Accountability for Court Packing

Abstract

How does the public respond to court-packing attempts? Longstanding accounts of public support for courts suggest voters retaliate against incumbents who seek to manipulate well-respected courts. Yet incumbents might strategically frame their efforts in bureaucratic terms to minimize the public’s outcry or use court-packing proposals to activate a partisan base of support. Drawing on a series of survey experiments, we demonstrate that strategic politicians can minimize electoral backlash by couching court reform proposals in apolitical language, and institutional legitimacy’s shielding effect dissolves in the face of shared partisanship. These results shed new light on how ambitious politicians might avoid electoral consequences for efforts to bend the judiciary to their will.

Expanding the size of the judiciary and filling those seats with allies can be a useful tool in the arsenal of an ambitious, policy-minded politician. In the short-term, a favorable judiciary can legitimize the majority coalition’s chosen policies (Dahl 1957); over the long run, the judiciary can protect those policies against efforts by new majorities to alter or change them. Because judicial tenures often outlast the terms of elected incumbents, majority coalitions who pack courts may entrench their policy views in the judicial branch as a form of “insurance” that will pay political dividends, even after the coalition is out of power (Ginsburg 2003; Epperly 2019).

Despite the instrumental benefits of court-packing, prominent theories of interbranch relations and judicial politics suggest that attempts to expand the judiciary carry political risk: so long as the judicial branch enjoys widespread popular support, politicians who attempt to change its structure will face electoral reprisal (e.g., Caldeira and Gibson 1992; Vanberg 2015). Yet, such attempts are not rare. In the United States, legislatures in Arizona, Georgia, Florida, Iowa, and South Carolina have seriously considered (and, in some cases, successfully enacted) court-packing reforms over the past fifteen years (Levy 2020). Many of these reforms came and went with minimal public outcry.

What explains the lack of electoral reprisal in response to these and other court-packing proposals? One possible explanation lies in the strategic behavior of elites. Anticipating public backlash, an incumbent seeking to pack the judiciary may disguise their efforts in neutral, apolitical, or seemingly positive terms, framing the proposal as one that will increase efficiency rather than an attempt to sway policy outcomes. Indeed, although his famous fireside chat incorporated some discussion of the political aspects of his infamous court-packing plan, Franklin D. Roosevelt (1937) stated that the first goal of that plan was benign: “to make the administration of all Federal justice speedier and, therefore, less costly.” Roosevelt is not alone. Each of the recent proposals in the US states sought to dramatically reshape the ideological composition of high courts but were advanced under the guise of apolitical motivations, such as promising a more transparent, efficient, and expedient administration of justice (Raferty 2016; Ducey 2016; Rankin 2017). These experiences suggest that politicians may be able to minimize the costs they face from a skeptical public by couching court reforms in carefully devised rhetoric.

These prominent theories might also overstate the threat of court-packing because incumbents may view a possible benefit to advancing court-packing plans: they might activate a base of electoral support or carry with them minimal risk. Not only have court reforms been described as ideal position-taking activities (Clark 2009), others have shown that the public’s professed loyalty to its judiciary may be informed by partisanship (e.g., Bartels and Johnston 2020; Bartels, Horowitz, and Kramon 2021).¹ By this logic, proposals to stack the courts with like-minded judges might animate government supporters, especially those who understand the value of long-term policy entrenchment. Minimally, the public backlash to court-packing may well be muted when advanced by entrusted, copartisan politicians (Driscoll and Nelson, Forthcoming).

To this end, we theorize that the public’s response to court-packing attempts will be swayed by politicians’ justifications for the reform, as well as by partisanship and the legitimacy they ascribe to judicial institutions.² Drawing on three survey experiments, we consider the public’s responses to both purportedly bureaucratic and explicitly politicized efforts at court reform, taking inspiration from the sorts of justifications incumbents often use to defend these proposals. Further, we examine how support for these proposals varies according to respondents’ preexisting support for the judiciary, hypothesizing that the supposed shielding qualities of institutional legitimacy are absent in the presence of shared partisanship between citizens and legislators.

Although those before us have explored the relationship between judicial politicization and support for national courts (Baird and Gangl 2006; Gibson and Nelson 2017; Nelson and Gibson 2019), we are among the first to experimentally probe the possibility that proposals to politicize the judiciary through court-packing efforts will spur electoral retribution (or burnish electoral support). Further, most previous research focuses on the determinants of institutional legitimacy rather than its effects, and focus nearly exclusively on public attitudes regarding pinnacle courts. We advance this literature by treating diffuse support as an independent variable, examining how this variable affects citizens’ willingness to punish an incumbent for efforts to stack lower courts with ideological appointees.

Our research is one of a small number of studies that examine public support for lower federal courts. We build on a well-established body of scholarly research that centers on public support for the US Supreme Court and other pinnacle courts and apply those studies to the lower federal judiciary. To be sure, there are many reasons why applying findings from one level of court to another requires care: lower courts differ from apex courts on many dimensions, including their size, salience, jurisdiction, and caseload. Still, it seems better to us to use the literature on these apex courts as a starting point, even though theories about high courts might not apply perfectly to our research setting. In this way, we follow in the well-trodden path of other literature on lower courts, which must often rely on what measures are available, knowing they may not perfectly align with concepts of primary theoretical importance (e.g., Benesh 2006) or may measure important concepts but be limited in geographic or temporal scope (e.g., Achury et al. 2022). We hope future research on non-apex courts will remedy these deficiencies.

We contribute to our understanding of court reform by examining how politicians’ justifications for court-packing can sway the public’s response to these proposals; we are the first to test the possibility that politicians can blunt electoral backlash or animate followers simply by framing judicial reform proposals in benign and apolitical motives. Indeed, many efforts to undermine democracy and the institutional separation of powers are framed by reformers as an effort to improve democratic functioning by rooting out corruption or making the judicial system more efficient (Levitsky and Ziblatt 2018). If it is the case the public is convinced by these framing efforts, this may clear the way for institutional manipulation that would otherwise be met with public hostility.

We find some evidence that the public takes issue with efforts to politicize courts and will use the electoral connection to hold would-be court-packers into account: proposals couched in bureaucratic language are more palatable, on average, than those that are explicitly political in motive. But partisanship matters too: shared partisanship of the proposer and citizen colors support for the incumbent alongside the rationale for the proposal. Additionally, we demonstrate across multiple experiments that the supposed shielding effect of institutional legitimacy is weaker than previous accounts have suggested. Voters who are more strongly supportive of the judiciary increasingly withdraw support from an outpartisan incumbent who makes a court-packing proposal, but that relationship is nonexistent when a copartisan incumbent attempts the same court-packing plan.

This finding directly contradicts existing accounts that suggest that legitimacy uniformly shields institutions against fundamental alterations to their membership and structure (e.g., Caldeira and Gibson 1992; Vanberg 2015). Instead, we join a chorus of researchers who emphasize that, even in a country where the courts and justice system are generally well-regarded (Gibson, Caldeira, and Baird 1998; Gibson 2007), the public’s support for judicial institutions may have a strong partisan basis (Bartels, Horowitz, and Kramon 2021; Bartels and Kramon 2020). Further, undermining or coopting independent courts may be a winning strategy for incumbents, especially when an elite frames the proposal in mundane bureaucratic terms and has a strong base of copartisan support.

The electoral risks of court-packing

Existing scholarship suggests that the public might protect courts against court-packing attempts. Scores of theoretical models are based on the premise that incumbents who meddle with the institutional framework of legitimate judiciaries should expect to face electoral costs (e.g., Vanberg 2000, 2001; Stephenson 2004; Staton 2006; Helmke 2010a).³ However, while many have examined the determinants of this public support, far fewer studies examine the effects of this support, especially on citizens’ electoral behavior: the very relationship that animates these theoretical models (c.f., Driscoll and Nelson Forthcoming).

Additionally, many candidates campaign on the promise of expanding the size of the judiciary and appointing supporters to fill those new seats. In the 2020 US presidential election, high profile presidential hopefuls like Pete Buttigieg and Elizabeth Warren championed court-packing; the Democratic party’s official 2020 platform also endorsed “structural court reforms to increase transparency and accountability” (Democratic Party 2020, 58). This suggests that, while court-packing may carry with it some electoral or long-term institutional costs, the risk of electoral retaliation seems to be viewed by many incumbents as minimal.

Unfortunately, little research has sought to explain the electoral consequences of modern court-packing attempts. The most common way that scholars have sought to understand the public’s support for judicial reform has come in studies of judicial legitimacy, which treat support for court reform as an indication that an individual views an institution as lacking legitimacy (e.g., Gibson, Caldeira, and Spence 2003). While studies in this vein have examined support for policies like court elimination or judicial removal, support for court-packing has not traditionally been incorporated in studies of this concept. When scholars have examined support for contemporary court-packing, these efforts have mainly been observational studies focusing on the presence (or absence) of partisan or ideological gaps in support for these proposals (Liptak 2020).⁴

However, two strands of the legitimacy literature are particularly relevant. Hibbing and Theiss-Morse (1995) note that the judiciary enjoys more support than the elected branches of government because it is separate from the rough-and-tumble politicking that characterizes the Washington system. Studies suggest that the public is unsupportive of judicial actions that are viewed as strategic and more willing to withdraw institutional support from judicial institutions that appear to be ‘politicized’ or partisan (Baird and Gangl 2006; Gibson and Nelson 2017). Further, the effects of cues that the Supreme Court is politicized vary according to one’s support for the source (Nelson and Gibson 2019). Scholars have studied the effects of politicization cues on support for the judiciary, most notably in Nicholson and Howard’s (2003) study of framing effects after Bush v. Gore; their study echoes the harmful effects of politicization for diffuse support for the Court, finding that framing that controversial decision in terms of expressly political terms – trying to end a presidential election – is associated with a decrease in diffuse support.

A second stream of contemporary research suggests that diffuse support for courts may have a partisan tinge. The best evidence on this front comes from Bartels, Horowitz, and Kramon (2021) who find that support for judicial power – a concept linked closely to judicial legitimacy – fluctuated after the Kenyan Supreme Court annulled an incumbent president’s electoral victory, only to then uphold his election in a new electoral process (see also Bartels and Kramon 2020). This evidence – based on results from a panel survey – has the unique ability to track partisan-based changes in support for judicial power within the same individuals over time, and in response to a dramatic reversal in the court’s position vis-a-vis the incumbent government.

This data from Kenya diverges from most models of diffuse support for the US Supreme Court, which find no differences in support on the basis of respondents’ partisan attachments. This is true both in papers that argue that judicial legitimacy has a strong instrumental basis (Bartels and Johnston 2013), and those that contend there is no instrumental basis to diffuse support for the Court (Gibson 2007; Gibson and Nelson 2014). What unifies all of these studies, however, is their focus on explaining the role of partisanship as a correlate of diffuse support. Existing research pays less attention to the partisan consequences of judicial legitimacy: that is, does judicial legitimacy provide a protective effect such that voters will punish incumbents for court-packing attempts? Does judicial legitimacy provide the same protective effect for both copartisan and outpartisan court-packing incumbents? It is no doubt important to understand who supports courts; but it is also essential to understand the extent to which those abstract commitments have real-world behavioral consequences (Driscoll and Nelson Forthcoming).

Where scholars have studied court-packing, they have focused on support for the proposal; we, by contrast, focus on support for the proposer. The study most closely related to ours is Albertus and Grossman (2021), which examines support for court-packing in the United States, Argentina, Mexico, and Brazil. Using a survey experiment, the authors conclude that voters are more likely to support reforms to expand the judiciary when they are advanced by copartisans, although voters do not react differently to court-packing attempts based on how the proposal is justified. We, by contrast, are interested in the electoral consequences of these proposals for the incumbents who seek to enact them.

To summarize, many studies have suggested that voters might punish incumbents who seek to pack the judiciary, especially if it is done for expressly political purposes. However, tests of this logic are scarce. And, because politicians might be strategic in their framing choices and support for courts might have some partisan basis, the electoral costs to court-packing might be minimal. Therefore, our goal in this paper is to bring data to bear on the electoral costs of court-packing.

Evaluating court-packing attempts

The previous discussion has highlighted two central features that might color citizens’ reactions to incumbents who seek to expand the courts: the rationale incumbents use to justify their proposals and the partisan match of the incumbent and the voter. Armed with these central concepts, we now turn to developing a set of empirical expectations to explain how they might affect citizens’ evaluations.

First, we anticipate that voters’ evaluations of incumbent proposers will vary based on how the proposal is framed or justified (Staton 2004). Voters are at an informational disadvantage, especially on relatively low-salience questions such as institutional reforms, making their opinions particularly malleable to framing effects. This is particularly salient on issues like the structural design of institutions that can have important implications for the quality of democracy. In particular, Levitsky and Ziblatt (2018) single out efforts to make the judiciary “more efficient” as a stealth authoritarian technique to capture the judiciary successfully (5). Because “efficiency” tends to be a broadly popular goal, antidemocratic proposals framed as efficiency-enhancing may garner support from the citizens, leading to the adoption of antidemocratic proposals that the public does not realize could have deleterious consequences. By contrast, proposals that expressly aim to exacerbate political cleavages may be more likely to encounter popular resistance.

We consider the ways in which court-packing proposals might be justified in two separate ways. In the first, incumbents may speak plainly to supporters, transparently admitting their instrumental motives for seeking to expand the courts with reference to shared priorities and values with judges they would seek to nominate. A more opaque and bureaucratic motivation, by contrast, justifies an attempt to expand the judicial ranks with reference to enhanced efficiency of case resolution and a more expedient administration of justice, the exact stealth authoritarian technique emphasized by Levitsky and Ziblatt (2018).⁵

In practice, politicians use both types of justifications. The US Democratic Party’s 2020 platform employed a broadly bureaucratic rationale for that party’s promise to nominate many new judges to the lower federal courts, citing vast increases in caseloads in recent decades as creating a need for the federal judiciary to be expanded. Advocacy groups have taken an explicitly politicized tone, pressing President Biden to prioritize political concerns in his court reform-related polices. Demand Justice (2021), one of these groups, provides an example of these more explicitly politicized calls to action: “To undo the damage Republicans did by stealing multiple Supreme Court seats, we should immediately add seats to the Supreme Court and appoint justices who will restore balance.”

Existing evidence on the politicization of the judiciary suggests a relationship between the justification for a court-packing proposal and the public reaction to it. As noted above, the public does not respond positively to explicit attempts to politicize the judiciary, withdrawing support from judicial institutions as perceived politicization increases (Gibson and Caldeira 2009; Christenson and Glick 2015). Even outside of the judicial branch, an array of evidence suggests that Americans dislike explicitly politicized processes and prefer more routinized, bureaucratic ones (Hibbing and Theiss-Morse 1995). Thus, we expect that the justifications given for the reform will shape citizens’ attitudes towards the candidates proposing them, such that candidates that advance reforms purported to be bureaucratic in nature will be evaluated more positively than those those that aim to politicize the judiciary.

Second, we entertain the possibility that the public’s response to court-packing will be informed by partisan considerations. Voters who share a party affiliation with a candidate have some shared fealty to the same policy goals, and comparative accounts of court reform elsewhere suggests that incumbents advance these initiatives with the support of their partisan followers (Staton 2004; Helmke 2017, 2010b). Clark and Kastellec (2015), for example, present experimental evidence that the public is willing to accept some attacks on courts when they approve of the attacker (c.f. Nicholson and Hansford 2014). Nelson and Gibson’s (2019) research substantiates these dynamics, demonstrating that Donald Trump’s attacks on the judiciary were only threatening to the US Supreme Court’s legitimacy among the minority of the public who expressed trust in President Trump. For the many Americans that hold President Trump in low regard, his attacks actually backfired and increased the Court’s support. Thus, it seems as though the identity of the proposer is consequential in explaining public evaluations and support.

Indeed, elites’ issue positions tend to rub off on their supporters (Lenz 2012). Broockman and Butler (2017) present field experimental evidence on this point in the case of state legislators: voters often adopted a state legislator’s issue position after learning of it, even when the position was accompanied with little justification. Work by Armaly (2018) extends this finding to the judiciary, finding that Americans react more favorably to attacks on judicial independence when they come from a presidential candidate the voter feels warmly about. As a result, even voters who do not care specifically about partisan entrenchment of the judiciary may decline to punish incumbents to attempt court-packing simply because they trust their copartisans in a general sense. Therefore, we expect that copartisan candidates will be supported at higher levels than outpartisan candidates.

Third, we anticipate that these reactions to a politicized or bureaucratic-framed proposal differ by the partisanship of the incumbent. A voter’s negative response to an outpartisan’s attempt to pack the courts is likely to be exacerbated when that incumbent frames his proposal in starkly political terms: not only is the proposal being made by “the other team” but the outpartisan is doing so in a way that advertises the unsavory, “political” nature of the proposal. On the other hand, copartisans who frame their proposals in politicized terms might even see some sort of boost from voters: when voters are likely to approve of the judges who will be placed in these new seats, they might like their copartisan’s attempt to entrench the judiciary with like-minded judges and reward her accordingly. This, we expect that the effects of rationale and copartisanship are interactive, not additive. In particular, the effect of a politicized proposal will be lessened for copartisan, rather than outpartisan, incumbents.

Finally, we expect that respondents’ evaluations of incumbents who propose court-packing vary according to their commitment to the institutional integrity of the courts. This is the logic that underscores prominent theories of interbranch relations: those who hold the judiciary in high esteem should be less tolerant of attempts to pack the judiciary than those who display low levels of institutional loyalty. Underlying these discussions has been an assumption that the public’s support for courts is nonpartisan in basis: institutional commitment should have a protective quality without regard to partisanship (e.g., Almond and Verba 1963, 357–9).

Given new research about the relationship between partisanship and support for courts (e.g., Bartels, Horowitz, and Kramon 2021), we anticipate that support for incumbents who propose packing the court will vary according to both the partisanship of the proposer and the citizen’s level of support for the court. When the incumbent is an outpartisan, we should observe the oft discussed shielding effect of institutional legitimacy: citizens with a high level of support for the judiciary will punish that incumbent more severely than those who hold the judiciary in low esteem. After all, if one doesn’t value the judiciary, there is no reason to punish someone for meddling with it!

But, we expect that the protective effect of diffuse support will dissipate in the face of shared partisanship between a citizen and an incumbent. When the would-be reformer is a copartisan,“valuing” the judiciary works at cross-purposes: an individual might value the judiciary in the sense that she does not want to see its structural integrity harmed, but, she also understands that she is more likely to get favorable policy decisions from a judiciary that is stacked with like-minded judges. We therefore expect that support for copartisan proposers will remain constant regardless of a citizen’s level of support for the judiciary.

Research design

We therefore have a series of expectations regarding (a) the rationale offered for the proposal, (b) the partisanship of the legislator who seeks to pack the courts, (c) the interaction of those two concepts, and (d) preexisting diffuse support for courts. We test these hypotheses in a focal case: the United States. The high level of political polarization between the two major parties and the high-profile efforts on both sides of the aisle to nominate partisan supporters to the bench makes this an auspicious opportunity to understand how stated motivations, partisanship, and preexisting institutional commitment inform the public’s evaluation of these court-packing efforts. Further, the positive public evaluation of the US Supreme Court suggests that the shielding effects of public support should be particularly likely to manifest in the United States (Gibson, Caldeira, and Baird 1998).

We tested these expectations in three separate experiments: an experiment embedded in the 2018 Cooperative Congressional Election Study (CCES), and in separate MTurk surveys fielded in 2018 and 2019.⁶ All three of our vignettes informed respondents of an effort by an incumbent Senator to increase the number of seats in the federal judiciary by about 40%, which was framed in either bureaucratic or political terms. Each of the three experiments have complementary strengths and weaknesses; that we achieve consistent results across all three samples is comforting evidence of the robustness of our conclusions.

We begin with the CCES study due to its high-quality respondent pool. We queried respondents to the 2018 CCES for their reactions to a survey vignette describing a hypothetical incumbent US Senator’s court-packing proposal to the federal judiciary. The vignette varied (a) the proposer’s rationale (bureaucratic or politicized), and (b) the partisanship of the proposer (Democratic or Republican). The bureaucratic rationale read “Legal experts from both parties have discussed the Senator’s proposal and agree that this proposal is an attempt to enhance the efficiency of the federal judiciary, enabling courts to better manage a backlog of cases.”⁷ Respondents who were assigned the politicized rationale read “Legal experts from both parties have discussed the Senator’s proposal and agree that this proposal is an ideological attempt to stack the federal judiciary with like-minded judges.” The two treatments were fully crossed. An example treatment (the Republican-Politicized treatment) read as follows:

An incumbent Republican Senator from a nearby state who is seeking reelection in November is campaigning on a plan that would expand the size of the federal judiciary, adding 64 new federal circuit court (appellate) judges (a 37% increase), and 189 new district court (trial) judges (a nearly 30% increase). He plans to introduce this bill in the U.S. Senate, if the Republicans win control of Congress in November. Legal experts from both parties have discussed the Senator’s proposal and agree that this proposal is an ideological attempt to stack the federal judiciary with like-minded judges.

Following the vignette, respondents indicated whether they would vote for the proposer in a hypothetical upcoming election, assessed the proposer’s job performance, and indicated their level of support for the proposal. The three items form a reliable unidimensional scale ( $ \alpha \hskip0.35em =\hskip0.35em 0.83 $ ); we, therefore, combine the three items into a single outcome measure, scaled from 0 to 1, which we call Proposer Support.⁸ Higher values of the outcome variable indicate more support for the proposer. More information on question wording and the properties of the scale is available in the Supplementary Appendix B.

Five design considerations deserve particular discussion. First, we are among the first to explicitly consider if an incumbent’s justification for a reform can actually sway the public’s support. The framing of the reform is especially important in light of the fact that the substantive outcome of the proposals is identical, irrespective of how it is framed. In either case, the proposed reform would result in the expansion of the number of judges in the US federal judicial hierarchy, opening a flood of new positions that would be filled by congressional majority. If it is the case that simply couching a judicial reform in bureaucratic terms has the effect of fostering broad public support, then this simple rhetorical shift may prove an effective tool for institutional capture and cooptation, all under the guise of legitimate and popularly supported political reforms.

Second, we based the vignette on court reform proposals that attracted some public attention in the lead-up to our experiment. We modeled the treatment after a proposed judgeship bill by Northwestern Law Professor Steven G. Calabresi, which proposed “that Congress should — at a minimum — authorize 61 new circuit judgeships… and 200 district court judgeships” (Calabresi and Hirji 2017, 21). Likewise, congressional Democrats introduced legislation in 2021 that would add more than two hundred seats to the federal judiciary (Adler 2021). In short, our hypothetical proposal is one that could have been credibly advanced by Senators from either side of the aisle, enhancing the experiment’s external validity.

Third, although survey experiments to evaluate public response to judicial decision-making are increasingly common (e.g., Baird and Gangl 2006 Bartels and Johnston 2013), existing experimental designs often present respondents with a hypothetical court decision, randomizing the particulars of the procedure or outcome and evaluating the extent to which citizens’ support shifts as a result. Where scholars have used an experimental approach to study interbranch relations, they mainly examine support for the curb or for the court as the outcome variable (Clark and Kastellec 2015; Armaly 2018; Nelson and Gibson 2019). By contrast, our outcome variables directly evaluate the public’s reaction vis-á-vis the incumbent proposer, and are therefore consistent with the mechanism that existing theories of judicial legitimacy imply, but have been rarely tested directly. These outcomes are also consistent with the framing of court reform proposals as incumbent position-taking activities meant to rally a base of electoral support (Clark 2011) and allow us to differentiate empirically from the support for the proposal itself, support for the candidate, and a respondent’s self-reported vote choice.⁹

Fourth, we study a court-packing proposal, rather than an actual court-packing attempt. We focus on proposals for two major reasons. First, and most importantly, we cannot randomize the presence of actual court-packing; thus, to provide some claim to external validity, our vignettes feature Senatorial proposals of court reform, which have the advantage of being relatively common in the past five years. Second, evidence from Clark (2009) demonstrates that the mere presence of court reform proposals (including those that would pack the courts) is powerful enough to shape judicial behavior. It is always possible the respondents questioned the credibility of the threat of the Senator’s proposal, yet the fact that we find consistent results before and after the 2018 midterms (when the Republicans lost full Congressional control) gives us comfort that respondents were evaluating hypothetical candidates’ campaign positions and not the broader reform environment. Of course, future research on this topic should attempt to move beyond a focus on proposals to examine the effects of implemented court expansion efforts.

Fifth, we crafted our hypothetical proposal with an eye for external validity. Whereas not every state has a senator from both parties, we instead discussed an incumbent “from a nearby state.” This is similar to the approach taken by Butler and Powell (2014) who queried respondents about state legislative elections in “a nearby state” in order to credibly randomize the partisanship of the party in control of the state legislature. We acknowledge that the hypothetical nature of the vignette is not ideal; however, such an approach was necessary to be able to credibly and randomly assign the partisanship of the proposer whilst not misleading respondents about real-life legislative initiatives. We are heartened by the findings of Brutger et al. (2022), who demonstrate that the consequences of hypothetical experimental treatments are negligable.

CCES results

The first three panels display the average value of the outcome variable across the experimental conditions. The lower-right panel shows the predicted value of Proposer Support as Federal Court Legitimacy varies, using estimates from Model 4 in Table C1. The whiskers represent 95% confidence intervals. Higher values of the y-axis indicate more support for the proposer. We first examine the direct effect of the experimental treatments on Proposer Support. Recall that we expected that proposers who couch their proposals in bureacratic language would be judged more favorably than those that seek to politicize the judiciary. As the upper left-hand panel of Figure 1 demonstrates, we find support for our hypothesis: respondents are on average less supportive of incumbents whose proposals are judged by experts to be politicized in nature ( $ p<0.01 $ ); the size of the effect is one-half of a standard deviation in size. Strategically describing a court-packing proposal in neutral or apolitical terms can bolster the public’s support or, minimally, blunt a possible backlash.

Figure 1. Experimental Results: CCES.

The upper-right panel of Figure 1 shows the average treatment effects for respondents who learned of a proposal by an outpartisan incumbent and those who learned of a proposal by a copartisan incumbent.¹⁰ The average level of support for copartisan incumbents who introduce a court-packing proposal is much greater than that for outpartisan incumbents who do the same. These findings are in line with our second hypothesis and consistent with other research that suggests that partisanship structures public opinion regarding the courts and incumbents’ efforts to reform them (Clark and Kastellec 2015; Armaly 2018).

Our third hypothesis related to the interactive effect of the two treatments. The lower-left panel of the Figure demonstrates that respondents are least likely to support an outpartisan incumbent advocating a politicized plan and most likely to support a copartisan whose proposal is ostensibly a bureaucratic fix. The difference is very large, accounting for a difference of more than one-half of the range of the outcome variable. However, these effects are additive; there is no evidence of an interaction between the two treatments ( $ p\hskip0.35em =\hskip0.35em 0.27 $ ). This finding is contrary to our third hypothesis.

Our final hypothesis related to the conditional effect of diffuse support. We expected that support for an outpartisan proposer would decline with one’s level of institutional support but that, when the proposer was a copartisan, support would be constant across levels of institutional legitimacy. This is exactly what we see in the lower right-hand panel of Figure 1, which provides predicted values from a linear regression that includes a multiplicative interaction term between the partisanship treatment and respondents’ level of federal court legitimacy.¹¹ For respondents who read of an outpartisan court-packing proposal, support for the proposer declines with respondents’ levels of diffuse support ( $ p\hskip0.35em =\hskip0.35em 0.03 $ ). Yet, there is no evidence that support for a copartisan proposer varies according to the respondent’s level of legitimacy ( $ p\hskip0.35em =\hskip0.35em 0.55 $ ).

These results help us to make two preliminary conclusions. First, incumbents can sway public opinion regarding institutional reform proposals simply by framing their efforts in apolitical terms. Second, legitimacy shields courts against subversion by outpartisans, but copartisans appear to be able to stack courts with relative impunity. This suggests that support for the institutional architecture of our democracy may be tightly entwined with one’s own partisanship.¹²

MTurk replication studies

Struck by these findings, we analyze their persistence in two MTurk samples. We fielded a survey of 2,500 respondents on MTurk in 2018 and a second survey of 1,500 respondents in July 2019.¹³ While the CCES sample provides a higher-quality sample, our inferences were limited by several design factors we sought to overcome with our MTurk surveys. First, the measure of judicial legitimacy in the CCES study is a single indicator rather than the typical multi-item index commonly used to assess the concept (Gibson, Caldeira and Spence 2003), a measurement limitation we remedied in our MTurk studies.¹⁴ Further, due to sample size considerations, the CCES experiment contained no control condition, limiting our ability to analyze these effects relative to a condition where respondents have no information about the proposer’s partisanship. Finally, our CCES design involved a judgment made by experts regarding the likely effects of the court-packing proposal. While we hoped this provided credibility to the rationale for the reform proposal, we acknowledge that respondents’ trust in experts could vary and confound the analysis. Accordingly, we changed this aspect of the design in our MTurk studies.

The MTurk experiments use a 3 × 3 fully crossed experimental design that adds a control condition to each experimental manipulation. For both the partisanship and rationale manipulations, the experiment contains treatments that omit the relevant information: the partisanship of the proposer and the proposer’s rationale for the court-packing attempt. The experiments also include a full battery of diffuse support indicators (see in the Supplementary Appendix B). We modified the standard battery of diffuse support questions suggested by Gibson, Caldeira, and Spence (2003) to the broader federal judiciary. As the Supplementary Appendix B demonstrates, the measure has a high degree of reliability and validity.

2018 MTurk results

The first three panels display the average value of the outcome variable across the experimental conditions. The lower-right panel shows the predicted value of Proposer Support as Federal Court Legitimacy varies, using estimates from Model 4 in Table C2. The whiskers represent 95% confidence intervals. Higher values of the y-axis indicate more support for the proposer.

The first three panels of Figure 2 show the direct effects of the experimental manipulations. Looking to the first panel, we observe that respondents were more supportive of incumbents who couched their proposals in bureaucratic language than those without a rationale and less supportive of those who used politicized language. These differences across effects are all statistically significant and provide additional support for our first hypothesis.

Figure 2. Experimental Results: 2018 Mechanical Turk Sample.

Next, we turn to partisanship. As shown in the second panel, respondents who read about a copartisan’s court-packing proposal expressed higher levels of support than those assigned to the control condition ( $ p<0.01 $ ); those respondents also expressed higher levels of support than respondents who read of an outpartisan’s court-packing proposal ( $ p<0.01 $ ). There is no difference in the level of support espoused by respondents in the control condition and those in the outpartisan category ( $ p\hskip0.35em =\hskip0.35em 0.09 $ ).

Moving to our third hypothesis about the interactive effects of the treatments, we see stark gaps in support based on the rationale given for the proposal. Respondents in all three partisanship treatments were more supportive of proposals couched in bureaucratic language than proposals with no rationale (the control group) and those described as an attempt at partisan entrenchment. However, as with Study 1, these effects are all additive; none of the multiplicative interaction between pairs of treatments achieves statistical significance at $ p<0.05 $ .

Finally, to examine the conditional effect of legitimacy, we turn to lower right-hand panel of Figure 2.¹⁵ Here, we observe the same pattern we observed in the CCES. For those in the outpartisan condition, respondents with higher levels of diffuse support express lower levels of support for the incumbent than those respondents with low levels of diffuse support. We see the same relationship in the control condition ( $ p<0.01 $ for both effects). But, there is no evidence that institutional legitimacy shields incumbents against actions taken by copartisan incumbents. There is no evidence that support for a copartisan incumbent varies according to a respondent’s preexisting level of institutional loyalty ( $ p\hskip0.35em =\hskip0.35em 0.73 $ ).

Combining the findings from Figure 2, a clear but striking picture emerges: when respondents are given no information about the identity of the proposer (the control condition), the public is generally unsupportive of court-packing efforts. This is consistent with the conventional wisdom: in a country where the courts and judicial system is broadly supported, the public is on average uncomfortable with efforts to reform the courts. Yet, our research reveals that this reticence disappears when they are aware of the shared partisanship of the proposer. Moreover, while respondents are willing to punish a proposer of unknown partisanship for packing the courts, shared partisanship exerts a powerful protective effect, regardless of a respondent’s level of institutional loyalty.

2019 MTurk results

The 2019 MTurk survey took an additional step, altering the wording of the vignette such that the rationale is spoken directly by the incumbent rather than an expert. For the politicized treatments, we took advantage of the fact that Senators have not been shy about mentioning the ideology of judges when discussing the Senate’s efforts to confirm presidential judicial nominees. Consider the remarks of two recent chairmen of the US Senate Judiciary Committee. Lindsey Graham (2018) tweeted that one goal as Chairman was to “push for the appointment and Senate confirmation of highly qualified conservative judges to the federal bench.” Likewise, when addressing the Federalist Society, Chuck Grassley (2018) bragged about “our success confirming qualified, conservative judges.”

The updated politicized treatment therefore read, “The Senator claims that this proposal will help ensure that the judiciary is filled with qualified, [likeminded/liberal/conservative] judges” (where “likeminded” is used when the Senator’s partisanship is not stated and liberal or conservative accompanies a proposal by a Democratic or Republican proposer).¹⁶ The respondents answered the same three outcome variables. Of respondents, 26% would vote for the incumbent, 49% approved of the proposal, and 53% approved of the Senator’s job performance. Again, we combine the three indicators into a single measure of support for the proposer.

The first three panels display the average value of the outcome variable across the experimental conditions. The final panel shows the predicted value of Proposer Support as Federal Court Legitimacy varies, using estimates from Model 4 in Table C3. The whiskers represent 95% confidence intervals. Higher values of the y-axis indicate more support for the proposer.

The first three panels of Figure 3 tell a now-familiar story. The first two panels show that proposers who use bureaucratic language are judged more favorably than those without a rationale, and those who use politicized language are judged least favorably. In line with our hypothesis regarding the effects of partisanship, there are stark gaps based on the partisanship of the proposer in the both of the middle panels, with respondents rating copartisan proposals more highly than outpartisan proposals; all of the differences across conditions are statistically significant. And, as the third panel shows, these effects are additive, not interactive; again, none of the two-way interactions between pairs of treatments is significant at $ p<0.05 $ .

Figure 3. Experimental Results: 2019 Mechanical Turk Sample.

The lower right-hand panel of Figure 3 enables us to examine how respondents’ support for the proposer varies as a function of their preexisting support for the judiciary.¹⁷ These results are very similar to those reported in the other two samples. Respondents in the control and outpartisan conditions display lower levels of proposer support as their level of federal court legitimacy increases ( $ p<0.01 $ for both effects). Yet, this is not true for respondents who read of a copartisan proposal ( $ p=0.35 $ ). We reiterate that this finding – consistent across three samples, each with slightly different experimental designs – challenges the conventional wisdom about the “costs” of court-packing: it appears that copartisans are shielded from reprisal for these types of proposals.

Discussion

We set out to investigate to what extent the public would punish incumbents’ attempts to pack the judicial branch with political allies. A long heralded explanation of the maintenance of judicial independence suggests that high levels of public support for judicial institutions provides courts a proverbial ‘shield’ from interbranch meddling (Caldeira and Gibson 1992; Vanberg 2008, 2015). This research suggests that attempts to politicize the courts would be met with hostility, especially in places where courts are broadly respected, and among those who held the courts in highest regard. By contrast, we suggested that incumbents might be able to sidestep the deleterious consequences of these proposals through strategic framing to minimize a possible backlash and among their copartisan supporters.

Across three separate studies, each with their own advantages and limitations, our evidence points in consistent directions. We find that the public is more amenable to reforms described in apolitical terms than those that would politicize the courts and more supportive of copartisans than outpartisans who advance an identical proposal. Further, the public’s preexisting support for the court does not provide an unqualified shield that would prevent court-packing. The court-loving public’s willingness to withdraw support is conditional on the partisanship of the proposer: when presented with a generic incumbent, or an incumbent from the opposite party, support for an incumbent who seeks to pack the court declines with one’s level of support for the court. Yet when the would-be reformer is a copartisan, even those with most reverence to the courts were no more likely to withdraw support.

Given the strong effects of framing and copartisanship, we conclude that the electoral connection may only provide a feeble shield for the courts in the face of subtle incumbent cooptation. In this way, our findings fit well with other recent studies that have shown that shared partisanship presents an obstacle to the defense of democratic institutions, especially in highly polarized political contexts (Graham and Svolik 2020; Svolik 2020). Accordingly, due to the high levels of polarization in the United States, further scrutiny of this hypothesis across a variety of electoral contexts is a necessary path forward.

To the extent that politicians stack new judicial posts with allies, our findings corroborate accounts of authoritarian consolidation in the context of democratic backsliding and make clear the calculus for court-packing leaders around the world. The burgeoning literature on democratic decay suggests that the sort of framing effects we explored here are especially useful to those incumbents who seek to weaken democratic institutions. Varol (2015) emphasized “stealth authoritarianism” as a key mechanism behind successful cooptation or capture of democratic institutions. Ginsburg and Huq (2018) describe a similar process (“constitutional regression”), which they characterize as “incremental (but ultimately substantial) decay in [the] basic predicates of democracy” (83); these authors note that this process often results in the erosion or elimination of institutional checks and balances. Finally, and particularly on point, Levitsky and Ziblatt (2018) note that many attempts to undermine democratic institutions are posed as “attempts to improve democracy” such that “[d]emocracy’s erosion is, for many, almost imperceptible” (5–6). Thus, while citizens might otherwise reject a court-packing proposal and punish an incumbent for suggesting it, creative justifications on behalf of those who would seek to stack courts might make citizens unable to recognize court-packing proposals and, by extension, unable to punish incumbents for proposing such measures (e.g., Carey et al. 2020; Svolik 2020). Alternatively, unsubstantiated here but still possible, is the chance that the public might reward incumbents for efforts to stack the courts with political allies (Clark 2009). Future research should consider – across a broad cross-section of electoral environments – the public’s ability to recognize these politicization attempts and interventions that can increase the public’s awareness of so-called reform attempts that might otherwise be enacted unnoticed.

One important point of discussion concerns the generalizability of our results in terms of both temporal and institutional context. Our experimental design focuses on plans to pack the lower federal courts, and our experiments were fielded before calls to pack the US Supreme Court had picked up steam. With regard to the temporal aspect, the recent American experience dovetails well with the findings we report here: the public is, in many cases, supportive of court curbing that moves the judiciary’s expected policies toward their own favored outcomes. Yet, with regard to institutional context, we note that Supreme Court and lower court-packing are not synonymous. For one, the bureaucratic justifications that we demonstrate are effective for swaying public support on these proposals for the lower court have the benefit of being true for the lower federal courts: they are drowning in cases to resolve. The Supreme Court, on the other hand, decides far fewer cases per year than it did decades ago. Further, because the Supreme Court’s role is particularly policy-minded, the public may be less concerned about expressly politicized rhetoric than the results we report here for lower courts.

Further, we note that there are two equally plausible interpretations of our findings.¹⁸ On the one hand, our results might suggest that the public is full of partisans who are enthusiastic about the prospect of bending the federal judiciary toward their favored political outcomes. At the same time, our results could be reflective of a trustee model of democratic representation, wherein the public entrusts copartisan incumbents to enact policies they favor. This would imply that citizens judge their copartisans’ court-packing proposals more favorably not because they appreciate the value in a judicial branch with partisan judges, but rather because they have faith in their copartisans’ efforts, no matter what they may be. Future research should disentangle these possibilities.

As with any experiment, external validity concerns deserve some additional discussion. In particular, we view these effects as likely maximal effect sizes. In real-world conditions, many respondents will not learn of court-packing efforts, may not understand which party is pushing to expand the Court, may receive conflicting framing about the rationale for the proposals, or may learn about a proposal supported by a bipartisan coalition (Barabas and Jerit 2010). By contrast, every respondent in our experiment received a single frame attached to a single proposal made by a single senator. Future work should examine how two-sided frames and conflicting information cues affect the dynamics we uncover here.

Likewise, we focus our efforts on the effects of court-packing proposals on the candidate proposer; future work should also investigate the effects of these efforts on the court. The literature on judicial legitimacy suggests that if citizens view judges as part of normal give-and-take of partisan politics then the esteem in which their institution is held by the public dissipates dramatically (Hibbing and Theiss-Morse 1995; Baird and Gangl 2006). In other words, for regimes to engage overtly in court-packing is to risk lowering the very legitimacy that they require for the judiciary to be effective. The 2021 Biden Commission, reviewing possible structural changes to the US judiciary, suggested that court-packing, judicial term limits, and other types of judicial reforms might weaken the judiciary’s legitimacy. Testing this logic is an obvious path forward for future research.