How can political and legal decision makers increase the likelihood of a favorable response from other actors? Scholars have identified several mechanisms. Presidential administrations can control resources to exert influence on independent regulatory commissions (Reference MoeMoe 1982). Presidents can staff agencies with appointees who will remain ideologically compatible (Reference Wood and WatermanWood & Waterman 1991). Congress can monitor bureaucratic agencies by responding to “fire alarms” (Reference McCubbins and SchwartzMcCubbins & Schwartz 1984) and can constrain federal judges by passing statutes with more detailed language (Reference Randazzo, Waterman and FineRandazzo, Waterman, & Fine 2006). Courts can achieve higher compliance from executive agencies by writing clear and explicit opinions (Reference SpriggsSpriggs 1996) and can encourage positive treatment from lower courts when they speak with a unified voice in the form of a unanimous opinion (Reference Benesh and ReddickBenesh & Reddick 2002). Within the courtroom, litigants and witnesses can increase their credibility, competency, and trustworthiness if they testify in a “style characteristic of high-status people” (Reference BlackBlack 1989: 18). Given that policymaking typically requires some level of cooperation from another political actor, a vital question facing legal actors is how to ensure a favorable response to their decisions.
We argue that a basic but unrecognized tool is available to policy makers; it involves the certainty or “authoritativeness” of language. Language is important because it is the primary way that political and legal actors communicate with each other. Moreover, institutional and legal actors primarily communicate with each other using written language, which means the receiver of the communicated message must interpret the language before responding to it. Thus, we argue that by varying the degree of certainty or authority expressed in the language, the sender of the message has an opportunity to enhance the favorability of the response.
We apply this theory to a principal–agent framework. Specifically, we examine U.S. Supreme Court opinions, analyzing how lower courts respond to Supreme Court precedents based on the variation in the degree of certainty expressed in Court opinions, though we should note that our findings hold larger implications for use of certainty in other legal settings, which we discuss more at the end. A key part of this hierarchical relationship—or principal–agent relationship—concerns the relevance of “stare decisis.” It requires lower courts to defer to support the outcome and reasoning established by a higher court, irrespective of their preferences. As the Supreme Court has stated, “unless we wish anarchy to prevail within the federal judicial system, a precedent of this Court must be followed by the lower federal courts no matter how misguided the judges of those courts may think it to be” (Hutto v. Davis 1982: 375). Yet, several studies show that lower courts' treatment of Supreme Court precedent is far from perfect, with lower courts citing some precedents positively while citing others negatively (e.g., Reference Hansford and SpriggsHansford & Spriggs 2006; Reference JohnsonJohnson 1979). This suggests there may be something different across opinions—related to the language of the opinion—that enhances or mitigates the response. Understanding this difference becomes vital because the legal reasoning underlying a particular decision is crucial in determining its precedential value and in understanding the evolution of the law. In fact, Reference SpriggsSpriggs (2003) argues that justices care more about substantive policy outcomes and observes that the bargaining over opinions is largely concerned with the language of the opinion. We suggest that the degree of certainty expressed in the opinion is a strong indicator of how lower courts treat Supreme Court precedents.
In this article, our goal is to understand the extent to which the language of the majority opinion influences how that opinion is treated by the lower courts. Do lower courts respond to the content of court opinions? Specifically, if the language of the opinion is more authoritative, are lower courts more likely to positively treat that opinion? In order to answer these questions, we utilize linguistic software designed to assess the certainty—or the authoritativeness—of the words used by the justices. Specifically, we use the certainty of the language expressed in the opinion to capture the authoritativeness of the opinion.Footnote 1 Our approach expands on the growing trend in empirical legal scholarship to employ computer content analysis to understand judicial decisionmaking (e.g., Reference BlackBlack et al. 2011; Reference CorleyCorley 2008; Reference Owens and WedekingOwens & Wedeking 2011; Wahlbeck, Spriggs, & Reference Wahlbeck, Spriggs and SigelmanSigelman 2002). It also has the benefit of taking opinion content seriously, and directly links the content of Court opinions to lower court policy outcomes. Additionally, this research further informs our understanding of how the courts of appeals treat Supreme Court opinions, a significant topic given that the courts of appeals are the de facto courts of last resort in the federal system (Reference Hettinger, Lindquist and MartinekHettinger, Lindquist, & Martinek 2006). We find that the language used in the majority opinion does influence the extent to which the lower courts positively treat Supreme Court precedent. Specifically, we find that as the authoritativeness of the words used in the majority opinion increases, the lower courts are more likely to positively treat the Supreme Court decision. Importantly, we believe our findings about the influence of the language are portable to other broader contexts.
Language and Persuasion
We propose studying the importance of certainty through the principal–agent framework. Principal–agent models were initially developed to examine questions of incomplete information and risk sharing, eventually were applied to organizational theory in economics, and subsequently introduced into other fields (Reference MoeMoe 1984). This approach, which treats the Court as the principal and the Courts of Appeals as the agent, is common within the courts literature (e.g., Reference BeneshBenesh 2002; Reference George and YoonGeorge & Yoon 2003; Songer, Segal, & Reference Songer, Segal and CameronCameron 1994).Footnote 2 Furthermore, courts that are subordinate to the Supreme Court are “subject to an absolute duty to follow its precedents” (Reference KimKim 2007). However, lower federal courts still retain discretion when deciding cases and we argue that a persuasively written opinion is more likely to lead to positive treatment.
An appellate opinion must be written persuasively. Writing a persuasive opinion can hold together a majority coalition or enhance the reputation of the opinion (and its author) in the legal community (Reference HumeHume 2006). Additionally, Supreme Court justices wish to write legally strong, persuasive opinions because they desire to produce good law and good policy and in order to be cast in as favorable a light as possible (Reference Corley, Collins and CalvinCorley, Collins, & Calvin 2011).
Finally, a strong, persuasive opinion can enhance the extent to which opinions are implemented by lower courts. “To properly communicate the disposition of a case the judge must enable the reader to understand and accept the judge's decision. Thus the document communicating that decision must be clear and persuasive” (Reference GeorgeGeorge 2007: 4). The success of this persuasive attempt is largely based on the ability to distinguish opinion from fact and portray assertions in persuasive ways. The Court “can firmly endorse rules or they can equivocate …” (Reference HumeHume 2006: 817). In other words, the Court can choose to use words that reflect a high degree of certainty or a low degree of certainty.Footnote 3 Thus, we argue a key feature of an opinion's level of certainty, or “authoritativeness,” stems from the language used in the content of the opinion. Judges wield enormous power in shaping the law and influencing society and, accordingly, “there is pressure on them to speak decisively” (Reference SolanSolan 1993: 2). Specifically, judges are taught that opinions “should … carry conviction …” (Federal Judicial Center 1991: 19).Footnote 4
This argument is supported by empirical research that finds more assertive messages to be more persuasive (Reference Hazelton, Cupach and LiskaHazelton, Cupach, & Liska 1986; Reference Sniezek and BuckleySniezek & Buckley 1995). This then begs the question, why is certainty important for persuasiveness? Reference Yates, Price, Lee and RamirezYates et al. (1996) attribute part of the answer to extremeness as an indicator of competence when they found that consumers prefer sources that make extreme confidence judgments. In addition, Reference Sniezek and Van SwolSniezek and Van Swol (2001) find that principals who express more confidence are trusted more, even when the agent has less expertise, and this leads to their advice being followed more often. In sum, source certainty is important because it helps resolve that decision maker's cognitive burden.Footnote 5
When viewed from the perspective of persuasion then, certainty is relevant and important in judicial opinions because “[w]riting opinions is a lot like writing briefs. Both are, at bottom, efforts to persuade. Lawyers want to satisfy clients and win cases. Judges want to persuade lawyers, litigants, and the community at large that the decision they have made … is the absolutely correct one” (Reference HigdonHigdon 2010: 1242). Furthermore, the persuasive power of courts is integral for judicial effectiveness because it is the only leverage courts have (see Reference Baird and JavelineBaird & Javeline 2007). Given that courts must rely on other actors to implement their decisions, possessing neither the power of the purse nor the sword (Reference HamiltonHamilton 1788, Federalist Paper 78), courts instead must rely on their persuasive power, which encourages implementation of their decisions.
What are the elements of persuasive legal argument? Reference GardnerGardner (1993) argues that one of the key strategies is “to create the impression that the judge has no choice” (54, emphasis in original). While Gardner admits that judges exercise discretion, he is quick to point out that judges do not like to exercise discretion because it increases the difficulty of the choice (i.e., increases the cognitive burden on the judge). As part of his formula, one of Gardner's key principles is “Establishing Certainty of Authoritativeness” (Reference Gardner1993: 55).
Gardner's advice is joined by many others. Reference Rieke and StutmanRieke and Stutman (1990), in their book on communication in legal advocacy, spend an entire chapter on source credibility where some of the key components are the assertiveness and confidence of the message. For example, Rieke and Stutman write, “when receivers perceive a source to be confident, they confer the source higher credibility” (Reference Rieke and Stutman1990: 120). But this advice is not limited to a handful of texts. Rather, advice like it can be found in almost any guide to legal advocacy or writing. For example, Reference Waicukauski, Sandler and EppsWaicukauski, Sandler, and Epps (2001) write on the importance of a speaker's “ethos,” which is the Greek word for character or credibility, and argue for the importance of “Convey[ing] your Conviction” (Reference Waicukauski, Sandler and Epps2001: 41). Finally, Reference Lebovits and HidalgoLebovits and Hidalgo (2009), advising law clerks how to draft their first judicial opinion, counsel them to “[b]e definite … not tentative” (35) and in The Judicial Opinion Writing Handbook, judges are advised, when writing opinions, to “be definite. …” (Reference GeorgeGeorge 2007: 27). Importantly, while we focus on judges, this is not limited to judges. In his sociological look at the justice system, Reference BlackBlack (1989) finds that how litigants and witnesses speak in court matters for their success and credibility in front of judges and juries. Thus, we argue that the more certainty expressed in Supreme Court opinions, the more persuasive those opinions will be, leading to an increase in compliance.
Although previous scholars have used different measures to capture the authoritativeness of a Supreme Court decision, including whether it was unanimous, whether it was a minimum winning coalition decision, the size of the voting majority, the number of dissenting justices, the number of dissenting opinions, and the number of special concurring opinions (Reference Benesh and ReddickBenesh & Reddick 2002; Reference Hansford and SpriggsHansford & Spriggs 2006; Reference JohnsonJohnson 1979),Footnote 6 those measures have neglected an important feature of our legal system—the content of the opinion. In sum, by using a higher degree of authoritative language in the opinion, the opinion is more likely to be persuasive. This leads to the following hypothesis: As the degree of certainty in an opinion increases, lower courts will be more likely to interpret the precedent positively than negatively or neutrally.
Data and Method
To determine whether the language of the opinion affects the treatment of Supreme Court precedent by the lower courts, we identify and examine lower court treatments of a random sample of 110 Supreme Court cases from the 1976 to 1986 terms.Footnote 7 The unit of analysis is a Court of Appeals decision that has interpreted one of our 110 Supreme Court opinions, which includes circuit court cases from 1976 to 2005. For our analysis, we identified 2,772 Courts of Appeals decisions through Shepard's Citations via Lexis. Shepard's is a legal resource that provides for each Supreme Court decision a list of all the subsequent cases (Supreme Court, Courts of Appeals, District Courts, and state courts) that cite the decision. Although Shepard's does not capture whether lower courts are ignoring Supreme Court precedent, Reference Benesh and ReddickBenesh and Reddick (2002) find that the Courts of Appeals do not disregard precedent they disagree with. In fact, they did not find a single opinion that overtly ignored the overruling decision.Footnote 8
Dependent Variable
Important for purposes here, Shepard's offers an editorial analysis indicating how the subsequent decision (the “citing” case) legally interpreted the previous decision (the “cited” case). The goal of Shepard's is to ascertain whether the precedent is still good law, or whether it has been diminished based on how it is being treated (Reference Hansford and SpriggsHansford & Spriggs 2006). To be judged by Shepard's, the subsequent case must contain specific language that legally interprets the cited case (see Reference Spriggs and HansfordSpriggs & Hansford 2000). In other words, a cited case is not considered to be “legally interpreted” just because it is cited.Footnote 9 Shepard's offers for each citing case the following types of legal interpretations that are relevant to this study: “Question,” “Limit,” “Criticize,” “Distinguish,” “Follow,” “Explain,” or “Harmonize.” Shepard's labels “Followed” as positive treatment, “Explained” and “Harmonized” as neutral treatment, and “Question,” “Limit,” “Criticize,” and “Distinguish” as negative treatment. Although Shepard's codes treatments of precedent in concurring and dissenting opinions, we focus only on treatments that occur in majority opinions.
Shepard's uses “Followed” to indicate that a citing case's majority opinion “expressly” relied on a cited case as precedent (Reference Spriggs and HansfordSpriggs & Hansford 2000: 330). Examples of language that lead to an opinion being coded by Shepard's as “Followed” are “controlling,” or “determinative” or “such a conclusion is required by” (Reference Spriggs and HansfordSpriggs & Hansford 2000: 330). Thus, we code a circuit case that Shepard's indicates “Followed” a Supreme Court decision as positive.Footnote 10 Consistent with Shepard's typology of legal treatment, we code a case that “Questioned,” “Limited,” “Criticized,” or “Distinguished” a Supreme Court decision as negative.Footnote 11 Finally, we code a case that “Explained” or “Harmonized” a Supreme Court decision as neutral. “Explained” indicates that the citing opinion “clarifies, interprets, construes or otherwise annotates the decision in the cited case” and “Harmonized” means “that the cases differ in some way; however, the court has found a way to reconcile and bring into harmony the apparent inconsistency” (Reference Hansford and SpriggsHansford & Spriggs 2006: 44).
Reference Spriggs and HansfordSpriggs and Hansford (2000) empirically tested the reliability of Shepard's analysis of Supreme Court opinions and assessed the validity of Shepard's treatment codes, finding them to be reliable and valid (see also Reference Hansford and SpriggsHansford & Spriggs 2006). Specifically, when collapsing the treatment codes into three broad categories—positive treatment, neutral treatment, and negative treatment—they found that the negative treatment code is the most reliable, and the neutral treatment code is the least reliable, although it is still considered reliable. Thus, we categorize the treatments by the circuit courts into three types: positive treatment, neutral treatment, and negative treatment. In our data, 62 percent of the cases received positive treatment, 13 percent received neutral treatment, and 25 percent received negative treatment. Because our dependent variable is nonordered and categorical, we estimate a multinomial logit (importantly, we get similar results if we estimate an ordered logit).Footnote 12
Primary Independent Variable
Our main independent variable, degree of certainty, measures the degree of certainty expressed in the majority opinion. We generate this measure using the computer content analysis program Linguistic Inquiry and Word Count (LIWC). LIWC is a dictionary-based program, meaning that it contains lists of words that correspond to separate dictionaries that represent a larger concept. Specifically, we use LIWC's dictionary for “certainty,” which we explain more fully below. LIWC was developed by psychologists to measure a variety of things, such as expression of emotions, cognitive thought processes, use of pronouns, as well as several others (Reference Tausczik and PennebakerTausczik & Pennebaker 2010). Using dictionaries, thesauruses, and questionnaires, an initial selection of words for each category was made by research assistants. Groups of three judges then independently rated whether each word was appropriate for that category. Those category word lists were updated and a word remained in the category list if two out of the three judges agreed it should be included, a word was deleted if at least two judges agreed it should be excluded, and a word was added to the category if at least two of the judges agreed it should be added. That process was then repeated by a separate group of three judges.
Dozens of studies have used indicators from LIWC to explain various phenomena, with these results demonstrating predictive validity. Moreover, LIWC's validity and reliability on a variety of its indicators have been established by several studies (e.g., Reference AlpersAlpers et al. 2005; Reference Bandum and OwenBandum & Owen 2009; Reference CohenCohen 2012; Reference KahnKahn et al. 2007). For example, Reference CohenCohen (2012) demonstrates the concurrent validity of LIWC's “certainty” indicator by showing that it correlated with a corpus-based dictionary of cognitive rigidity. In short, LIWC appears to be widely accepted as a text analysis tool. We should note, however, as with any linguistic software program, LIWC has its limitations.Footnote 13
The 2007 LIWC dictionary for the concept “certainty” contains 83 words. Some examples include: absolutely, always, certain*, clearly, commit, completely, every, exact*, extremely, forever, indeed, inevitab*, must, never, perfect*, positiv*, precis*, totally, truly, undeniab*, undoubt*, unquestion*, where the asterisk allows the program to count any variation of the word with that stem. Appendix A contains the full list of words.Footnote 14 The LIWC program works simply by searching the text for these words and counts their occurrence as a proportion of the total number of words, yielding a percentage for each category. In our sample of 110 Supreme Court cases, the certainty category ranges from .61 to 2.34, with a mean of 1.25 and standard deviation of 0.323. Higher values are theorized to measure higher levels of certainty expressed by the writer. While these percentages may seem small at first glance, they are, in fact, larger upon closer inspection. For example, a document with 5,864 words (the mean opinion length in our dataset) with 1 percent “certain” words will have about 59 “certain” words, which roughly equals 3.5 “certain” words per page in the U.S. Reports (this assumes a 17-page opinion with an average of 350 words per page). In short, while 3.5 words per page do not seem overwhelming, it is the repetitive effect over the course of an opinion compared to an opinion with far fewer “certain” words.Footnote 15
To better illustrate how our measure relates to a Supreme Court opinion, we provide an example. Nixon v. Administrator of General Services, 433 U.S. 425 (1977), the legal dispute over former President Nixon's White House tapes, has a certainty score of 1.44, which is slightly above the sample mean. In the majority opinion's discussion of separation of powers and the “abundant” statutory precedent for mandatory disclosure of documents that the Executive branch possesses, consider Brennan's use of the word “never.” “Such regulation of material generated in the Executive Branch has never been considered invalid as an invasion of its autonomy” (446), where “never” is one certainty word that the LIWC program captures. Further, consider this sentence, “As the careful research by the District Court clearly demonstrates, there has never been an expectation that the confidences of the Executive Office are absolute and unyielding” (450, emphasis added). While these two sentences represent only a small sample of what the LIWC program captures, its significance becomes more noteworthy when one considers alternative ways to construct the sentence. For example, Brennan could have simply wrote that “… there is no expectation that the confidences of the Executive Office are absolute and unyielding” and it would have carried the same substantive meaning, but it would have lacked the force that the added certainty brings.Footnote 16
Control Variables
Previous research shows that a number of other factors influence treatment of Supreme Court precedent by the lower courts. The first factor is age of precedent, measured in years. There are two views of how the age of a precedent might influence treatment by lower courts, suggesting competing hypotheses. The first view suggests that older decisions have become fundamental to the Court and lower courts would be more likely to positively treat those cases. The second view argues recent precedents deserve more respect from the lower courts because the Supreme Court is not likely to overturn recently established precedents (see Reference Brenner and SpaethBrenner & Spaeth 1995).
Research also suggests competing directional hypotheses for the complexity of a case. Reference WasbyWasby (1970) views complex decisions as confusing to the lower courts and thus expects them to limit positive treatment. Alternatively, Reference JohnsonJohnson (1987) found that complex decisions were followed more often and Reference Benesh and ReddickBenesh and Reddick (2002) viewed complex decisions as fostering higher levels of positive treatment since they engender a closer reading. For complexity, we count the number of legal provisions relied upon and the number of issues raised in the precedent (Reference SpaethSpaeth 2006).
We also include the ideological consistency with the Supreme Court majority opinion. Ideology influences lower court judges (see, e.g., Reference Hall and BraceHall & Brace 1992; Reference Songer and HaireSonger & Haire 1992), and as the distance between the ideology of the Supreme Court decision and the members of the deciding appeals court panel increases, the likelihood of the panel treating the precedent positively should decrease. We use the Judicial Common Spaces score (Reference EpsteinEpstein et al. 2007a) for each federal court of appeals judge, district court judge,Footnote 17 and each Supreme Court justice, a measure of personal ideology that places them in the same policy space. We take the absolute value of the difference between the median of the appeals court panel and the median of the precedent's majority coalition. This distance should capture whether the appeals court panel is ideologically consistent with the Supreme Court decision.
We also control for case importance. Although some scholars argue that important Supreme Court cases are more likely to be followed by lower courts since they are more visible (see Reference Benesh and ReddickBenesh & Reddick 2002), important cases are also more likely to be controversial, and a number of scholars (Reference BaumBaum 1978; Reference GruhlGruhl 1980; Reference WasbyWasby 1970) have suggested that controversial Supreme Court decisions are more likely to receive negative treatment by lower courts. Thus, we use two measures to tap into the importance of a Supreme Court case.Footnote 18 The first is a measure of political importance, a dichotomous variable coded 1 if the case is a major case using the New York Times measure, and 0 otherwise (Reference Epstein and SegalEpstein & Segal 2000; Epstein et al. Reference Epstein2007b).Footnote 19 The second is a measure of legal importance, also a dichotomous variable, coded 1 if the case struck down a law as unconstitutional or overturned an existing precedent, and 0 otherwise (Reference SpaethSpaeth 2006).
Next, we control for the possibility that lower courts sometimes engage in anticipatory behavior. How far away has the Supreme Court moved from the precedent? The deciding appeals court panel may engage in anticipatory behavior, taking into consideration the ideology of the Supreme Court that is sitting at the time the lower court interprets the precedent relative to that of the Supreme Court that handed down the precedent. This may be because the judges fear reversal by the Supreme Court or because they believe that is their proper role. Reference KleinKlein (2002) found evidence that two federal appellate judges indicated they sometimes engage in anticipatory decisionmaking. In addition, Reference GruhlGruhl (1981) found that federal court decisions were more likely to act in anticipatory compliance. Thus, we control for a change in Supreme Court ideology and we use the same ideology scores as above and calculate the change in Supreme Court ideology from the time of the precedent by taking the absolute value of the difference between the median of the Court sitting at the time the lower court treats the decision from the median justices that issued the precedent. We expect that, as the distance grows, the lower courts will be less likely to treat the Supreme Court case positively.
We also account for the treatment of precedent by Supreme Court. Reference Hansford and SpriggsHansford and Spriggs (2006) found that lower courts respond to how the Supreme Court has interpreted its own precedent. If the Court treats a case positively, by following it and declaring it to be good law, then the authority of the case is enhanced. Conversely, if the Court negatively interprets a case, then the authority of the case is diminished. To account for this, for each Supreme Court case decided during the 1976–1986 terms, we use Shepard's to identify all subsequent Supreme Court cases that positively or negatively treated it. We then count the number of times the Court's majority opinions interpreted the precedent in a positive or negative manner at the time the lower court treats the decision.Footnote 20 We take the difference between the number of prior positive and negative interpretations. Thus, positive values of this variable indicate that the Court has interpreted the precedent positively more often than negatively at the time the lower court treats the decision and negative scores indicate that the precedent has had more negative treatments than positive. We expect that the more often the Court has treated the precedent positively than negatively, the more likely the lower courts will treat the precedent positively.
We also control for how much support the opinion has garnered. Thus, we include a variable, which is equal to 1, if the vote in the case is unanimous, or 0 otherwise.Footnote 21 To control for Supreme Court cases that have been overruled, we include a dummy variable, which is equal to 1 if the case has been overruled by the Supreme Court, or 0 otherwise. We use Shepard's to identify cases that have been overruled and expect that these cases are less likely to receive positive treatment by the lower court.
Another potential factor is opinion clarity. Principal–agent theory suggests that agents have a more difficult time evading a principal's commands when those commands are clear (Reference BrentBrent 2003). In other words, a clearly written opinion leaves less discretion for the lower court. Reference SpriggsSpriggs (1996) found that opinion clarity mattered for agency compliance. In addition, Reference Staton and VanbergStaton and Vanberg (2008) identify that when judges value policy outcomes (rather than managing institutional prestige), judges will value clarity over vagueness (as part of the tradeoff). Additionally, if an opinion is written very clearly, it may be more persuasive. While there are many possible ways to measure an opinion's clarity, for one proxy we use the average number of words per sentence. This is a variant of the commonly used readability measures, which capture surface characteristics of a text (such as the average sentence length) to use as a proxy for the difficulty of reading the text. We expect that as opinion clarity increases, lower courts should treat precedents more positively.
While the readability measure captures one dimension of legal clarity, we also control for a second type of opinion clarity—the attention to detail.Footnote 22 In this Reference Randazzo, Waterman and Finevein, Randazzo, Waterman, and Fine (2006) use a measure of statutory constraint, borrowed from Reference Huber, Shipan and PfahlerHuber, Shipan, and Pfahler (2001) and Reference Huber and ShipanHuber and Shipan (2002), that suggests the more detail provided in a statute will constrain other actors who are responsible for implementing the policy. It is simply the log of the total number of words in the opinion.Footnote 23 We expect clear precedents to be treated more positively. Finally, we also include in the model all of the dummy variables for each circuit, excepting the First Circuit which was used as the baseline, so that each dummy can be interpreted as the impact of a given circuit relative to the First Circuit.Footnote 24
Results
Does more authoritative language affect the treatment of Supreme Court precedent by the Circuit Courts of Appeal? Table 1 suggests that opinions that have a higher level of certainty are more likely to be treated positively by the lower courts.
Table 1. Multinomial Logit Model of the Impact of “Certainty” on Lower Court Compliance
N = 2,772; * P < 0.05, **P < 0.01, ***P < 0.001 (one-tailed tests where directionality hypothesized).
Note: Fixed effects for each circuit are not reported.
Specifically, even after controlling for alternative explanations, the coefficient for degree of certainty is statistically significant. Thus, the fact that the decision is more authoritative appears to affect treatment by the lower courts.Footnote 25 As the level of certainty increases, circuit courts are more likely to treat the majority opinion positively and less likely to treat that opinion negatively. However, the coefficient for neutral treatment is not statistically significant, suggesting that the degree of certainty does not influence whether the lower court is less likely to treat the opinion neutrally. Specifically, as the degree of certainty increases by one standard deviation above the mean, the probability of positive treatment goes from 0.634 (the baseline) to 0.662, an increase of 0.028. The probability of negative treatment drops to 0.218 from 0.251, a decrease of 0.033. When the certainty score is at its highest compared to its lowest, the probability of positive treatment increases from 0.571 to 0.716, an increase of 0.145. This finding suggests that the Supreme Court can increase compliance by using more certain authoritative language in its opinions.
To better illustrate the magnitude of our findings, Figure 1 displays the predicted probabilities of the three types of treatment based on the level of certainty, holding the other variables at their mean or modal values. Each shaded region corresponds to a different treatment: positive, neutral, and negative. Figure 1 supports our hypothesis, as it illustrates that as certainty increases, the probability of a positive treatment increases while the probability for a negative treatment decreases, with the probability of a neutral treatment staying the same. To further highlight the substantive effect a change in certainty might have, consider how a modest change in certainty will increase the number of positive treatments of an opinion. For example, with a one standard deviation increase in certainty (which is approximately 1 more certainty word per page of a 17-page opinion), our model would predict 40 more positive treatments of Supreme Court precedent. This becomes even more substantial when we consider the possibility that one Supreme Court opinion might get multiple positive treatments from several Courts of Appeals decisions.
Figure 1. Predicted Probabilities of Each Type of Treatment.
Our model also includes a series of control variables and there are several variables related both to increased negative treatment and neutral treatment.Footnote 26 Although opinion clarity is not statistically significant (P = 0.072, one tailed), attention to detail is statistically significant and signed in the expected direction. Specifically, more detailed precedents are treated more positively than either negatively or neutrally. As attention to detail increases, positive treatment goes from 0.634 to 0.693, an increase of 0.059. This suggests that opinion clarity, along with certainty, matters when it comes to treatment of Supreme Court precedent by lower courts, which is consistent with earlier research by Reference Huber and ShipanHuber and Shipan (2002), Reference Randazzo, Waterman and FineRandazzo, Waterman, and Fine (2006), Reference SpriggsSpriggs (1996), and Reference Staton and VanbergStaton and Vanberg (2008).
As we noted, scholars have disagreed about whether older or more recent Supreme Court cases are more likely to be treated positively by the lower courts. The results of this study show that the age of the Supreme Court precedent has a positive impact on lower court treatment. As precedents age, lower courts are less likely to treat a Supreme Court decision negatively and neutrally. If the Supreme Court precedent is 15 years old (one standard deviation above the mean) compared with 8 years old (the mean), the probability of positive treatment goes from 0.634 to 0.708, an increase of 0.074. Thus, the results show that older decisions have become fundamental and lower courts are more likely to follow those decisions and less likely to negatively interpret those decisions.
In addition, if the decision has more positive treatments than negative treatments by the Supreme Court, the lower courts are less likely to treat the case negatively. If the Supreme Court has treated its own precedent more positively than negatively, the probability of positive treatment goes up by 0.049 (0.634–0.683).
More complex cases are more likely to be treated neutrally than positive, while legally important cases are more likely to be treated negatively than positively. Additionally, as the difference in ideology between the Supreme Court case and the deciding appeals panel increases, the lower court is more likely to treat the precedent negatively versus positively. However, the political importance of the case does not appear to manifest any systematic influence on the lower courts' treatment of the precedent.
With respect to changes in Supreme Court ideology, the odds of the lower court neutrally treating the case rather than positively treating the case increase. Specifically, the probability of positive treatment decreases by 0.029 when the distance between the ideology of the Supreme Court sitting at the time the lower court treats the precedent and the ideology of the Supreme Court that handed down the precedent increases (0.634–0.605). Thus, the lower court is treating the case in a less positive manner when the Supreme Court has moved away from the precedent. However, the case still stands as precedent. This suggests that the appeals court panel is not more likely to negatively treat the case. Thus, it appears that the circuit courts are somewhat engaging in anticipatory behavior. Finally, if the precedent was a unanimous opinion, lower courts are more likely to positively treat the precedent than negatively treat it, with the probability of positive treatment increasing by 0.047 (0.634–0.681).
Conclusion
Past research was mixed regarding whether the authoritativeness of a Supreme Court opinion influenced lower court compliance. However, those studies defined authoritativeness based on the amount of support the majority opinion had. In contrast, we examine the authoritativeness of the majority opinion based on the language the justices use in the opinion. “Opinion writing is public writing of the highest order; people are affected not only by judicial opinions but also by how they are written” (Reference Lebovits, Curtin and SolomonLebovits, Curtin, & Solomon 2008: 237). Given that the judiciary's power comes arguably from its words alone, it is important to understand how the language the Court uses in its opinions can influence how lower courts treat those decisions.Footnote 27
Lower courts are more likely to positively treat Supreme Court precedents when the precedent contains more certain language, suggesting that there is a connection between the content of court opinions and implementation by other actors. By using more certain language, lower court judges may be more persuaded by the opinion. Although we argue that lexical choice is an important feature of the Supreme Court's persuasiveness, we are by no means assuming that lexical choice is the only tool that judges have to persuade. It is also possible that other nonlexical linguistic features, such as presuppositions used to provide background information on sentences that convey an author's purported assumption that the proposition in question is already assumed true by the addressee, may also persuade lower court judges to treat a Supreme Court precedent more favorably. However, an examination of presuppositions would entail a different type of analysis that requires a close reading of a much smaller number of opinions and is beyond the scope of the current article. Importantly, we believe that both lexical choice and other nonlexical tools, such as presuppositions, sentential syntax, and discourse coherence, all make important contributions to an opinion's ability to persuade.
Additionally, the degree of certainty used may not be completely a conscious judicial strategy. It is also entirely likely that some justices are (just as some laypeople are) more inherently gifted communicators, allowing them to write and speak “automatically” in a manner that is more convincing and certain. Some judges and lawyers undoubtedly have a “gift” or “knack” for phrasing arguments in just such a way that it makes it very difficult to disagree with them. This argument is echoed in Reference BlackBlack's (1989) book where he finds that speaking style matters greatly for the credibility of witnesses who testify. For those where expressing certainty is partly an unconscious act, we think, the legal domain is no different than other domains. The presence (or absence) of this personality characteristic, however, does not diminish the fact that certainty can also be a conscious strategy that is used to try and enhance the persuasiveness of a message.
This raises a couple of important broader questions. Namely, what does certainty stand for? As well as, is it legally relevant and might it apply to other areas of legal decisionmaking? With respect to the first question, a rudimentary examination of the definition of certainty might suggest it stands for a firm conviction or belief that something is reliably true. That someone is willing to phrase an opinion with more certainty not only means that a particular response is desired, it also suggests that the person's reputation and keen judgment are being called to speak for the legal actor. It is another way of saying “trust me” without having to explicitly reference those terms, and its strength lies in the fact that one can use this phrasing of language to communicate with either friend or foe. It is a linguistic mechanism designed to signal that only a certain logic could have led the debate to a particular point and that it leads to only one proper conclusion. In sum, certainty stands for something that can help tip the scales in a case.
As for certainty's legal relevance, we posit that it is especially relevant to appellate court judges and hierarchical relations in the judiciary. Importantly, that is the only relationship we tested for in this article, yet we believe it also applies to other judges (e.g., lower court judges applying a higher court's ruling on the admissibility of evidence) as well as lawyers who might, when advocating for a client, phrase an important case precedent or fact in such a way to make it appear the judge has little other choice but to decide in their favor. We also think certainty is relevant for juries and their decisionmaking, as well as litigants. Consider this quote from Donald Black:
It should finally be mentioned that the success of litigants also depends on how they speak. Recent experiments show that the credibility of people in court increases if they testify in a style characteristic of high-status people. We can distinguish between “powerful” and “powerless” speech by witnesses in courtrooms … those testifying in the powerful mode have more credibility. To a judge or jury, they seem more competent and trustworthy … In various ways, then, how people speak allows the social structure of a case to insinuate itself into the courtroom when it might otherwise be unknown. (Reference BlackBlack 1989: 18–19)
We believe it even applies to broader coverage of social movements, with how the news media cover and frame a reaction, a political protest or social response to a Court ruling. For example, Reference Gamson, Modigliani and BraungartGamson and Modigliani (1987) document the trends in which certain affirmative action frames are used over time by columnists. They find that columnists' usage of the “delicate balance” frame, which argues the government should maintain a proper balance between remedying past discrimination and avoiding future discrimination, peaks at the time of the Bakke decision, something that is notable because of the “balance” that was eventually struck by the Court, striking down racial quotas but allowing the use of race as a criterion in school admissions. This suggests that if the media frames the case in such a way that it emphasizes certainty of an outcome or some ramification, then the case might have a much broader impact, possibly greatly influencing the momentum of a social movement or even change the social structure of future cases.
Furthermore, these findings raise important empirical and normative questions. From an empirical standpoint, given our findings, one might suggest that all justices need to do to increase compliance is to add language to increase the certainty of an opinion. However, as we noted above, if justices want to be regarded as credible and respectable jurists, they need to exercise their own discretion, realizing that sending the signal of high certainty “all the time” will lose its value. In other words, justices need to demonstrate some modesty and temper any inclination that demands perfect compliance.
From a normative perspective, one might wonder whether having an ability to increase compliance (by changing the certainty of the language in an opinion) is a “good” or “bad” thing for the law as well as legal change in society. Although we are not entering the normative debate, we recognize that increasing the certainty of opinion language to ensure compliance may (or may not) have negative consequences that may be intended or unintended. For example, Reference Brewer and BurkeBrewer and Burke (2002) found that a more confident witness was perceived by jurors to be more credible, as indicated by the jurors' higher likelihood of believing a crime was committed, regardless of whether the witness was consistent or inconsistent in testimony (see also Reference Whitley and GreenbergWhitley & Greenberg 1986). In contrast, increased certainty can also have positive effects. For example, it has long been widely accepted that one of the Supreme Court's main purposes is to clarify the law. In other words, when multiple circuit court decisions are in conflict with each other, creating uncertainty in the law, many view it as an important function of the Court to reduce this conflict. In situations where there is lower court conflict, it can be beneficial if the Supreme Court increases its level of certainty in an opinion to better ensure compliance, thus helping to alleviate conflict in the lower courts. In sum, our larger point is to emphasize the importance of documenting and highlighting the presence of the empirical finding of certainty and how it influences judges on the Courts of Appeals.Footnote 28
Beyond this article's primary contribution to a greater understanding of the connection between the language of court opinions and treatment by lower courts, this research corroborates the value of using computerized text analysis to understand judicial opinions. Much can be learned by employing computer-based text analysis programs (Reference Owens and WedekingOwens & Wedeking 2012), such as the LWIC software used here (e.g., Reference Owens and WedekingOwens & Wedeking 2011) as well as other automated methods (e.g., Reference CorleyCorley 2008; Corley, Collins, & Reference Corley, Collins and CalvinCalvin 2011; Laver, Benoit, & Reference Laver, Benoit and GarryGarry 2003). For example, future research might use the LWIC software to evaluate whether more certain language used in parties' briefs leads to more favorable outcomes or whether it influences the extent to which the Supreme Court borrows from the parties' briefs. We believe that the addition of systematic research into this area will provide more insight into understanding how the law is crafted.
Appendix A
List of “Certain” Words in the 2007 LIWC Dictionary
absolute, absolutely, accura*, all, altogether, always, apparent, assur*, blatant*, certain*, clear, clearly, commit, commitment*, commits, committ*, complete, completed, completely, completes, confidence, confident, confidently, correct*, defined, definite, definitely, definitive*, directly, distinct*, entire*, essential, ever, every, everybod*, everything*, evident*, exact*, explicit*, extremely, fact, facts, factual*, forever, frankly, fundamental, fundamentalis*, fundamentally, fundamentals, guarant*, implicit*, indeed, inevitab*, infallib*, invariab*, irrefu*, must, mustn't, must'nt, mustn't, mustve, must've, necessar*, never, obvious*, perfect*, positiv*, precis*, proof, prove*, pure*, sure*, total, totally, true, truest, truly, truth*, unambigu*, undeniab*, undoubt*, unquestion*, wholly.
Multinomial Logit Model of the Impact of “Certainty” on Lower Court Compliance
