions generally). (49) These claims therefore occupy much of our attention in the rest of the Article.
Unfortunately, almost nothing is known about the specific subject of constitutional approval. Major American surveys (such as the American National Election Studies, the General Social Survey, and the National Annenberg Election Survey) do not ask about it. (50) Nor do most important foreign surveys (such as the Eurobarometer, the Latinobarometer, and the World Values Survey). In fact, the only poll we have found that (sometimes) includes a germane question is the Afrobarometer. In three of its five rounds, it asked respondents whether they agree or disagree with the statement, "Our constitution expresses the values and hopes of the [country's] people." (51) As far as we can tell, only one academic paper, by Devra Moehler and Staffan Lindberg, has taken advantage of the resulting data. (52) Moehler also has surveyed Ugandans on whether they support their most recent constitution. (53)
Fortunately, this is not the end of the story. Many American polls routinely ask people about their approval of other governmental institutions: the Supreme Court, Congress, the President, state governments, and so on. Many scholars also rigorously investigate the consequences and causes of approval of these bodies. Their findings are not directly applicable to this project, so we do not dwell on them at great length. But they do illuminate many of the factors that might be linked to constitutional approval, thus setting the stage for our own examination.
Beginning with the consequences of institutional approval, only one of them, institutional legitimacy, has been assessed empirically. Most of the relevant work has concluded that specific support for courts is a statistically significant predictor of diffuse support for them. (54) A noteworthy study by Vanessa Baird, Caldeira, and Gibson, for instance, found that specific support is related to diffuse support for the national high court in nineteen out of twenty countries (Russia being the lone exception). (55) Similarly, the most recent article on the topic, by Gibson and Michael Nelson, showed that performance satisfaction is tied to institutional support for the U.S. Supreme Court even controlling for a host of other variables. (56) These results validate the predictions of Easton's theoretical model. As the model anticipates, diffuse support arises--in part but not exclusively--from specific support. (57)
Turning to the causes of institutional approval, people's demographic attributes, first, are not especially influential. Moehler found that support for the Ugandan constitution does not vary significantly by respondents' gender, age, education, or wealth. (58) Caldeira and Gibson detected only "[t]rivial bivariate correlations ... between Court attitudes and gender and age," though African Americans are less likely to back the Court. (59) According to David Jones and Monika McDermott, "definitive evidence remains elusive" as to "whether or not socioeconomic status affects public approval of Congress." (60) And the coefficients for gender, age, and education are insignificant in most of Kelleher and Wolak's models of state governmental approval, though blacks again are less supportive of all three branches. (61) At least in this area, demography does not seem to be destiny (except possibly for African Americans).
Second, the impact of knowledge about the institution varies by the body at issue. As to the Court, Caldeira, Gibson, and others have found consistently that people who are more informed about it also support it more strongly. (62) "[A] considerable body of earlier research" establishes that "as knowledge of the Supreme Court increases, so too does loyalty toward the institution." (63) But as to Congress, knowledge is related negatively to approval. John Hibbing and Elizabeth Theiss-Morse, (64) Jones and McDermott, (65) and Jeffery Mondak and his coauthors, (66) all have determined that "Americans who know Congress the best like it the least." (67) (Though Mondak and his coauthors have explained that this may be because high-knowledge and low-knowledge people assess Congress differently, not because knowledge directly affects approval. (68))
Third, people's attitudes toward other institutions, and toward government generally, usually influence their opinion of any particular institution. Ugandans who support their country's ruling party tend to back their constitution. (69) Americans who think government takes their views into account, and who adhere to broad democratic values, approve of the Court at higher rates. (70) And respondents who trust government to do the right thing are more likely to evaluate Congress's performance positively. (71) However, the evidence is mixed as to whether support for one governmental branch is linked to support for the other branches. Some studies find that it is, (72) while others conclude to the contrary. (73)
Fourth, the literature is nearly unanimous that people approve more strongly of bodies whose policy outputs are more congruent with their preferences. As to the Court, Durr, Martin, and Wolbrecht found that the more its decisions diverge ideologically from the public's views, the less the public supports it. (74) Likewise, Brandon Bartels and Christopher Johnston showed that respondents' ideological disagreement with the Court is linked to reduced backing for it. (75) And as to Congress, Jones and McDermott, (76) Mondak and his coauthors, (77) and Mark Ramirez (78) determined that its approval declines, respectively, as respondents' ideological distance from the majority party grows, as people perceive that their views are worse represented, and as it deviates further from the public mood. This connection between approval and policy congruence is the closest this body of scholarship comes to a consensus.
And fifth, relatively little is known about what we earlier called the non-substantive features of institutions. In her Ugandan study, Moehler found that more extensive participation in the process of constitutional ratification did not result in greater backing for the document. (79) In their analysis of national high courts, Baird, Caldeira, and Gibson showed that "[t]here is a very strong relationship ... between the age of the court and the level of satisfaction with its outputs." (80) And in their work on American state governments, Kelleher and Wolak determined that legislative professionalism and the voter initiative reduce legislative support while term limits increase it, that the gubernatorial recall has no effect on gubernatorial support, and that the type of judicial election is unrelated to judicial support. (81)
Much more could be said about the literature on institutional approval, which is impressively rich and varied. But for present purposes, there are two essential points. First, the literature barely addresses constitutional approval--and overlooks U.S. constitutions entirely. Our poll is the very first to ask Americans how strongly they back their state and federal charters. And second, the literature does suggest an array of consequences and causes of constitutional approval. The potentially significant consequences are why we think support for the constitution is worth studying, while the hypothesized causes guide much of our analysis of the concept. It is to this analysis that we now turn.
Our main tool for exploring whether and why people back their constitutions is a nationally representative survey that we carried out in October 2014. The survey asked Americans about their support for their federal and state charters, as well as a host of other issues that might influence approval levels. The survey also focused on state constitutions because their considerable variation makes them an ideal laboratory for studying the impact of constitutional design. While there is only one federal Constitution, state constitutions diverge widely in their substantive content, in their non-substantive features, and in the populations they aim to govern. They thus enable us to test many more hypotheses about constitutional approval than does the federal Constitution alone.
Another advantage of our emphasis on state constitutions is that our results for them may be more generalizable to constitutions around the globe. It is true that subnational constitutions tend to be more obscure than their national counterparts, (82) and that they do not need to address certain issues covered by the higher-level charters. (83) However, as one of us has recently shown, American state constitutions actually are more similar to foreign countries' constitutions than is the U.S. federal Constitution. (84) Like most foreign constitutions, state charters tend to be long and detailed, to grant plenary rather than limited powers, to be amended or replaced frequently, and to be fairly unfamiliar to their publics. (85) Judged by these characteristics, it is the U.S. federal Constitution that is the true outlier on the international stage. (86) So while we study its backing as well, it is our state-level findings that may be more applicable to constitutions worldwide.
A. Survey Design
To determine the levels and causes of constitutional approval, we designed and then administered a nationally representative online survey. (87) The survey was conducted by Survey Sampling International (SSI), a firm that specializes in online polling research. SSI distributed our survey to a panel of respondents that was nationally representative in terms of gender, age, race/ethnicity, and census region (Northeast, Midwest, South, and West). (88) To build the panel, SSI used relationships with partnership organizations through which respondents had agreed to participate in online polls. For example, some respondents signed up through United Airlines, and were rewarded with frequent flyer miles. Others signed up through iPad applications, and were rewarded with iTunes dollars. While the rewards varied, all respondents received compensation of about fifty cents per five minutes of survey time. The survey was online for two weeks in October 2014. In total, 2215 people took the poll, which is a large enough sample for us to draw inferences about the national population as a whole. (89)
The survey is included in its entirety as Appendix A, so rather than reproduce all of the questions here, we direct interested readers to the end of the Article. The survey began by asking respondents to identify basic information about themselves, such as their gender, age, race/ethnicity, (90) state of residence, education level, (91) and household income. (92) After soliciting this demographic data, the survey provided a short introduction to state law, state constitutions, and the federal Constitution. The purpose of this introduction, which is excerpted below, was to better acquaint respondents with the documents they would then be asked about:
As you may know, each state has its own constitution, which takes precedence over other kinds of state law such as statutes and regulations. If ordinary state law conflicts with the state constitution, it is the state constitution that has to be followed. State constitutions cannot contradict the federal United States Constitution, but they can provide additional protections and cover additional areas. State legislators often face a choice between including policies in the state constitution or in ordinary state law. There are several differences between these options. First, when policies are placed n the state constitution, they are harder to change in the future. Amending a state constitution is always more difficult than amending a regular state law. Second, policies that are in the state constitution are often considered more "fundamental" than policies in ordinary state law. States commonly include policies that they see as especially important in the state constitution. And third, policies that are in the state constitution can be used by courts to invalidate policies that are in ordinary state law. In other words, if ordinary state law violates the state constitution, the ordinary state law must be struck down.
Following this passage, the survey presented respondents with a list of twenty-nine substantive policies along with brief explanations of what their adoption would entail. For example, the "right to unionize" "would allow workers to join unions even when their employers object to their membership." The "obligation to establish a state university" "would create an obligation for the state to fund a state university that is available to admitted residents at a subsidized rate." The "right to gender equality" "would ensure that women are treated as equal to men by the state." And the "prohibition of the death penalty" "would ensure that the death penalty is never imposed, even for the worst crimes." Again, the full list of policies and explanations is available in Appendix A.
We formed this list by perusing the texts of current state constitutions in search of provisions that (1) represent substantive policy choices; and (2) are found in multiple state charters but not in the federal Constitution. On the one hand, we wanted to identify provisions that are actually plausible elements of state constitutions. On the other, we did not want to include provisions that are also present in the federal Constitution, since their greater familiarity could induce respondents to support including them in state constitutions irrespective of their merits. Based on these guidelines, we compiled policies in the areas of employment, education, welfare, marriage, criminal justice, the environment, and several others. We also had two experts on state constitutional law inspect our list, and are grateful for their feedback. (93)
For all of these policies, the survey asked respondents whether they would like to see them included in their state's constitution. The answers to these questions are our core measure of people's substantive constitutional preferences. However, one potential concern with this approach is that people might conflate their constitutional with their ordinary legal preferences. In other words, they might respond based on whether they approve of the policy generally, not whether they want it enshrined in their constitution specifically. (94) To mitigate this risk, the survey first asked respondents, for the same twenty-nine provisions, whether they would like to see them included in their state's regular laws. Only after respondents answered these questions were they asked whether they would like to see the provisions incorporated into their state's constitution. The aim here was to encourage people to separate their constitutional from their ordinary legal preferences, and to take into account the ways in which constitutions differ from conventional legislation.
The survey next asked respondents to assess, on a scale from one to ten, how strongly they approve of the federal Constitution and of their state's constitution. (The survey also asked about support for state law, again in order to prompt people to distinguish between their constitutional and non-constitutional attitudes.) These questions capture constitutional (and statutory) backing, and generate the dependent variables for all of our models. Their wording also is essentially identical to prior polls of other institutions' approval, thus increasing our confidence in the questions' validity and facilitating inter-institutional comparisons. (95) However, unlike those other polls, these questions have never been posed before, and so their answers are of particular interest.
The survey further included items that inquired about respondents' constitutional and civic knowledge, as well as their partisanship and patriotism. These items all correspond to additional hypotheses about the sources of constitutional approval, and they were drafted as follows: Three questions asked respondents to rate their (self-professed) familiarity with the federal Constitution, their state constitution, and their state's laws on a scale from one to five. Two questions asked respondents how closely they (claim to) follow the national and local news, with possible responses ranging from "not closely at all" to "very closely." Two more questions asked respondents how proud they are, on a ten-point scale, to live in the United States and in their particular state. And a final question asked respondents about the political party to which they belong (Democratic, Republican, independent, or other).
The survey ended with two quizzes that tried to test whether respondents read our explanations carefully and understood our questions. Specifically, we asked (1) whether state constitutions are easier or harder to amend than ordinary state law; and (2) whether state constitutions are more or less fundamental than ordinary state law. Respondents who gave wrong answers to both questions were removed from our sample. (96)
B. Congruence Scores
By themselves, the survey responses allow us to evaluate some but not all of our hypotheses about constitutional approval. One claim we cannot assess on this basis alone is the proposition that people prefer constitutions whose substantive content more accurately reflects their preferences. To test this hypothesis, we need a measure of congruence that compares people's constitutional views with the documents' operative provisions, and then determines how close the fit is.
The survey itself captured respondents' preferences because it asked them, for twenty-nine policies, whether they would like to see the measures included in their state's constitution. The missing piece is thus the actual content of each charter, the provisions it in fact happens to enshrine. To obtain this information, two research assistants coded all fifty state constitutions and recorded whether each of the twenty-nine policies is present in each document. The assistants agreed in their judgments in the vast majority of cases, and all discrepancies were resolved by the authors. (97)
Our basic coding rule was that, for a provision to count as included, the constitution must explicitly require the state to carry out the policy. When the constitution merely provides that the state "may" do something, or "shall have the power" to do something, we did not mark the policy as present. (98) Our coding also was based only on the state constitution's text, and did not take into account judicial interpretations of the language. We believe this approach is appropriate because, unlike the federal Constitution, state constitutions are long and detailed documents that are subject to frequent revision. As a result, there are fewer opportunities for courts to interpret many of their provisions. (99)
We calculated respondents' congruence scores by pairing their constitutional preferences with our coding of their states' constitutions. That is, each score represents the proportion of a respondent's preferred policies that are actually found in her state's constitution. We consider there to be congruence both when a respondent supports a given policy and this policy is included in her constitution, and when a respondent opposes a policy and it is not included. The resulting scores thus have a theoretical range of zero (none of the respondent's preferences are incorporated) to one (all of the respondent's preferences are enshrined).
C. Non-Substantive Features
A final hypothesis the survey itself cannot address is the possibility that constitutions' non-substantive features--in particular, their age, length, and amendment frequency--influence constitutional approval. We therefore collected data on these characteristics from a number of sources. Westlaw and the Green Papers, among other resources, list when all state constitutions were adopted; with this information it is trivial to calculate each charter's current age. (100) One of us previously determined the total number of words in each state constitution, and we reuse those figures here. (101) Lastly, The Book of the States documents the total number of amendments to each state constitution since its adoption. (102)
III. Descriptive Exploration
Having perhaps taxed our readers' patience with this long buildup, we are now in a position to present our findings on constitutional approval. Our presentation in this Part is mostly descriptive; we first identify the levels of federal and state constitutional support, and then explore how they vary along several dimensions. In the next Part we turn from description to explanation. We build regression models in stages for federal and state constitutional backing, thus illuminating several of the factors responsible for them. In both Parts, we consider the same five sets of hypotheses in the same order, involving (1) demographic attributes, (2) civic knowledge, (3) institutional attitudes, (4) constitutional congruence, and (5) non-substantive constitutional features.
A. Overall Levels
We begin with the overall levels and distributions of constitutional approval. By and large, Americans strongly back their federal Constitution. Its average approval score is 7.8 out of 10, and its median score is even higher at 9. As the dotted density curve in Figure 1 indicates, a full 20% of respondents give it the maximum approval score of 10, while only about 5% rate it below 5. In partial contrast, the average approval score for state constitutions is 6.7 out of 10, and the median score is 8--noticeably, though not dramatically, lower. As the solid density curve in Figure 1 shows, only about 8% of respondents award their state charter the maximum score of 10, while more than 10% rank it below 5. It thus is fair to conclude that state constitutions are somewhat less popular than their federal counterpart. (Though popularity, of course, is not the same as merit. As we observed earlier, it is possible for constitutions to be too admired given their actual design and performance. (103))
B. Geography and Demography
We next consider how constitutional approval varies geographically and demographically. The two maps in Figure 2 depict the average approval scores in each state for the federal Constitution and the state constitution. They reveal that, in most states, support is in line with the national average, but that there are a few modest exceptions. Specifically, average federal constitutional approval falls below 7.5 in Massachusetts, New York, and Vermont (all liberal northeastern states), while it exceeds 7.8 in Alabama, Idaho, Mississippi, Montana, Nebraska, Oklahoma, South Dakota, and Utah (all conservative southern and western states). Similarly average state constitutional approval is less than 6.6 in Alabama, Georgia, Mississippi, South Carolina, and Utah (mostly conservative southern states), and more than 6.8 in Arizona, Kansas, Nebraska, New Hampshire, North Dakota, South Dakota, West Virginia, and Wyoming (mostly conservative western states).
We also note that these state-level estimates were not produced through crude disaggregation (that is, simply calculating averages for each state's respondents), but rather through a more sophisticated technique known as multilevel regression and poststratification (MRP), which we describe briefly in the margin. (104) MRP is considerably more accurate than disaggregation when (as here) the samples in each state are relatively small. In these circumstances, though, MRP also tends to produce more tightly bounded estimates because it uses information from the entire country to estimate public opinion in each state. This feature of MRP explains why the states' constitutional approval scores are all relatively close to one another. The limited state-specific information is outweighed by the greater volume of national data.
Shifting to demographics, Figure 3 shows how constitutional approval differs by race, gender, education, income, and age. The dots in each chart represent the means for the various groups, while the vertical lines represent the 95% confidence intervals around the means. When the confidence intervals for groups do not overlap, the differences between the groups' means are statistically significant. (105)
As to race, African Americans' approval scores are lower than those of Caucasians and other groups (Asian Americans, Pacific Islanders, and Native Americans). (106) As to gender, men's approval scores are higher than women's. As to education, the approval scores of respondents with at least four years of college are higher than those of people with less schooling. As to income, the approval scores of respondents whose households make at least $50,000 per year are higher than those of less well-compensated people. And as to age, the approval scores of respondents over thirty-five are higher than those of their younger peers. In many cases, however, these differences are relatively small and, as we show in the next Part, not statistically significant once other variables are incorporated into the analysis.
C. Civic Knowledge
Proceeding to respondents' civic knowledge, Figure 4 plots average approval scores for different levels of federal and state constitutional familiarity, along with their 95% confidence intervals. It indicates that those who (think they) know their charters better also back them more strongly. In particular, respondents who rate their knowledge of the federal Constitution as 1 out of 5 have an average approval score of 4.1, while those who claim their knowledge is a perfect 5 have an average score of 9.0. Likewise, respondents who assess their knowledge of their state constitution as 1 have an average approval score of 4.5, while those who assert maximum knowledge have an average score of 8.7.
While these results suggest that constitutional familiarity and approval are intertwined, some caution is in order. Our questions did not test respondents' actual knowledge of either the federal Constitution or their state charters. It is possible that the relationship between genuine knowledge and approval is quite different. It is also possible that respondents believe they are familiar with their constitutions because they support them, or that both support and professed familiarity stem from the same general positive attitude toward the documents. Unfortunately, our research design does not allow us to probe further these psychological aspects of respondents' answers. (107)
However, our survey responses do allow us to say a bit more about individuals who claim intimate familiarity with their constitutions. Respondents who rate their knowledge of their state constitution as 4 or 5 are better educated than other subjects (45% versus 37% with at least four years of college), younger (median age of 42 versus 49), wealthier (58% versus 46% with household income above $50,000), more liberal (38% versus 31% with left-of-center views), more likely to be Democrats (47% versus 39%), and more likely to be male (57% versus 45%). (108) There are similar differences between respondents who rank their knowledge of the federal Constitution as 4 or 5 and other subjects. (109) Also notably, purportedly higher-knowledge respondents do not score any better on our quizzes than their ostensibly lower-knowledge peers. In fact, they score worse in one case, with 42% of higher-state-knowledge subjects wrongly stating that state constitutions are easier to amend than ordinary state law versus 20% of lower-state-knowledge respondents. (110) These figures hint (but do not prove) that a gap may exist between professed and actual understanding of the constitution. They thus confirm the need to take our findings about constitutional knowledge with a grain of salt--and to study the concept further in the future.
Constitutional knowledge, though, is not the only form of familiarity that appears to be related to constitutional approval. The same is true for knowledge of public affairs more generally. Figure 5 displays average federal and state constitutional approval scores, along with their 95% confidence intervals, for different levels of familiarity with the national and local news, respectively. Possible responses range from following the news "very closely" to "not closely at all." At the federal level, those most attentive to the news have an average approval score of 8.4, while those least attentive have an average score of 5.8. The same pattern holds at the state level: the most avid local newshounds have an average approval score of 7.3, compared to 5.2 for those who do not follow the local news at all. These findings further suggest that knowledge is connected to approval--albeit with the same caveat as before about actual and claimed observation of the news not being the same.
D. Institutional Attitudes
The final characteristics of respondents themselves (as opposed to their constitutions) for which we have information are their general institutional attitudes: that is, their feelings toward their country, state, and party. Figure 6 plots average federal and state constitutional approval scores, along with their 95% confidence intervals, for different levels of national and state pride, respectively. It shows that approval and jurisdictional pride are closely correlated. At the federal level, people who rate their national pride as 1 out of 10 have an average approval score of 3.8, while those who are maximally proud of their country have an average score of 8.5. Similarly, at the state level, people who are least proud of their state have an average approval score of 3.3, while those who are most proud have an average score of 8.2. Jurisdictional pride thus seems as strongly related to constitutional approval as either of the knowledge variables we considered above.
Our results for party affiliation are also notable, though not quite as stark. Figure 7 displays average federal and state constitutional approval scores, along with their 95% confidence intervals, for Democrats, Republicans, and independents. At both levels, Republicans are stauncher constitutional supporters than other parties' members. Their average federal approval score is 8.1, compared to 7.8 for Democrats and 7.5 for independents. Likewise, their average state approval score is 7.1, compared to 6.8 for Democrats and 6.4 for independents. (111) These differences are larger than most of the demographic gaps we identified earlier. But they are substantially smaller than the variations by civic knowledge and institutional pride.
E. Constitutional Congruence
We now turn to factors that involve not just respondents' own attributes but also those of their constitutions. Again, these factors are available solely at the state level since, at the federal level, there is only one constitution and so no possibility of comparison. Starting with constitutional congruence, we explained above that we calculated it by comparing respondents' preferences on twenty-nine constitutional policies with the provisions actually present in their states' charters. (112) The resulting congruence scores range from 0.14 (4 out of 29 preferred policies included in the constitution) to 0.97 (28 out of 29 preferred policies included). (113) This wide spectrum is itself quite interesting, as it indicates that state constitutional law differs greatly in its fit with people's views.
Figure 8 plots average state constitutional approval scores, along with their 95% confidence intervals, for each level of congruence (displayed here as the raw number of each respondent's preferred policies that are constitutionally enshrined, and varying from 4/29 to 28/29). Strikingly, and unlike the results in Figures 5 to 7, Figure 8 does not reveal an ascending pattern with approval positively related to congruence. Instead, many respondents whose constitutional preferences are barely satisfied still rate their charter favorably, while many respondents whose constitutional views are largely heeded still do not support their charter very strongly. For most levels of congruence, the average level of approval stays roughly constant and seems impervious to variations in fit. While this analysis does not control for other possible causes of constitutional approval, it does hint that congruence may not be as closely tied to it as the theoretical literature asserts.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Abstract through III. Descriptive Exploration E. Constitutional Congruence, p. 113-152|
|Author:||Stephanopoulos, Nicholas O.; Versteeg, Mila|
|Publication:||Washington University Law Review|
|Date:||Dec 1, 2016|
|Previous Article:||Antitrust in zero-price markets: applications.|
|Next Article:||The contours of constitutional approval.|