Theory

Infrequent amendment of the US Constitution has prompted development of diverse interpretive philosophies as judges seek to apply a document from a foreign era to modern disputes. While the public may not fully appreciate the legal community’s nuanced distinctions between originalism and textualism (cf. Bork Reference Bork2009; Scalia Reference Scalia2018) or between living constitutionalism and purposivism (cf. Ackerman Reference Ackerman2006; Breyer Reference Breyer2006), the public does grasp the general distinction between originalism and living constitutionalism loosely defined (Greene, Persily, and Ansolabehere Reference Greene, Persily and Ansolabehere2011; Krewson and Owens Reference Krewson and Owens2021). As popularly understood – and, therefore, as used in this article – originalism assumes that the US Constitution’s writers chose their words carefully, placing the burden on the amendment process rather than on judges to update that language to meet modern needs; by contrast, living constitutionalism seeks each provision’s implicit purpose, then applies that purpose to modern circumstances consistent with the living generation’s sensibilities.Footnote ⁴

Though the public lacks detailed political knowledge (Converse Reference Converse and Apter1964), Americans do have general views about the US Constitution (Levinson Reference Levinson1988; Stephenopoulos and Versteeg Reference Stephenopoulos and Versteeg2016; Zink and Dawes Reference Zink and Dawes2016; Brown and Pope Reference Brown and Pope2018; Dawes and Zink Reference Dawes and Zink2021) and its proper interpretation (Greene, Persily, and Ansolabehere Reference Greene, Persily and Ansolabehere2011). Greene, Persily, and Ansolabehere (Reference Greene, Persily and Ansolabehere2011) find that public support for originalism is explained by variables which underlie justifications for originalism among jurists. For example, people who say judges should “strictly follow the law no matter what people in the country may want” are more likely to identify as originalist, even controlling for demographic and political characteristics. As they note, their findings “by and large conform to what we would expect of respondents who are engaging the questions in a serious and relatively informed manner… . [I]t parallels the divide within professional constitutional discourse among lawyers, judges, and academic commentators” (411, emphasis in original).Footnote ⁵

When people evaluate judges, only a judge’s ideology clearly explains more variation in support than whether a judge is originalist or espouses living constitutionalism. Judicial philosophy is more important than a judge’s personality, religion, age, race, or sex (Krewson and Owens Reference Krewson and Owens2021, Reference Krewson and Owens2022). Like most political matters, judicial philosophy correlates with partisanship: Republicans lean toward originalism while Democrats lean toward living constitutionalism (Krewson and Owens Reference Krewson and Owens2021).Footnote ⁶ Nevertheless, judicial philosophy has an independent effect on how people evaluate court decisions, even after accounting for partisanship and diverse other considerations (Krewson and Owens Reference Krewson and Owens2022; Rivero and Stone Reference Rivero and Stone2023).Footnote ⁷ Simply put, disputes between originalist and living interpretations of the US Constitution divide not only judges but the public, and the public’s philosophical differences appear to be at least somewhat sincere.

Meanwhile, two fundamental differences between state constitutions and the US Constitution mute interpretive disputes at the state level. First, and most importantly, state constitutions undergo such frequent amendment and even replacement that they tend to accommodate legislators’ statutory interests, reducing the interpretive burden on state judges (Brown Reference Brown2018; Dinan Reference Dinan2018; Marshfield Reference Marshfield2018). The median state constitution receives more amendments in 20 years than the US Constitution has had in 200, and every state constitution has been amended more recently than the US Constitution (Brown Reference Brown2023). Most states have replaced their constitution entirely – often more than once – so that the median state constitution is roughly half the age of the US Constitution.Footnote ⁸ Unlike the US Constitution, state constitutions directly address modern concerns – especially in states with the highest amendment rates, where the legislature’s failure to propose a modernizing amendment implies acceptance of the status quo (Kincaid Reference Kincaid1988, 13; but see Brady Reference Brady2021).

Second, when difficult state constitutional questions do arise, state judges have ready access to source materials shedding light on original intentions. Many state archives retain full transcripts of their constitutional conventions (Baum and Fritz Reference Baum and Fritz2000; Dinan Reference Dinan2006). By contrast, we have no transcript of the federal convention except Madison’s notes – apparently heavily massaged (Bilder Reference Bilder2015). In addition, the amendments that occur so frequently in the states leave a paper trail of legislative hearings, voter pamphlets, and other informational materials (Dinan Reference Dinan2018; see also Salvucci Reference Salvucci1997; Gilbert Reference Gilbert2012).

For these reasons, “the critique of federal original-intent jurisprudence does not apply … with equal force” to state jurisprudence (Tarr Reference Tarr1991, 851; see also Williams Reference Williams2009, 318–19). Instead, original intent jurisprudence reigns as “the primary canon of [state] constitutional interpretation,” applied consistently in nearly every state for over two centuries (Christiansen Reference Christiansen2017, 341). State constitutional change occurs through a “democratic, public, participatory, popular, and socially transcendental” process, legitimizing this reliance on original intent (Farinacci-Fernós Reference Farinacci-Fernós2020, 23).Footnote ⁹ These differences between state and federal jurisprudence demonstrate that interpretive philosophies reflect an interaction between an individual’s prior political attitudes and core constitutional features like the age or changeability of a constitution’s provisions.

This emphasis on interpretive attitudes risks losing the forest for the trees, however. Judicial philosophy is merely a means to an end, after all: the final judicial decision. Thus, taking the next logical step, we might expect the same conditions that mute interpretive disputes in the states – in short, the relative youthfulness of state constitutional provisions – to also mute concerns over judicial review more generally. State and federal judges have claimed the authority to strike down legislative acts as unconstitutional since the nation’s founding (Prakash and Yoo Reference Prakash and Yoo2003).Footnote ¹⁰ Judicial review is not definitionally anti-democratic, of course. Indeed, judicial review may enhance democracy by mitigating the crucial principle-agent problem of popular sovereignty, revealing government misbehavior, and coordinating the public’s ability to respond effectively (Law Reference Law2008). In practice, the US Supreme Court’s willingness to exercise this authority reflects extralegal factors like public support (Vanberg Reference Vanberg2001; Ura and Wohlfarth Reference Ura and Wohlfarth2010; but see Ura and Wohlfarth Reference Ura and Wohlfarth2022), particularly relative to the popularity of the president (Gardner and Thrower Reference Gardner and Thrower2023), and of Congress (Merrill, Conway, and Ura Reference Merrill, Conway and Ura2017; see also Bassok Reference Bassok2013). Similar findings exist at the state level (Kastellec Reference Kastellec2016; Langer Reference Langer2002), where we also observe that state supreme court invalidations are less frequent in states with higher amendment rates (Brown Reference Brown2018; Reference Brown2023; Marshfield Reference Marshfield2018). All this makes sense: since courts cannot enforce their rulings, judges who fear executive non-implementation (Hall Reference Hall2014; Gardner and Thrower Reference Gardner and Thrower2023) or retaliatory legislation (Clark Reference Clark2009) exhibit self-restraint.

Given that judicial review hinges critically on public support, it is striking how little we know about the origins of public support for judicial review. Traditionally, the primary concern regarding judicial review has been that unelected judges may substitute their will for that of elected bodies that can claim popular mandates (Bickel Reference Bickel1986; Rogers and Ura Reference Rogers and Ura2020). As it turns out, however, citizens are “no more likely to accept the use of judicial review by an appointed court than an elected court” (Crabtree and Nelson Reference Crabtree and Nelson2019, 287), casting significant doubt on this received wisdom. There must therefore be more that motivates American skepticism of judicial review.

This observation leads us back to the relative youthfulness of state constitutions. When, as in many states, judicial review is based on a more recently updated constitution, it strengthens the case that invalidation merely enforces a higher-order social contract rather than substituting the court’s will for the legislature’s. In our argument, whether judges are appointed or elected matters less than whether the documents they interpret bear the living generation’s fingerprints. Again, legal elites are aware of the differences between the US Constitution and state constitutions, leading to the dominance of original intent jurisprudence in state constitutional discourse and reducing controversy over the use of judicial review. Our core prediction is that common citizens will respond similarly as legal elites if only they are informed of how state constitutions differ, so that respondents presented with a younger constitutional provision will manifest more support for judicial review than respondents presented with an older one.Footnote ¹¹

To this point, we have spoken of constitutional youthfulness as an abstract theoretical concept. In practice, we have in mind the age of a specific constitutional provision. To a lesser extent, we also consider the overall age or amendment rate of a constitution as a whole. With these operationalizations in mind, we arrive at these (pre-registered) hypotheses:

As described more fully in the next section, we test these hypotheses with a vignette experiment focused on H1c and conjoint analysis that evaluates both hypotheses. With an eye on competing explanations, we test a few additional hypotheses, also pre-registered. To be clear, we have not developed a strong theoretical foundation for these ancillary hypotheses; we include them only to provide an explicit test of the most likely confounders. Recognizing, for example, that respondent support for judicial invalidation of a particular statute may hinge on the statute’s general popularity, we test this hypothesis:

We consider the possibility that respondent judicial philosophy is merely an instrumental means to a partisan or policy end by testing two additional hypotheses:

Design

In August 2022, we recruited a nationally representative sample of 1,978 American adults using Lucid (cf. Coppock and McClellan Reference Coppock and McClellan2019), removing those who failed attention checks.Footnote ¹² Complete question wording appears in a supplement. We first asked respondents about their agreement with seven policies, with four framed in a conservative direction, and three in a liberal direction. After several intervening questions, one of these policies was randomly selected to appear in our vignette experiment at the end of the survey. We next collected respondent demographics.

Prior to any manipulation, we then collected respondent views about the US Constitution. To familiarize them with the format used in our subsequent conjoint experiment, we displayed a table showing six facts about the US Constitution: the year it was written (1787), the year of its last amendment (1992), its length (fifteen pages), its number of amendments (twenty-seven), its focus (structuring government), and the country’s partisan lean (competitive).Footnote ¹³ Using a seven-point scale, we asked respondents about their agreement with two separate statements about interpretation of the US Constitution: “Judges should interpret the United States Constitution as generally understood at the time the constitutional provisions were written” (to measure support for originalism), and “Judges should interpret the United States Constitution as generally understood according to current views and sensibilities” (to measure support for living constitutionalism). To measure support for invalidation we asked, “How much would you approve of judges using the United States Constitution to strike down state laws?” Finally, to gauge views of the US Constitution, we measured respondent agreement with four statements drawn from past research (Brown and Pope Reference Brown and Pope2018; Dawes and Zink Reference Dawes and Zink2021): “The creators of the US Constitution were looking out for themselves,” “The US Constitution should be amended more often,” “The US Constitution is outdated,” and “The US Constitution was written by visionary people,” each on a five-point scale.

Respondents then read these instructions: “As you may know, every state has its own constitution. Some states have replaced their constitution from time to time, while others have kept a single constitution since statehood. State constitutions vary in their length and their amendment rate. They also vary in whether they focus on erecting government structures rather than establishing specific laws and policies. On the next few screens, we will briefly describe a few hypothetical state constitutions and ask for your reaction. You will consider 4 different state constitutions.” We then presented a conjoint experiment, iterated four times (Hainmueller et al. Reference Hainmueller, Hopkins and Yamamoto2014). Each time, respondents saw a table like the one used earlier, but with an additional column providing information about a hypothetical state constitution. An example appears in Figure 1.

Figure 1. Illustration from Instrument.

Values for each hypothetical state constitution were drawn randomly from a realistic range reflecting actual state constitutions. “Year constitution established” was one of 1780, 1868, 1890, 1949, or 1986; “year of last amendment” was 2000, 2005, 2010, 2015, or 2020; “length of constitution” was 16, 30, 50, 85, or 142 pages; “frequency of amendment” was 16, 30, 50, 85, or 142 times; “focus of constitution” was “structuring government” or “making policy”; and “partisan lean of state” was “Republican,” “Democratic,” or “Competitive.”Footnote ¹⁴ Below each table, respondents answered the same questions they answered previously about the US Constitution, but about the state constitution before them. We use this conjoint experiment to assess H1, H2, and H4.

After this conjoint experiment, we concluded with our vignette experiment: “We would like to ask you about a recent controversy in an American state. Last year, this state enacted a law [policy description piped in]. This law passed [easily, with over 70% of state lawmakers voting in favor OR narrowly, with barely 50% of state lawmakers voting in favor]. A few months ago, this state’s highest court struck down this law, saying that it violates a provision of the state constitution that was ratified in [1896 OR 2016]. In general, is it appropriate for a court to strike down a law in this sort of situation?” The policy description piped into the vignette aligned with one of the policies we asked about at the beginning of the instrument, enabling us to consider the respondent’s agreement or disagreement with the policy in question. Randomly selecting one of several possible policies should cause idiosyncrasies associated with any particular policy to average out. Moreover, placing the vignette experiment after the four iterations of the conjoint analysis meant that respondents first had the opportunity to gain some sense for how state constitutions vary along these dimensions. We use this vignette experiment to assess H1, H3, and H5.

Results of vignette experiment

Our vignette experiment considers the age of a specific constitutional provision, creating a direct but challenging test of our theory that a constitution’s youthfulness can reduce concerns about judicial review. We asked respondents to evaluate a state court’s invalidation of a specific state statute (from 1, “very inappropriate,” to 4, “very appropriate”) after presenting a vignette varying along the following dimensions. First, whether the statute passed easily (coded 0) or narrowly (coded 1) through the state legislature. Second, whether the state court relied on a constitutional provision from 2016 (coded 0) or from 1896 (coded 1). And third, the respondent’s previously measured agreement with the particular policy that was piped into the questionnaire (from 1, “strongly oppose,” to 6, “strongly support”).

Ordinary least squares regression confirms – unsurprisingly, and consistent with H5 – that respondents show more support for judicial invalidation of a policy they personally oppose (p < 0.01). More to the point, we also observe that respondents are more favorable toward judicial invalidation – even invalidating a policy they support – when the court relies on a more recently adopted provision (p < 0.01), consistent with H1.Footnote ¹⁵ Incidentally, we find no effect of legislative divisiveness, contrary to H3 (p = 0.64). Figure 2 plots predicted values and 95% confidence intervals over the observed range of policy agreement, with separate lines for newer (dark) and older (light) constitutional provisions. (Full results appear in the supplement, along with pre-registered alternative specifications such as ordered probit.)

Figure 2. Results of Vignette Experiment.

Changing from a new to an old constitutional provision has the same estimated effect as moving 1.26 points along the 6-point policy agreement scale – substantively, a very large estimated effect that suggests Americans may react strongly to the age of the specific constitutional provision underlying a judicial invalidation. Judicial review apparently has greater legitimacy when rooted in a recently adopted – and thus more democratic – constitutional provision.

Results of conjoint analysis

We now turn to our more exploratory conjoint analysis. Unlike the vignette experiment, the conjoint analysis examines features of state constitutions as a whole rather than features of a specific constitutional provision. This generality leads us to expect somewhat weaker effects. This part of our analysis considers two dependent variables: respondents’ support for judicial review (to test H1) and respondents’ support for original intent interpretation relative to a living interpretation (to test H2).

We begin with descriptive information about these dependent variables, beginning with our measure of support for judicial review. Using a 5-point scale from “strongly disapprove” to “strongly approve,” we asked respondents whether they would approve of a court using a particular constitution to strike down a state law. When evaluating the US Constitution, 30% of Democrats and 28% of Republicans disapproved of judicial invalidation, while 34% of Democrats and 41% of Republicans approved – relatively minor partisan differences. When evaluating state constitutions (averaging across the four iterations of the conjoint analysis), 29% of Democrats and 30% of Republicans disapproved of judicial invalidation, while 45% of both Democrats and Republicans approved.Footnote ¹⁶ Thus, at a broad level, we see more acceptance of judicial invalidation in the context of state constitutions, among both Republican and Democratic respondents.

For our other dependent variable, we constructed a measure of respondent judicial philosophy from two questions. For the US Constitution and for each state constitution in the conjoint analysis, we included one item gauging support for an originalist interpretation and another gauging support for a living interpretation. With responses to both items spanning a 7-point scale, subtracting the former from the latter yields a 13-point net score, with positive net scores indicating greater support for a living approach, and negative scores indicating greater support for an originalist approach. For the US Constitution, average scores on this 13-point scale land at 1.63 for Democrats and −1.12 for Republicans; more intuitively, 57% of Democrats favored a living interpretation (with 27% neutral and 16% favoring an originalist interpretation), and 48% of Republicans favored originalist interpretation (with 27% neutral and 25% favoring a living interpretation). Averaged across the four iterations of our conjoint analysis of state constitutions, average scores were 0.76 for Democrats and −0.73 for Republicans; 52% of Democrats favored a living interpretation (with 24% neutral and 23% favoring originalist interpretation), and 46% of Republicans favored an originalist interpretation (with 21% neutral and 33% favoring a living interpretation). These patterns are consistent with past research showing a strong correlation between judicial philosophy and partisanship (Krewson and Owens Reference Krewson and Owens2021).

Figure 3 plots the mean values and 95% confidence intervals for these dependent variables. On average, respondents showed greater support for judicial review when evaluating state constitutions in the four iterations of the conjoint analysis than when evaluating the US Constitution (p = 0.02 two-tailed).Footnote ¹⁷ Similarly, respondents showed reduced support for a living interpretation (relative to an originalist interpretation) when evaluating state constitutions than when evaluating the US Constitution (p < 0.01 two-tailed). These significant differences are broadly consistent with H1 and H2, since the average state constitution in the conjoint analysis was adopted more recently, amended more often, and amended more recently than the US Constitution. However, we cannot tell from this level of analysis which feature of state constitutions drove this result. We now turn to an analysis to uncover the various features of state constitutions potentially driving these results.

Figure 3. Overall Support for Dependent Variables, Comparing US Versus State Constitutions.

H1 predicts greater support for judicial invalidation when respondents are presented with a state constitution featuring higher amendment rates (H1a), more recent amendment (H1b), or more recent constitutional adoption (H1c). In Figure 4, horizontal lines depict 95% confidence intervals around marginal means, treating each component in the analysis as a categorical variable. (Because respondents participated in four iterations of the conjoint experiment, standard errors are clustered by respondent throughout.) We do not find support for H1a. To the contrary, support for invalidation appears superficially to fall rather than rise with increased amendment rates. However, the effect is driven by evaluations of a constitution with 142 amendments − the only category to differ significantly (p = 0.03) from respondents evaluating a constitution with sixteen amendments − and may therefore be a fluke. Nor do we find support for H1b; the marginal means vary little by year of last amendment, perhaps reflecting the narrow range of years that we tested.

Figure 4. Support for Judicial Invalidation (Estimated Marginal Means).

H1c is more interesting. Respondents show much greater support for invalidation when evaluating a twentieth century than nineteenth century state constitution. Relative to respondents evaluating an 1868 constitution, for example, support for judicial review was much higher among those evaluating a 1949 (p = 0.03) or 1986 (p < 0.01) constitution.Footnote ¹⁸ If we had not included a condition for constitutions adopted in 1780, we would have seen these results as consistent with H1c and concluded that support for invalidation rises with younger constitutions. As it turns out, however, respondents presented with a 1780 constitution supported judicial review as strongly as those presented with a twentieth century constitution. We suspect that 1780 may have unintentionally primed respondents to think about the US founding, however. There is precedent for this admittedly post-hoc hunch; Brown and Pope (Reference Brown and Pope2018) report that respondents’ views of the US Constitution have more to do with patriotic support for America’s founding narrative than for the Constitution itself, and we may have invoked a similar response by including this date. Regardless, when paired with the results of our vignette experiment, we can confidently say that respondents show greater support for judicial review when presented with a twentieth century than nineteenth century constitution.

Though we made no prediction about a relationship between partisanship and support for invalidation, we observe in passing that respondents show much greater support for judicial invalidation when their partisanship aligns with the state’s partisan lean. Apparently, respondents believe that judicial invalidation is more likely to involve a cooperative court reigning in a wayward legislature than an activist court preventing a popular legislature from following a voter mandate (Law Reference Law2008; Rogers and Ura Reference Rogers and Ura2020). These findings may be worthy of future study.

We turn now to respondents’ preferred judicial philosophy. Using the same approach as the previous figure, Figure 5 plots marginal means for our second dependent variable, where higher scores indicate support for a living interpretation and lower scores indicate support for originalism. The effect of a constitution’s age is much clearer here than in the previous figure. Respondents presented with a state constitution adopted in 1986 had an average score 0.30 points lower (p < 0.01) than respondents presented with a constitution adopted in 1780 − the largest effect by far of any comparison in Figure 5. More generally, we find a statistically significant linear relationship (p < 0.01) when constitution age is modeled as a continuous variable using ordinary least squares regression, controlling for the other components shown in Figure 5.Footnote ¹⁹ The younger the constitution, the more our respondents favor an originalist interpretation.

Figure 5. Support for Living or Original Interpretation (Estimated Marginal Means).

H2 predicts that respondents will show greater support for originalist interpretation when informed of a state constitution’s (a) higher amendment rate, (b) more recent amendment activity, or (c) relatively recent adoption. These significant results for constitution age are consistent with H2c. We do not observe meaningful effects for amendment rates (H2a) or amendment recency (H2b),Footnote ²⁰ nor do we observe effects for components inserted as controls − the constitution’s length or substantive focus. In an ancillary hypothesis (H4), we predicted greater support for originalism when the respondent’s partisanship aligned with the state’s lean, but we do not observe this relationship.Footnote ²¹

In summary, this conjoint analysis provides evidence that respondents favor a more originalist interpretation as a constitution grows younger. We also find that respondents show more support for judicial invalidation when presented with a twentieth century than a nineteenth century constitution, though a constitution from 1780 produces counterintuitive results. These findings are consistent with H1c and H2c and support our broader argument about the differences between the US Constitution and a typical state constitution We do not find support for our predictions about amendment frequency (H1a and H2a) or recency (H1b and H2b).