The Supreme Court and Congress: Information,Threat,and the Separation of Powers

Abstract

Separation of powers research concerning the Supreme Court focuses on when and how the political branches can constrain the Court. Recent work theorizes that the justices are willing to defer to the legislative branch in order to protect the Court’s institutional legitimacy, arguing that it’s not the threat of specific congressional override that ‘checks’ the Court. Rather, broader threats fulfill this function – threats in the form of court-curbing bills and eventual laws. I reexamine the assumptions and key operationalizations of these institutional maintenance frameworks, and seek to parse threat conditions that can produce legislative retribution away from the more general and non-legislative threat environment of political criticism. To do so, I utilize a type of non-legislative House floor speech called one-minutes, revealed by recent research to be a potent indicator of constituent and party concerns, even more closely reflecting public priorities than committee work or bill introduction. My results indicate that the Court is sensitive to this broader environment of political criticism and praise – as operationalized through congressional one-minute speeches.

Keywords

supreme court separation of powers judicial behavior congress

Introduction

“We see it [checks and balances] particularly displayed in all the subordinate distributions of power, where the constant aim is to divide and arrange the several offices in such a manner as that each may be a check on the other – that the private interest of every individual may be a sentinel over the public rights.”

-James Madison, Federalist 51

If the U.S. constitution functions as it was intended – or at least as Madison’s Federalist 51 suggest it should – a system of checks and balances affords each branch a measure of power over the others. The Supreme Court, for example, can declare federal legislation unconstitutional or use statutory construction to shape a law’s meaning; Congress can statutorily override any such construction and also controls the judiciary’s size, structure, budget, and appellate jurisdiction. Separation of powers research, then, focuses on whether or not these checks and balances affect the behavior of relevant institutional actors. Despite the theoretical appeal, initial scholarship in this field produced few positive results; for the most part the Court does not alter its agenda-setting and decision-making behavior, even under conditions where the Court might expect an override from other branches (Owens, 2010; Segal, 1997; Sala, James, & Spriggs, 2004). Later work, however, suggests scholars might be looking in the wrong places when searching for separation of powers effects. Lacking the power to enforce its own decisions, these works posit that the Court must cultivate and sustain public confidence in the judiciary so that (1) the public will accept and comply with judgments, and (2) the political branches are willing to provide enforcement that the Court cannot (Clark, 2009; Mark & Zilis, 2019; Segal et al. 2011). In this configuration, it’s not a specific Congressional override that the Court fears, but a broader threat to institutional legitimacy, where threat has previously been measured via court-curbing bill introduction and ideological distance between the Court and relevant political players.

In this paper, I reexamine the assumptions and key operationalizations of these institutional maintenance frameworks, and seek to parse threat conditions that can produce legislative retribution away from the more general and non-legislative threat environment of political criticism. My specific contribution is in attempting to more precisely determine what the Court is concerned with when trying to protect its institutional legitimacy: does it want to avoid court-curbing legislation by being cautious when the political branches are ideologically situated to pass such legislation (Segal, Westerland and Lindquist, 2011) or when court-curbing bills have actually been introduced (Clark, 2009) – or, does the Court wish to limit political criticism, maybe both? To do so, I utilize a type of non-legislative House floor speech called one-minutes, revealed by recent research to be a potent indicator of constituent and party concerns, even more closely reflecting public priorities than committee work or bill introduction. Key with one-minute speeches is that they don’t carry a threat of legislation, so convey a Congressman’s criticism or praise without the automatic possibility of court-curbing bills or laws.

To forward this argument, I examine Supreme Court decision-making behavior, specifically when the constitutionality of a federal law is under consideration. Instead of attempting to model the Court’s behavior on rarely-used constitutional checks, I focus instead on legislative position-taking in the form of one-minute speeches given on the floor of the House of Representatives. To do so, I next summarize the literature of relevant work. Section three explicitly lays out my theory and expectations for testing this theory. Section four describes the data and methods used to test these hypotheses. Section five describes the results and section six contains discussion about the results and their larger implications.

Court Legitimacy & Congressional Threat

Scholars have long been interested in how the separation of powers affects the Court, and specifically whether or not the Congress can alter justices’ decision-making behavior (Clark, 2009; Epstein, Knight, & Martin, 2001; Hansford & Damore, 2000; Mark & Zilis, 2019; Segal et al. 2011). To this end, Segal, Westerland and Lindquist (2011) examine two potential avenues for affecting the Court’s decision-making behavior: this influence can come in direct, policy-based considerations, or in more broad legitimacy-based considerations. But why would the Court – tucked away in its ivory tower and composed of members with lifetime tenure, insulated from the populous writ large through appointment rather than election – care about policy and legitimacy? The simple answer is that the Court cannot enforce its decisions and so must rely on a ‘reservoir of good will’ in order to see its policy effectively implemented (Bartels & Johnston, 2013; Benesh, 2006; Caldeira & Gibson, 1992; Casillas, Enns, & Wohlfarth, 2011).

To further parse out these motivations – policy and legitimacy – Segal, Westerland and Lindquist (2011) test two competing theories of Congressional influence on the Court’s decision-making behavior: they set the commonly-used rational anticipation model against their creation, the institutional maintenance model. Early studies rely on the rational anticipation model which charges justices with anticipating whether or not their decisions would be overturned in any specific case (Ferejohn & Shipan, 1990). That is, this model assumes justices make game-theoretic type calculations, mapping out each pivotal actor along with the likelihood of congressional override. These calculations are made with the sole outlook of preserving the Court’s most-preferred and achievable policy position – ostensibly, one that if chosen by correctly mapping the ideological landscape, will be immune to Congressional override. Work relying on this theoretic backdrop has, however, failed to produce generalizable and convincing results (Epstein & Knight, 1998; Spiller & Gely, 1992).

As an alternative, Segal, Westerland and Lindquist (2011) propose the institutional maintenance model , which relaxes some of the rational anticipation assumptions and assumes the Court has more to worry about than a direct override of its decisions. There are, for example, stagnating budgets, Congressional control of the Court’s appellate jurisdiction, and policy compliance to consider. Operating in this broader political landscape, justices aren’t necessarily concerned with shielding themselves from direct congressional override on a case-by-case basis, but rather are sensitive to the Court’s broader reputation and therefore seek to avoid court-curbing legislation. Since issuing too many unpopular decisions can result in indirect ‘purse and sword’ retaliation from Congress which in turn harms the Court’s legitimacy, “justices need only sense when they might be in institutional danger and should ratchet back confrontational behavior with Congress in those circumstances” (2011, 93). The authors then operationalize ‘sens [ing] when [the justices] might be in institutional danger’ in two ways: (1) via a series of variables measuring the ideological distance between the Court and relevant political actors (i.e. the President, Congress), and (2) with the number of court-curbing bills introduced the prior year.

Use of court-curbing bills comes from Clark (2009), who adopts a similar theoretic framework, casting the preservation of Court legitimacy as the primary goal in separation of powers conflicts. Clark, however, takes his assumptions one step further and theorizes that the Court also fears for its institutional legitimacy – when the conditions are right (or maybe wrong) – under even the threat of any sort of sanction, as “…the Court is concerned about political criticism” (2009, 973). Clark paints the Court as informed and reflective – it recognizes the institution is vulnerable by looking to public opinion and prior congressional behavior. Legitimacy is protected, then, by acting in a criticism-reducing manner when public opinion is sufficiently low and when Congress has been sufficiently threatening.

To recap, various models that attempt to predict separation of powers effects between Congress and the Court all begin with defining what the Court might fear, and how it could avoid that outcome. Rational anticipation theorizes the Court self-censors a particular decision when it fears Congress might override that specific decision. Moving on from this, the institutional maintenance model relaxes this one-to-one requirement by redefining both the fear and the self-check. Justices concerned with institutional legitimacy fear broader court-curbing legislation such as a change in size or budget, and are therefore more willing to self-censor in any decision where federal legislation could be struck – the self-censoring is more and more likely when the two branches are more ideologically distant. Finally, Clark (2009) argues the Court fears not just the probability that override or court-curbing legislation could be passed, but that political criticism, in the form of even unsponsored, stuck-in-committee, never-going-to-pass court-curbing legislation, is harmful to the Court’s legitimacy when it’s already in a vulnerable position. Rather than measuring ideological distance from Congress like Segal et al. (2011), Clark operationalizes threat by counting the number of court-curbing bills introduced per year; I note, however, that he “…leave[s] open the possibility that other possible signals may exist” (2009, 977).

Congressional One-Minute Speeches

One-minute speeches occur at the beginning of each legislative day and allow any House Member to speak on any issue, whether they’re part of the majority coalition or not. To participate, a representative must arrive early and sit in the front row of their respective side of the chamber; the chair then alternates between parties beginning with a member from the majority. Once recognized, Members are given one minute (hence the name) to talk about whatever they want, subject only to general House rules (Hudiburg, 2020). Although providing for one-minute time is not part of the standing House rules, time for these speeches is allowed almost every legislative day and has developed through unanimous consent requests (Schneider, 2007). In recent years, there have been typically 15 Democrats and 15 Republicans allowed to speak during a given day’s one-minute time, but this is determined by the Speaker. Because this particular type of legislative speech occurs every day and is open to anyone and anything, one-minute speeches have been studied extensively, yielding two broad findings that are relevant to my research.

First, the sentiments reflected and topics covered in one-minute speeches align with the preferences and concerns of Members’ constituencies.¹ Drawing on Mayhew (1974)’s electoral connection, this growing body of research argues House Representatives deliberately and strategically use one-minute speeches to advertise and position-take on issues most important to concerned constituencies (Hughes, 2018; Kalaf-Hughes, 2020; Pearson & Dancey, 2011a, 2011b). Stated succinctly by Hughes and Koger (2022), “House members understand most voters will not witness these [one-minute] speeches on C-SPAN (Cable-Satellite Public Affairs Network) or read the transcript of the Congressional Record, so speeches are rebroadcast through members’ newsletters, websites, and YouTube channels” (p. 830). Additionally, since there are electoral risk-taking concerns involved with such public advertising and position-taking, MCs position-take and advertise preferences that are aligned with their constituents (Mayhew, 1974; Rocca, 2007). Hughes (2018), for example, categorizes all one-minute speeches given between 1989 and 2012 in order to compare issue attention by different types of Congressional behavior. He finds that the content of one-minute speeches more closely aligns with public interests than do other types of legislative behavior such as bill-introduction and committee hearings (p. 176–177). Hughes (2018) attributes this alignment with public opinion to the House Members’ close electoral connections.

Second, and in addition to this electoral connection, Hughes (2018) also attributes the overlap of one-minute speeches with public opinion to the recognition of one-minutes as valuable outlets for party platforms. Early research into non-legislative speech uncovered internal party leadership memoranda used to specifically coordinate one-minute messaging campaigns during the 101st Congress (Harris, 2005), and subsequent interviews conduced with Congressional staffers in 2014 show this type of coordination and solicitation has continued (Hughes & Koger, 2022). Drawing on prior Congressional scholarship, Harris (2005) theorizes that parties exert control over messaging in order to “construct and nurture the party’s image” since there are “collective benefits of a favorable party image” (p. 128). These collective benefits include “build[ing] a party record on which members can run for reelection, and shap[ing] an overall media image for the party” (p. 129). Parties therefore solicit and coordinate one-minute speeches as part of a larger effort to cultivate and sustain strong party brands (Harris, 2005; Hughes & Koger, 2022). Most importantly, however, the parties act as messaging gate-keepers – even when not actively working to organize and promote specific one-minute content – in order to protect their brands and the benefits that flow from them. This gate-keeping results in the practice of one-minutes departing in specific ways from House procedure, where procedure allows for one-minute speeches on any topic:as one Congressional staffer put it, “you never give a speech on the floor and go against the caucus” (Hughes, 2016, p. 200, emphasis added). This combination of organizing and gate-keeping means that the messages promoted in one-minute speeches either have some broader public appeal that a party hopes to tap into, or are at least not publicly unpopular in a way that could damage a party’s brand.

Taken together, these insights reveal one-minute speeches are potent signals, representing salient issues important to an MC’s constituents that align with, or are acceptable to, party leadership. This signal is further enhanced by the time and effort costs involved in giving one-minute speeches, especially in contrast to sending a tweet or signing off on the contents of a newsletter. Indeed, Members of the House must physically line up at the beginning of the legislative day as only 15 members of each party are recognized. This is not a responsibility that can be done electronically or staffed out – the MC must show up early and wait her turn in order to give a one-minute speech. While these 30 or so minutes may not seem like time and effort costs, legislators’ days are tightly scheduled, and 30 minutes could easily be used on a host of other issues. The point is that the costs involved serve to highlight the electoral and/or party importance of any topic an MC chooses to cover during a one-minute speech. In short, this combination of gate-keeping and MCs’ electoral motivations means one-minutes convey important constituent concerns that overlap with party positions since party leaders both organize and police their messaging. If the concern originates with a Member’s constituency, that concern is informally ok’d by party leadership prior to materializing as a one-minute speech (Hughes, 2016); if, however, the messaging is party-driven, individual MCs only volunteer to be its mouthpiece if the issue is aligned with or shared by their constituency (Hughes & Koger, 2022). Critically, both of these pathways carry a public opinion valence, where (1) party leadership organizes messaging that promotes the party brand through broad public appeal, or allows messaging that isn’t unpopular with the public and therefore damaging, while (2) MCs speak on issues that represent strong constituent concerns.

This transmission of public opinion through elites via one-minute speech-giving therefore represents the attentive public’s most intense preferences – preferences strong enough to elicit party recognition and MC responses. In addition to these theoretical advantages, however, one-minute speeches are also appealing for three related practical reasons. First, one-minute speeches are specific – they are concise and require no translation or guessing.² Relatedly, the sentiment of one-minute speeches can be positive rather than the purely negative nature of bill introduction. Indeed, as I report in more detail below, roughly one quarter of court-related speeches are positive, and these data allow researchers to test whether message sentiment matters when looking for inter-branch effects. Finally, and maybe most advantageous, one-minutes are non-legislative, and are therefore pure political criticism (or praise) without the attached threat of legislative action. This aspect means I can more cleanly test whether the Court is sensitive to this type of political criticism.

The Court, the Public, and the Political Branches

Despite there being no direct evidence the Court tallies Congressional votes or watches one-minute speeches, appearances and writings by the justices, themselves, provide evidence that the Court is generally aware of – and not immune to – current and political events. As early as 1928, Justice Cardozo opined, “…if there is anything of reality in my analysis of the judicial process, they [judges] do not stand aloof on these chill and distant heights…The great tides and currents which engulf the rest of men, do not turn aside in their course, and pass the judges by” (168). Nearly 60 years later, Justice Rehnquist used a similar analogy when delivering an invited lecture at Suffolk University Law School:

The judges of any court of last resort, such as the Supreme Court of the United States, work in an insulated atmosphere in their court-house where they sit on the bench hearing oral arguments or sit in their chambers writing opinions. But these same judges go home at night and read the newspapers or watch the evening news on television; they talk to their family and friends about current events. Somewhere ‘out there’ - beyond the walls of the courthouse - run currents and tides of public opinion which lap at the courthouse door (Rehnquist, 1986, 768).

Speaking before a House Appropriations Subcommittee in 2015, Justice Kennedy, for example, discussed the ‘information revolution’ and the near instantaneous availability of expert legal opinions, posted on blogs and news sites.³ Despite claiming that he, himself, did not read such blogs, Justice Kennedy was clearly aware of them and their contents – much in the way Justice Rehnquist described the tides of opinion lapping at the courthouse door. Similarly, in an interview with NPR in 2019, the late Justice Ginsburg indicated she knew about and was not in favor of proposals to increase the size of the Court, stating “Nine seems to be a good number.”⁴ At the time, such proposals were rumored and reported in the news, but had not yet been introduced as bills.

There is also ample evidence that justices follow presidential elections. It was widely reported, for example, that Justice O’Connor remarked, ‘this is terrible’ on election night of 2000 after CBS called Florida for then-Vice President Gore. A life-long Republican who served in the Arizona state senate, O’Connor was a savvy politician – even winning the Court’s 1992 election betting pool – the existence of such a betting pool even more evidence that justices follow politics. Tucked in Justice Blackmun’s vast collection of private papers turned over to the Library of Congress in 2004, Maltzman et al. (2004) found a four-page document detailing the wagering game and outcomes. Six of the nine justices participated by predicting which presidential candidate would win the popular vote in each state as well as the overall vote; thus, each participating justice made 51 predictions for 306 total selections. All told, there were 261 correct picks (85%), and interestingly, their accuracy increased along with margin of victory per state in 1988 (2004, 840). Put plainly, justices’ knowledge of the prior election helped them pick winners in the 1992 election.

Beyond this political awareness, the Court also has self-awareness to understand the complexity of its position vis-à-vis the public and elected branches. Dissenting in Baker v. Carr, Justice Frankfurter argued that maintaining legitimacy required the Court steer clear of political entanglements: “The Court’s authority - possessed of neither the purse nor the sword - ultimately rests on sustained public confidence in its moral sanction. Such feeling must be nourished by the Court’s complete detachment, in fact and in appearance, from political entanglements and by abstention from injecting itself into the clash of political forces in political settlements” (Baker v. Carr 369 U.S. 267, Frankfurter Dissenting). Writing nearly 40 years later, Justice Sandra Day O’Connor stated the matter more bluntly, “We don’t have standing armies to enforce opinions, we rely on the confidence of the public in the correctness of those decisions. That’s why we have to be aware of public opinions and of attitudes toward our system of justice, and it is why we must try to keep and build that trust” (O’Connor, 1999, p. 13). Bringing all elements together – Congress, the Court, and the public – one justice noted, “The Court is pretty good about knowing how far it can go…Congress is better than we are, especially the House. They really have their finger on the pulse of the public” (Clark, 2009, p. 973, emphasis added). This is all to say that the Court is (1) sensitive to the invisible guard rails of public opinion and ‘political entanglements’ as it is (2) politically-aware, and (3) understands its reliance upon sustained public confidence.

Theory and Hypotheses

Each of the separation of powers theories detailed above relies, to different degrees, upon elements of rational choice theory as they each paint the Court as a sophisticated actor that attempts to overcome constraint in order to maximize its preferences. Information plays a key role in models built upon rational choice theory: if actors have complete and perfect information, they control their fate and can successfully choose when to exercise constraint. We know, however, that the Court isn’t perfectly attuned to Congressional preferences, otherwise the Court would never be ‘checked’ by the other branches or fear for its legitimacy. It is precisely because the Court has incomplete information regarding the possible reactions of Congress and other external actors that the justices must be sensitive to their broader political environment in order to avoid legitimacy threats. Even in the less demanding institutional maintenance framework of Segal et al. (2011), avoiding court-curbing reprisals requires that justices (somewhat) accurately ascertain Congress’s view of the Court – operationalized there as a measure of vote-based ideological distance. While ideology scores or measures of ideological distance between institutional actors are often employed for such a purpose, I argue other and more nuanced information is available.⁵

Given this background, I posit valuable information can be gleaned from the sentiment expressed via one-minute speeches on the House floor – from negative speeches (Geer, 2006), but I also contend there is valuable information in positive speeches. If the Court perceives political criticism as a threat, and as a sophisticated actor moves to reduce threat, it should also recognize praise as providing a sort of ‘wiggle-room’ it can use to pursue its own preferences. The two variables previously used to measure potential institutional threat (court-curbing bills and congressional ideology) cannot account for this valence, and so my data provide a unique opportunity to do so. One-minute speeches about the federal courts are public, observable, intentionally disseminated for position-taking, occur at an average of more than once per week during legislative sessions, and their sentiment is easily understandable. Add to this the fact that MCs have limited time and effort, and so to praise a Supreme Court decision or to engage in a negative attack on the federal judiciary is itself a signal of the issue’s importance. Pivotally, Members of the House of Representatives share an ‘electoral connection’ with their constituencies and as such, these speeches offer a measurable account of intense constituent preferences. Most importantly for my theory, however, is that one-minute speeches are non-legislative in nature and therefore represent political criticism in its purest form, as detached from legislative threat.

Recall that institutional maintenance frameworks are built upon a fear of court-curbing legislation, but also the Court’s concern with “general political conditions” (Segal et al., 2011, 92) and “political criticism…how it is perceived by the public and members of the bar” (Clark, 2009, p. 973). While court-curbing bills and congressional ideology scores (calculated using roll call votes) are one way of measuring the generalized political conditions, I argue one-minute speeches offer advantages when specifically looking to test the Court’s sensitivity to criticism. First, court-curbing bills are actual pieces of legislation, and so while all are inherently critical, some are more symbolic in nature while others are potential legislative threats. Since there is no systematic method of cleanly parsing those bills which are purely symbolic from those carrying potential legislative threat (and doing so contemporaneously at bill introduction, rather than, e.g., seeing how far a bill progresses through the legislative process), a measure of court-curbing bill introduction represents some combination of both threat and criticism. Because this study aims to determine whether the Court truly is sensitive to reputational criticism absent legislative threat, one-minute speeches are a more attractive measure because they are not bills.

I want to pause for a moment to clarify that none of this should be read as saying that Clark (2009) or Segal et al. (2011) are wrong, and indeed I do not challenge their findings. My contribution is in pushing the operationalization of congressional preferences slightly further, while still keeping in line with their foundational assumptions. A Court motivated to maintain its institutional legitimacy certainly fears being legislatively curbed – having its jurisdiction limited or its size expanded – but it also fears loss of its ‘reservoir of good will’ such that implementation and compliance become problematic. Utilizing one-minutes helps determine if the Court is only sensitive to situations where passing court-curbing legislation is on the table (because a bill has been introduced and/or the Congress is ideologically willing to pass court-curbing legislation), or if it also constrains itself in response to non-legislative political criticism. Because the Court is sensitive to its information environment:

Negative Speeches: The Court will be less (more) likely to strike federal laws the higher (lower) the number of negative Court-related one-minute speeches per year.

Positive Speeches: The Court will be more (less) likely to strike federal laws the higher (lower) the number of positive Court-related one-minute speeches per year.

In sum, $H_{1}$ and $H_{2}$ test whether the Court taps into the broader political environment reflected in Congress’s judiciary-specific sentiment. Hostile and favorable speeches should work in opposite ways, making the Court fear (or not) criticism or retaliation if it strikes legislation and calls Congress’s authority into question. I next detail how I operationalize and model these hypotheses. Again, I don’t go so far as predicting that one-minute speeches – themselves – cause the Court to alter its decision-making behavior. Rather, I theorize that one-minute speeches reflect intense constituent and party preferences that are then also taken into consideration when pondering the constitutionality of federal legislation.

Data & Methods

I begin my data collection efforts by gathering one-minute speeches for the 101st - 114th Congresses (1989–2016). Pearson & Dancey, 2011b provide raw text files of one-minute speeches from 1989–2002; I update these data by downloading one-minute speeches from congress.gov. I next search each speech for the word ‘court’, drop non-germane speeches,⁶ and code for whether the speech addresses federal, state, or ‘courts’ in general. This allows me to isolate speeches relevant to the federal judiciary, and federal judicial power, more generally.⁷ From an initial pool of 646 speeches germane to courts of law, including only references to federal courts leaves me with a dataset of 544 one-minute speeches.⁸

I additionally read and hand-code all 544 one-minute speeches for negative, positive, or neutral sentiment. This coding process is straight-forward given the direct nature and time constraints of these Congressional speeches – Members simply do not have time to equivocate or address multiple topics. Specific examples of speeches coded negative include, “We have seen the Supreme Court…deal a cruel blow to many citizens…” (Oakar, 6/14/1989), as well as,

Mr. Speaker, the Supreme Court will soon hear oral arguments in the case commonly referred to as Hobby Lobby v. Sebelius. The outcome of this case will determine whether or not a for-profit company has the right to limit a female employee’s access to health care under the guise of religious freedom. Already the Supreme Court has wrongly declared that corporations have a right to ‘freedom of speech,’ as determined in the case of Citizens United. In just a few short years, this ruling has led to a flood of undisclosed money into our elections and corrupted our political system. Corporations’ latest attempts to secure the constitutionally-protected rights of citizens is equally as dangerous (Slaughter, 3/13/2013).

Examples of speeches coded in the positive category include language such as, “I rise to congratulate the Supreme Court on its decision…” (Lampson, 5/14/2002). Another example frames the Court as the people’s champion:

This week, the working guy or gal actually got a reprieve from one of these costly burdens when the Supreme Court placed a stay on President Obama’s so-called Clean Power Plan …I applaud the Supreme Court for placing a hold on this Big Government tax on my constituents. Finally, a win for the little guy (Yoder, 2/11/2016).

In addition, Members often use descriptive language such as, excellent, praise, or correctly-decided as well as thoughtless, shocking, or deplorable when giving positive or negative speeches, respectively.

Finally, speeches are coded neutral when referring passively – and with no value statements – to something the Court has done or might do in the future, that is “the Supreme Court in Beck found that only 19% of the dues collected were used for collective bargaining purposes” (Hancock, 3/15/1990).⁹ While the distribution of sentiment varies across Congresses, negative speeches account for the cumulative majority of one-minute speeches at 53.5%. Members of Congress give the fewest neutral speeches (22.4%) while just under a quarter are positive in sentiment (24.1%).

I model decision-making behavior by examining the Court’s willingness to strike down federal legislation as unconstitutional. To do so, I begin with Whittington (2022)’s Judicial Review Database, which catalogs all Supreme Court cases where the constitutionality of a federal law is at issue. My dependent variable takes on a value of 1 if the Supreme Court declared the Law Unconstitutional , and a value of 0 if the law was upheld. Since I’m attempting to model whether the Court’s likelihood of invalidating federal legislation shifts along with expressed sentiment from Congress, I include a count of the number of MCs Giving Negative Speeches and MCs Giving Positive Speeches per year – both calculated using my one-minute dataset.¹⁰ Recall that the scholarly literature demonstrates MCs give one-minute speeches to advertise and position-take on matters important to constituents; I therefore want to link my operationalization to how widespread those concerns are among the general public. A measure of different MCs who speak captures this more directly than counting speeches since one Representative could give the same one-minute speech every day for a month, artificially inflating the count, but doing very little to signal to the Court that their concern is a broader concern. These data are aggregated and lagged at the year level to determine if judicial behavior at time $t$ can be linked to Congressional behavior at time $t - 1$ .

I next add a group of variables to account for other Congressional and public opinion-based information that’s available. Following Clark’s (2009) findings, I include number of court-curbing bills introduced in the prior year by utilizing his initial replication data (101st–110th Congresses) as well as Mark and Zilis (2019)’s updates to the 114th Congress.¹¹ I also include public opinion and institutional constraint variables to control for the Court’s established propensity to act in a legitimacy-protecting manner. I utilize General Social Survey data with Court Disapproval which captures the % of respondents who have “hardly any confidence…in the people running the Supreme Court.” Following Segal et al. (2011), I use Judicial Common Space scores (Epstein et al., 2007) (JCS scores) to calculate Institutional Constraint . When the Court finds itself more ideologically extreme than both branches of Congress, Institutional Constraint is measured as the ideological distance between the Court’s median justice and the median member of the closest Congressional chamber; if the Court is in the middle, however, the variable is coded as 0. As I theorize the Court is concerned with political attacks, I also account for the President’s ideology by using JCS scores to measure the distance between the Court and the President .

Finally, I include a series of variables to control for relationships identified in prior work. I control for the effects of ideology via inclusion of the Martin-Quinn score (MQ score) for the Court’s median. To do so, I use the Supreme Court database’s lcDispositionDirection and petitioner variables to determine whether striking the law in question would produce a liberal or conservative outcome. From here, I align the Court median’s MQ score with that outcome. If, for example, striking the law would produce a liberal outcome and the Court’s median is liberal (and therefore the MQ score is negative), I take the inverse of this score; I therefore expect a positive effect for Court Ideology . Additionally, we know from (Collins, 2004) that amicus participation can impact outcomes, so I create amicus difference which captures the difference between the number of amicus briefs filed on behalf of the party advocating for a strike and for the law being upheld. In this formulation, positive values indicate more briefs favor a strike, so here, too, I expect a positive result. Another known effect is presence of the Solicitor General (Black & Owens, 2012), so I include a dummy variable that indicates whether the SG advocated on behalf of the law. Finally, as the dependent variable is dichotomous, Court decision-making behavior is estimated as a logistic regression model.

Results

The first numerical column of Table 1 depicts the full parameter estimates for my model; it performs well despite the relatively low

N = 171

, and my two speech-based variables of interest are statistically significant and correctly signed. At first glance, these findings support my hypotheses that predict negative (positive) speeches will decrease (increase) the likelihood that the Court will strike federal legislation. In addition to my two variables of interest, Amicus Difference is also significant, indicating that when more friend-of-the-Court briefs support overturning (upholding) a law, the Court is more likely to decide in that direction. Also note that no ideology, constraint, or public opinion variables reach statistical significance. Due to the difficulty in interpreting the coefficients of a logistical regression model, I provide a graphical depiction of my results.

Table 1.

Logistic Regression Results.

MCs giving Neg ${Speeches}_{t - 1}$	−0.100*	−0.100***
MCs giving Neg ${Speeches}_{t - 1}$	(0.039)	(0.026)
MCs giving Pos ${Speeches}_{t - 1}$	0.186***	0.185***
MCs giving Pos ${Speeches}_{t - 1}$	(0.052)	(0.041)
Institutional constraint	−0.123		−2.195
Institutional constraint	(1.456)		(1.222)
Distance bw court and president	0.007		0.402
Distance bw court and president	(0.935)		(0.785)
Court ideology	0.646	0.653	0.647
Court ideology	(0.496)	(0.493)	(0.455)
Dissatisfaction with court	−0.016		0.059
Dissatisfaction with court	(0.072)		(0.077)
Court-curbing ${Bills}_{t - 1}$	−0.041	−0.043	−0.038
Court-curbing ${Bills}_{t - 1}$	(0.025)	(0.026)	(0.037)
Amicus difference	0.106**	0.106**	0.123**
Amicus difference	(0.039)	(0.038)	(0.038)
SG for uphold	−0.432	−0.420	−0.400
SG for uphold	(0.379)	(0.345)	(0.351)
Constant	0.819	0.581	−0.241
Constant	(1.415)	(0.391)	(1.472)
N	171	171	171
Log-likelihood	−100.50	−100.52	−106.09
$χ_{(9,6,7)}^{2}$	45.88	44.68	21.03

Logistic regression of the probability the Court will strike a federal law as unconstitutional. Standard errors clustered on term and reported in parentheses below maximum-likelihood parameter estimates. Significance levels (two-tailed test) : *p < .05, **p < .01, ***p < .001.

Figure 1 depicts the effects of both of my variables of interest, MCs giving Negative Speeches (left panel) and MCs giving Positive Speeches (right panel). First, recall that I pare down each speech variable to drop duplicate observations by MC and sentiment per year. This means my variable captures the number of MCs who give positive or negative speeches per year, and looking at the strong negative and positive slopes of Figure 1, this operationalization captures substantial effects. Beginning with the left-hand panel, the Court becomes less likely to strike a law in year $t$ as the number of MCs who gave negative one-minute speeches about the Court in year $t - 1$ grows. To provide context, if the Court considers exercising its power of judicial review and an average number of MCs gave negative speeches about the Court the prior year, there is a 0.41 [0.32, 0.50] probability the law will be struck. If, however, nine more MCs (or approximately one standard deviation) express negative sentiments about the federal judiciary, this probability drops by nearly 50% to only 0.21 [0.10, 0.38]. The effect is more striking when moving from the minimum to maximum of MCs giving Negative Speeches – this change produces a 93% decrease in the probability of overturning federal legislation (0.60 vs. 0.04).

Figure 1.

Predicted probability curves that the Court will strike down federal legislation as unconstitutional based upon number of Representatives giving negative (left panel) or positive (right panel) one-minute speeches about the federal judiciary. All other variables held at median or modal values. Whiskers represent ninety-five percent confidence intervals.

Moving now to the right side of the figure, MCs giving Positive Speeches also shows a strong slope; despite this type of speech being more rare, its effect on the Court’s decision-making is similar in magnitude. While moving from 0 to 16 positive speeches per year (min to max) produces a 3.3-fold increase in the probability the Court strikes federal legislation, the effect is still strong one standard deviation from the mean. Again, beginning with the mean number of MCs giving positive speeches per year (3.5), and then moving one standard deviation above that, the resulting probabilities change from 0.41 [0.32, 0.50] to 0.58 [0.44, 0.70], or a roughly 41% increase.

To further investigate the robustness of my data, I include two additional models in Table 1. First, in the middle results column, I omit the three variables accounting for public opinion and constraint due to Congress or the President. Note that – in comparison to the full model – omitting these variables results in near identical significance and log-likelihoods, suggesting that – in addition to being statistically insignificant in the full model – these two variables do very little for the model; this is supported by Wald testing.¹² Additionally, I include a model that omits my speech variables to see whether these measures are simply tapping into the same latent concepts captured by the public opinion and constraint-based variables, where these variables may be over-lapping. The far-right column in Table 1, however, shows that my speech-based variables are capturing something unique. Note, especially, the much lower $χ^{2}$ value for this model specification (21.03 vs. 45.88) as well as the fact that none of the public opinion or constraint-based variables gain statistical significance.

Before moving on to discussion, I note that my results are robust to alternate operationalizations of my variables of interest as well as several key control variables. First, results are robust to a logistical transformation of my speech variables and to a count of positive or negative speeches (rather than a count of MCs giving such speeches).¹³ I also test alternates for Institutional Constraint . This includes using Bailey (2007) scores rather than JCS scores, factoring the President in my constraint conditions (rather than using a separate distance variable, as in the model depicted), and specifying constrained and unconstrained dichotomous conditions where the Court is constrained if it sits ideologically outside the two Congressional chambers and the president, and unconstrained if it falls somewhere between any of those three points. In each of these alternate modeling strategies, the constraint variable or variables fail to achieve significance while my variables of interest remain correctly signed and statistically significant. Additionally, while I include court-curbing bills as a count, Clark logistically-transforms bills in his initial study (2009). I chose not to do so since both prior studies most relevant to my work included his variable as a raw count (Ringsmuth & Johnson, 2013; Segal et al., 2011). To ensure this modeling choice isn’t the reason the court-curbing bills variable does not achieve statistical significance, I respecify my model using Clark’s transformed variable $(1 / (1 + \exp (- b i l l s_{t} / 2)) - . 5)$ . This operationalization does not affect results, and the logistically-transformed variable fails to reach statistical significance $(z = - 1.65, p > 0.10)$ . Finally, I also download Clark’s (2009) replication data and then include my one-minute speech variables along with his original data. He estimates five separate Poisson models predicting number of federal laws struck per year, where each model contains his variable of interest – court-curbing bills per year – along with varying operationalizations of public opinion and institutional constraint. I add my one-minute variables to his original models, and they are significant and correctly signed in all five. In sum, my speech variables are robust to a wide variety of different configurations and transformations, and are consistently significant and correctly signed when using alternate control variables, or even in alternate model specifications aggregated entirely to the year-level (therefore containing only 28 observations). These robustness checks help lend confidence to the veracity of my results.

Discussion & Conclusion

This work attempts to further identify and parse separation of powers effects due to the Supreme Court’s concern for its own institutional legitimacy. In line with prior research, my results support the notion that absent explicit constitutional checks to the Court’s power, the justices are still concerned with the “prevailing winds that drive reelection-minded politicians” (McGuire & Stimson, 2004, p. 1019). My specific contribution lies in providing theoretically sound and empirically tested ‘prevailing winds’ in the form of non-legislative, one-minute speaking behavior. While the theoretic appeal of one-minute speeches is found in their clear language, bimodal sentiment (negative or positive), and signal as constituent and party concerns, one-minutes are critically non-legislative. In other words, the speeches, themselves, do not constitute a threat of possible legislative court-curbing actions; this means we can parse the Court’s concern with legislative retribution from its sensitivity to political criticism and praise.

Specifically, I find the Court is willing to change its decision-making behavior in a class of arguably the most consequential cases – those where the Court must decide whether or not to strike down a federal law as unconstitutional. In this class of cases, the Court seems to respond to the broader political environment reflected in Congressional one-minute speeches, seeing a metaphorical green light when more House Members have spoken on the Court’s behalf, and a flashing red when negative sentiment is reflected in one-minute speeches – even when controlling for the effects of political conditions that could actually produce court-curbing legislation. Distinct to my work, however, is the inclusion of positive sentiment and its effect on the Court. Although MCs give positive speeches less often than they criticize, praise packs a powerful punch, as shown in Figure 1. Previous research has identified that a lack of constraint or court-curbing bills results in more (or more likely) invalidations of federal law, however those findings are fundamentally different than mine here, where the presence of positive sentiment impacts Court behavior. The off-setting effect of praise could be a key insight to broader judicial behavior that seems to defy conventional expectations given the Court’s historic low approval ratings.¹⁴

Within these findings, I note that while I include a count of court-curbing bills – a variable theoretically similar to mine – this variable does not achieve statistical significance in my case-level decision-making model. While this is, perhaps, unsurprising given similar non-significant results in prior work (Ringsmuth & Johnson, 2013; Segal et al., 2011), it is worth discussing when one-minute speech – that is non-legislative behavior – is significant instead. I attribute this to the close electoral connection present in the House of Representatives along with the informationally-rich and theoretically appropriate nature of my one-minute speech variables.

I also want to acknowledge the limits of my work. First, this type of nuanced data isn’t always available and is admittedly more time-consuming to gather than voting records or public opinion polls. The emergence of text-processing software – such as topic-modeling, natural language processing, and sentiment analysis – could be employed by scholars to reduce time and effort costs, but careful consideration of the source data is still required to avoid atheoretical data mining. I did not employ such automated tools, but variables built upon behavior rely upon the consistency of that behavior; a second potential limitation to my work is that – unlike roll call votes – consistency is not guaranteed. I build my theory around scholarship that demonstrates MCs use one-minutes as a means of advertising and position-taking on issues important to their constituents, while parties solicit and gate-keep these speeches. This means the utility of my speech-based variables is possibly temporal, constrained to some time period where such conditions are met. Indeed, research indicates that parties only began exerting influence over one-minutes in the 1980s, advancing to “a more formal role when Dick Gephardt and Newt Gingrich ascended to their respective leadership positions within the House of Representatives in 1989” (Hughes, 2016, p. 199). My data align with this period of party oversight, but future House leaders could abandon the strategy, or similarly, future House rule changes could diminish the electoral value of giving one-minute speeches. In short, my speech-based variables (and others similarly derived) are potentially limited in their applications, and so should only be employed after careful consideration of time period and context.

Despite the aforementioned limitations, political behavior is more nuanced and complex than is able to be captured in up or down votes, and so scholars ought to be looking to capture more nuanced information, especially in policy domains that rarely or never get up or down votes. Underscoring this point is the fact that the first two models of decision-making behavior depicted in Table 1 – where the second omits variables that capture public opinion and ideological constraint – are nearly identical: when using data from one-minute speeches about the federal judiciary – data that is informationally-rich and theorizes a strong electoral connection – the effects of ideological distance and public opinion polls disappear. Stated simply, this research seems to bear out what one anonymous justice articulated: “The Court is pretty good about knowing how far it can go…Congress is better than we are, especially the House. They really have their finger on the pulse of the public” (Clark, 2009, p. 973, emphasis added).

Footnotes

ORCID iD

Maron W. Sorenson

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Notes

References

Bailey

M. A.

(2007). Comparable preference estimates across time & institutions for the court, congress, & presidency. American Journal of Political Science, 51(3), 433–448. https://doi.org/10.1111/j.1540-5907.2007.00260.x

Bartels

B. L.

Johnston

C. D.

(2013). On the ideological foundations of Supreme court legitimacy in the American public. American Journal of Political Science, 57(1), 184–199. https://doi.org/10.1111/j.1540-5907.2012.00616.x

Benesh

S. C.

(2006). Understanding public confidence in American courts. The Journal of Politics, 68(3), 697–707. https://doi.org/10.1111/j.1468-2508.2006.00455.x

Black

R. C.

Owens

R. J.

(2012). The solicitor general & the United States Supreme court: Executive branch influence & judicial decisions. Cambridge University Press.

Caldeira

G. A.

Gibson

J. L.

(1992). The etiology of support for the Supreme court. American Journal of Political Science, 36(3), 635–664. https://doi.org/10.2307/2111585

Cardozo

B. N.

(1928). The nature of the judicial process. Yale University Press.

Casillas

Enns

P. K.

Wohlfarth

P. C.

(2011). How public opinion constrains the U.S. Supreme court. American Journal of Political Science, 55(1), 74–88. https://doi.org/10.1111/j.1540-5907.2010.00485.x

Clark

T. S.

(2009). The separation of powers, court-curbing, and judicial legitimacy. American Journal of Political Science, 53(4), 971–989. https://doi.org/10.1111/j.1540-5907.2009.00411.x

Clark

T. S.

(2011). The limits of judicial independence. Cambridge University Press.

10.

Collins

P. M.

(2004). Friends of the court: Examining the influence of amicus curiae participation in U.S. Supreme court litigation. Law & Society Review, 38(4), 807–832. https://doi.org/10.1111/j.0023-9216.2004.00067.x

11.

Epstein

Knight

(1998). The choices justices make. CQ Press.

12.

Epstein

Knight

Martin

A. D.

(2001). The Supreme court as a strategic national policymaker. Emory Law Journal, 50(2), 583–612.

13.

Epstein

Martin

A. D.

Segal

J. A.

Westerland

(2007). Judicial Common Space. Journal of Law, Economics, and Organization, 23(2), 303–325. https://doi.org/10.1093/jleo/ewm024

14.

Ferejohn

Shipan

(1990). Congressional influence on bureaucracy. Journal of Law, Economics, and Organization, 6(Special Issue), 1–27. https://doi.org/10.1093/jleo/6.special_issue.1

15.

Geer

J. G.

(2006). In defense of negativity. The University of Chicago Press.

16.

Giles

M. W.

Blackstone

Vining

R. L.

(2008). The Supreme court in American democracy: Unraveling the linkages between public opinion and judicial decision making. The Journal of Politics, 70(2), 293–306. https://doi.org/10.1017/s0022381608080316

17.

Hansford

T. G.

Damore

D. F.

(2000). Congressional preferences, perceptions of threat, & Supreme court decision making. American Politics Quarterly, 28(4), 490–510. https://doi.org/10.1177/1532673X00028004003

18.

Harris

D. B.

(2005). Orchestrating party talk: A party- based view of one-minute speeches in the house of representatives. Legislative Studies Quarterly, 30(1), 127–141. https://doi.org/10.3162/036298005x201482

19.

Hudiburg

J. A.

(2020). Nonlegislative debate in the U.S. House of representatives. Technical report. Congressional Research Service.

20.

Hughes

(2016). Assessing minority party influence on partisan issue attention in the US House of representatives, 1989–2012. Party Politics, 24(2), 197–208. https://doi.org/10.1177/1354068816654323

21.

Hughes

(2018). Identifying the causes of issue attention & policy change: Evidence from U.S. Offshore oil & natural gas drilling policy, 2008. The Review of Policy Research, 35(1), 170–188.

22.

Hughes

Koger

(2022). Party messaging in the U.S. House of representatives. Political Research Quarterly, 75(3), 829–845. https://doi.org/10.1177/10659129211029712

23.

Kalaf-Hughes

(2020). Representation by the minute: The influence of ethnicity, partisanship and district fit on legislator one-minute floor speeches. Congress and the Presidency, 47(1), 92–129. https://doi.org/10.1080/07343469.2019.1666437

24.

Landis

J. R.

Koch

G. G.

(1977). The measurement of observer agreement for categorical data. Biometrics, 33(1), 159–174. https://doi.org/10.2307/2529310

25.

Maltzman

Lee

Wahlbeck

P. J.

(2004). Supreme court justices really do follow election returns. PS: Political Science & Politics, 37(4), 839–842.

26.

Mark

Zilis

(2019). The conditional effectiveness of legislative threats: How court curbing alters the behavior of (some) Supreme court justices. Political Research Quarterly, 72(3), 570–583. https://doi.org/10.1177/1065912918798501

27.

Mayhew

D. R.

(1974). Congress: The electoral connection. Yale University Press.

28.

McGuire

K. T.

Stimson

J. A.

(2004). The least dangerous branch revisited: New evidence on Supreme court responsiveness to public preferences. The Journal of Politics, 66(4), 1018–1035. https://doi.org/10.1111/j.1468-2508.2004.00288.x

29.

Mondak

J. J.

Smithey

S. I.

(1997). The dynamics of public support for the Supreme court. The Journal of Politics, 59(4), 1114–1142. https://doi.org/10.2307/2998595

30.

O’Connor

S. D.

(1999). Public trust as a dimension of equal justice: Some suggestions to increase public trust. Court Review, 36(3), 10–13.

31.

Owens

R. J.

(2010). The separation of powers and Supreme court agenda setting. American Journal of Political Science, 54(2), 412–427. https://doi.org/10.1111/j.1540-5907.2010.00438.x

32.

Pearson

Dancey

(2011b). Speaking for the underrepresented in the house of representatives: Voicing women’s interests in a partisan era. Political Research Quarterly, 64(4), 910–923. https://doi.org/10.1017/S1743923X1100033X

33.

Pearson

Dancey

(2011a). Elevating women’s voices in congress speech participation in the house of prepresentatives. Politics and Gender, 7(4), 493–519. https://doi.org/10.1177/1065912910388190

34.

Rehnquist

W. H.

(1986). Constitutional law and public opinion. Suffolk University Law Review, 20(4), 751–769.

35.

Ringsmuth

E. M.

Johnson

T. R.

(2013). Supreme court oral arguments and institutional maintenance. American Politics Research, 41(4), 651–673. https://doi.org/10.1177/1532673x12471756

36.

Rocca

(2007). Nonlegislative debate in the U.S. House of representatives. American Politics Research, 35(4), 489–505. https://doi.org/10.1177/1532673x07300233

37.

Sala

B. R.

James

Spriggs

I. I.

(2004). Designing tests of the Supreme court and the separation of powers. Political Research Quarterly, 57(2), 197–208. https://doi.org/10.2307/3219864

38.

Schneider

(2007). One minute speeches: Current house practices. Congressional Research Service.

39.

Segal

J. A.

(1997). Separation-of-Powers games in the positive theory of congress and courts. American Political Science Review, 91(1), 28–44. https://doi.org/10.2307/2952257

40.

Segal

J. A.

Chad

Lindquist

S. A.

(2011). Congress, the Supreme Court, and judicial review: Testing a constitutional separation of powers model. American Journal of Political Science, 55(1), 89–104. https://doi.org/10.1111/j.1540-5907.2010.00480.x

41.

Spiller

P. T.

Gely

(1992). Congressional control or judicial independence: The determinants of U.S. Supreme court labor-relations decisions, 1949-1988. The RAND Journal of Economics, 23(4), 463–492. https://doi.org/10.2307/2555900

42.

Whittington

(2022). The judicial review of congress database (pp. 1789–2022). https://kewhitt.scholar.princeton.edu/judicial-review-congress-database.