This chapter takes a statistical approach to the analysis of the legal rulings that comprise the corpus. The analysis will provide insights into the possible relationships between the variables that make up the corpus, including types of invocations for counsel, the judges’ rulings, the presence of police follow-ups, and the judges’ presidential appointments, among an array of other legal and linguistic factors. To frame the discussion and understand the judges’ application of the law, the chapter starts with a description of the corpus, the selection of variables, and the research questions posed. The latter will be tested further using statistical analysis, specifically a difference in proportion test.

5.1 The Corpus

The corpus consists of 301 criminal federal cases. The search engine used for the collection of the corpus is Westlaw Edge. This search engine is used by law schools and provides a comprehensive database of federal and state rulings, as well as supplementary documents and opinions on an array of legal matters, including the Miranda rights, the invocation of rights, and legal analysis on police interrogation. Both unpublished and published opinions are included in Westlaw and comprise the corpus, albeit most of the opinions in the study are published opinions. An opinion is considered unpublished (not precedential, except when relevant under doctrine of law) if it is not published in the Federal Reporter. Yet, Westlaw publishes most of these opinions online, as well as in a print series called the Federal Appendix. Of note, the distinction between published and unpublished opinions is not relevant to the analysis of the book’s corpus. Both types of opinions were adjudicated, had an impact on the defendant’s legal journey, and reveal the judges’ stance on legal precedence.

The parameters for the search were limited to the three main levels of the federal court system: district courts (the trial court), circuit courts (first level of appeal), and the Supreme Court of the United States (final level of appeal in the federal system). Military courts were also included in the search since custodial interrogations occur in military settings and custodial suspects must be read their Miranda rights.Footnote ¹ Cases that are argued in military courts may be appealed at The United States Court of Appeals for the Armed Forces and up to the Supreme Court of the United States (see Chapter 1 for details on the federal court system).

The search terms include “unequivocal,” “equivocal,” “ambiguous,” and “unambiguous” invocations for counsel. The legal standard is for invocations for counsel to be unequivocal and/or unambiguous, hence these parameters (and their counterparts) capture the legal requirements of invoking counsel in the United States. As noted previously, the courts often do not make a clear distinction between the terms, albeit some courts have addressed them interchangeably. The date parameters for the search terms were 1966–2021. The result of this broad search were 1,050 cases. The search terms, however, included other types of “unequivocal,” “equivocal,” “ambiguous,” and “unambiguous” invocations, such as unequivocal invocations for silence and unequivocal pro se (self-representation) invocations, among others. Of the 1,050 cases, 316 cases fit the (initial) parameters of the search.

The initial corpus of 316 cases was adjusted to 301 cases, after a careful three-step review process aimed at ensuring (single) rater reliability and accuracy of the corpus entries. First, an Excel file was created to manually enter the variables examined in the study – these are described in detail in the next section. Every variable was entered per initial review of each case. Second, every variable entry was rechecked, per case, against the 316-case corpus. In this second review, the researcher noted fourteen double counted case entries in the database. These were removed from the Excel file, since they could alter the quantitative and statistical results. The revised corpus was reduced to 302 cases. Third, before completing the statistical analysis, the researcher performed an additional (final) check to ensure that there were no other double counts or errors in the corpus file. To accomplish this task, all variables were rechecked individually against the corpus to ensure that the entries were accurate and reflected the corpus to be analyzed. One additional error was noted: the inclusion of one state court criminal case. State court cases were not included in the federal courts’ Westlaw search parameters and it is unknown why it was included in the results. The case was subsequently removed from the research corpus. To reduce the possible effect of “tunnel vision” during data entry and when (re)checking for errors, the three-step review process occurred (approximately) one month apart for the first two checks and approximately six months apart for the third and final check. The review of the corpus resulted in an updated 301 case file, which is 29 percent of the total cases (1,050) that fit the broad search parameters of the study. The final corpus of court rulings constitutes a large data set for statistical analysis.

5.2 The Variables Tested

The variables that comprise the corpus are numerous and include: the courts that reviewed the cases, type of interrogator (police interrogator, FBI interrogator, DEA interrogator, and military interrogator), the date of the case ruling (pre-Davis, Davis/post-Davis), unequivocal/equivocalFootnote ² invocations for counsel (per court ruling,) linguistic formulation of invocation for counsel, vulnerable suspects (e.g., L2 speakers, juveniles, suspects’ with limited literacy and/or education), police follow-ups, requests for clarification, Davis citation, case ruling (e.g., affirmed, granted, denied), and presidential appointment of lead judge delivering the opinion of the court (majority, concurring, and dissenting opinion appointments, if applicable).

The quantitative composition of the variables, by group/type, follows, starting with the courts. Table 5.1 provides a summarized quantitative account of the types of courts represented in the corpus and the number of cases viewed by each of the courts.

Of note, the United States Court of Appeals for the Armed Forces is formerly known as the United States Court of Military Appeals. Furthermore, the district court cases listed in Table 5.1 are not divided by region of the court (e.g., eastern, southern, etc.). Appendix 1 includes a complete count of cases per district court and region.

Table 5.2 shows the five types of interrogators featured in the corpus. The numbers match the cases in which a type of interrogator is used, rather than the number of interrogators present in each case. In some instances, more than one type of interrogator questioned a suspect. The presence of a second police interrogator may affect the interrogation and interview process (see Reference Tkacukova, Oxburgh, Mason and RockTkacukova & Oxburgh, 2020), but for the purposes of our analysis the discursive strategies used to change the suspects’ invocation preferences are consistent and complimentary across the corpus. For this reason, and to avoid double counting per case, one type of interrogator is recorded. Of note, if two different types of interrogators (e.g., police interrogator and FBI agent) questioned the suspect, each type of interrogator (combination) is accounted for in the corpus.

As observed in Table 5.2 most cases in the corpus featured a police interrogator. There are a few cases in which the police and agents from the Federal Bureau of Investigation (FBI) jointly participated in an interrogation. Interrogators from the Drug Enforcement Administration (DEA) and military branches of the United States Armed Forces are also accounted for, but they are not as common in the corpus. In the cases that featured military interrogators, the suspects were Spanish speakers and Arab nationals. The latter’s native language is not part of the court record. A few of the military cases included FBI interrogators. It is unknown whether the Reid technique or one of the other methods listed in the High-Value Detainee Interrogation (HIG) Report (see Chapter 2) were used. Yet, the interrogators’ treatment of the invocation for counsel stage was consistent across interrogations irrespective of the interrogation technique employed. This is an insightful finding of the corpus and has important implications when discussing the connection between the law and police interrogation practices during the invocation stage of a custodial interrogation.

In addition to the type of interrogator, the corpus was also divided into two points in time: pre-Davis and Davis/post-Davis rulings. As Table 5.3 shows, most cases fall in the Davis/post-Davis time frame.

The judges’ rulings on type of invocation for counsel, from a legal perspective, are also captured in the corpus. Table 5.4 shows the judges’ rulings on invocations for counsel (unequivocal or equivocal) and the linguistic formulation of the invocations for counsel, if one is provided, per court record. These two variables, judicial ruling and type of linguistic invocation, are distinct and will be analyzed further in the statistical portion of the analysis.

A case may have more than one invocation for counsel. The defendant/petitioner, then suspect, may have attempted to invoke counsel more than once either after the reading of rights (per interrogation) or in the middle of an interrogation. If the case record accounts for more than one attempt at invoking counsel and the invocations were of different types (to avoid double counting) these instances were noted in the corpus. This did not occur often in the corpus. Of 301 cases, there are 329 instances of invocations for counsel, as reported in Table 5.4.

Table 5.4 also shows that most of the invocations were ruled as equivocal. When looking at the linguistic formulations of the defendants’ invocations, as argued in the court record, we see a different story. The defendants invoked counsel using multiple formulations. Some of the formulations were in-dispute/disagreement or not mentioned overtly in the court record. Yet, the defense argued that the suspect invoked counsel. Other invocations, albeit uncommon, included, the defendant requesting a detective or “phone calls,” refusing to sign a waiver form, requesting to be taken to jail, requesting someone other than an attorney (e.g., spouse, parent) or someone invoking counsel on the suspect’s behalf (e.g., juveniles) and what some judges refer to as the “talismanic” request for counsel, such as “I rather talk to an attorney.” Several invocation types were used more often, such as invoking as a direct statement (e.g., “I want an attorney”), invoking as a conditional request (e.g., “If I can call my mom, I’ll talk” or “I’ll talk, but it depends on the question”), invoking as a pro forma, indirect request (e.g., “Can I get an attorney?”), and responding directly to the Miranda waiver (e.g., “Yeah, I would like to do that”).

The most common invocations for counsel are “hedged” invocations, such as “If possible, can I call my lawyer?” and “I may need a lawyer,” and pre-requests (with an inquiry), such as “Should I get an attorney?” and “Do you think I need an attorney?” Invocations for counsel that are hedged were ruled as both equivocal and unequivocal invocations. The latter, particularly, in more contemporary rulings. Pre-requests, or pre-invocations, however, have been predominantly and historically ruled as equivocal, nonrequests. The judges’ rationale with these types of invocations is often that they reflect suspects’ attempts at obtaining information that improves their situation or as failed attempts at invoking counsel. In either assessment, the judges remind us that police have no legal obligation to provide information to, or advise, custodial suspects. The police, in essence, are not a “fiduciary of the suspect.”

As noted in Chapter 4, discursively pre-requests carry an expectation of a follow-up move. This could range from addressing the pre-request specifically, such as “I cannot advise you regarding counsel, but you do have the right to have counsel present during the interrogation,” to informing the suspect that they are unable to respond to their request (they are not legally required to do so), but reaffirming the suspect’s right to counsel. These responses would, at a minimum, although as the corpus suggests probably inadequately, address the suspect’s pre-request, which would allow the suspect to make another attempt at requesting counsel, waive their rights, or simply remain silent.

From the suspect’s perspective, the use of a pre-request, such as “Should I get an attorney?” also suggests that the suspect views the police interrogator as a type of fiduciary: the opposite interpretation of the law, per United States v. Rutledge (1990). In the context of a custodial interrogation, the suspect is often interacting solely with the police and can only seek responses from them, including assessing the question of counsel. This discursive process of seeking information, observed often in talk, including in institutional contexts, has been arbitrarily restricted by the law, as an unsuccessful move to invoke counsel. Consequently, the courts’ treatment of pre-requests may penalize suspects who employ these discursive strategies, benefitting in turn the interrogators who are playing the invocation game. This may affect all custodial suspects, but potentially more so vulnerable ones.

Table 5.5 shows the number of vulnerable suspects and nontraditional suspects (e.g., police officers) in the corpus.

Vulnerable and nontraditional suspects account for thirty-one (31/301) or 10 percent of the corpus. Vulnerable suspects include juveniles, persons with cognitive and/or intellectual disabilities, as well as persons with limited education and low (or no) literacy skills (as noted in the court record). L2 speakers of English are also part of the vulnerable population corpus and account for the largest number of suspects in this category. These suspects included Spanish speakers and a few Arab nationals. Nontraditional suspects, on the other hand, included two police officers. The category “cognitive, educational, and/or intellectual considerations” was noted three times in the corpus (two for juveniles and one for L2 speaker). This is a separate category from the other three, hence, it is accounted for individually. The total number of suspects in Table 5.5 is thirty-one, rather than thirty-four, for the purposes of the quantitative analysis. This classification of the corpus is also accounted for in the statistical analysis.

During the Miranda reading stage of a police interrogation, vulnerable populations, across the board, were subjected to identical interrogation strategies as those used with other suspects in the corpus. Of note, the police officer as suspect, albeit a small sample, is particularly interesting because despite stating knowledge of the law and/or familiarity with interrogations (as noted in the court record) they invoked counsel indirectly and ended up waiving their rights. In essence, the two police officers, as other suspects in the corpus, behaved in a similar manner when subjected to police interrogation, even though the police officers potentially had more knowledge of the Miranda rights and/or police interrogation techniques.

After suspects invoked counsel, the police interrogators often followed up the suspects’ statements or requests. These follow-up turns of talk did not always serve the linguistic function of requests for clarification, hence distinctions were made between follow-up moves and requests for clarification moves, linguistically defined. The frequency of use of these moves is illustrated in Table 5.6.

There were nine instances in the corpus in which the case record did not discuss the police interrogators’ response to the suspects’ invocations for counsel. A transcript of the exchange was not provided, hence it is unknown whether a follow-up or a request for clarification was employed in these cases.

As discussed in Chapter 4, judges often referred to police interrogators’ follow-ups as requests for clarification. Linguistically they were often not formulated as such. Table 5.6 shows the distinction between a discursive follow-up and a request for clarification. This is an important distinction, since the judges in the corpus often argued that despite the police officers not being required to clarify a suspect’s ambiguous/equivocal invocation for counsel, they did so showing due diligence on their part. This argument is particularly meaningful after the Davis ruling whose concurring opinion encourages police interrogators to err on the side of caution and clarify a suspect’s ambiguous/equivocal invocation for counsel.

The corpus also accounts for Davis being (overtly) cited in a ruling. This citation occurred in 219 or 73 percent of the cases. The final case rulings divided by type and frequency are shown in Table 5.7.

The cases sometimes have “universal” rulings that account for all motions and grounds for appeal. Some cases, however, have specific/separate rulings per motion and grounds for appeal. The rulings in Table 5.7 account for those that relate solely to a violation of the right to counsel, such as motions to suppress statements and/or requesting habeas relief due to Miranda violations. In addition, the distinction against and in favor of defendant was made since the appeals, although mostly filed by the defense, also included a small number of cases brought by the prosecution. These include cases that were reversed. Only one ruling that deals with the invocation for counsel matter is noted to avoid introducing extraneous variables that are not relevant to the analysis.

The rulings noted in Table 5.7 include affirming a lower court’s ruling, as valid and correct, dismissing a court ruling without imposing liability on the defendant, denying a motion for the court to review a case on appeal (a court may also rule to deny and dismiss), granting a motion, such as a motion to suppress statements or a writ of habeas corpus (the ruling may be granted in part), remanding the case to a lower court, vacating a ruling making a prior legal judgment legally void, reversing the judgment of a lower court because it was incorrect (the case may also be reversed and remanded), and deferring a judgment to a lower court. Some rulings also included dissenting, concurring, and dissenting/concurring in part opinions, as noted in Table 5.8.

A concurring opinion is an opinion that agrees with the majority opinion but does not agree with its rationale. Concurring opinions are not binding since they do not receive the majority of the court’s support. Dissenting opinions are a separate judicial opinion in which a judge who disagreed with the majority’s decision explains the disagreement. Opinions may also be concurring and/or dissenting in part. These were not common in the corpus. Agreement among the judges seems to be the norm, yet some of the most important cases tried in the Court, such as Miranda and Davis, had concurring and/or dissenting opinions that provide insights into the justices’ viewpoint on the protection of custodial suspects’ rights against the police’s ability to question suspects.

Finally, the analysis also looks at the presidential appointments of the judges who wrote the lead opinion for the courts. These appointments are listed in Table 5.9.

Now, some opinions were en banc (two) or per curiam (six). The former occurs when all judges of a particular court hear a case. The latter is a court opinion issued in the name of the Court rather than specific judges. In three cases tried in a military court, the presidential appointment of the judge who presided the case could not be established. The term “other” accounts for these eleven instances in which there is no presidential appointment.

The decision as to which judge writes the opinion(s) for the court varies by federal court. At the Court level, after the votes have been tallied, the Chief Justice, or the most senior justice in the majority (e.g., if the Chief Justice is in the dissent,) assigns a justice in the majority to write the opinion of the Court or writes the opinion themself. The most senior justice in the dissent can assign a dissenting justice to write the dissenting opinion or write the opinion themself.

At the federal appellate (circuit) court level, cases are decided in a panel composed of three judges. These judges do not always work together, but rather they rotate periodically. The panels are assigned cases using a mathematical formula, so that each judge hears one civil, criminal, and other types of cases within a specified period.

In the federal district court, judges are assigned cases following an order, which may be altered if the case, for example, requires specialization, such as with death penalty cases. A district court judge who specializes in those types of cases may be assigned to hear the case outside the prescribed order.

With regard to the presidential appointment of the lead judge writing the court’s opinion, the most common appointments include: Bill Clinton (17 percent), Ronald Reagan (16 percent), George H. W. Bush (13 percent), and George W. Bush (13 percent). These percentages are reflective of the total judgeship appointments from 1901–2021.Footnote ³ Bill Clinton had the highest number of appointments, followed closely by Ronald Reagan and George W. Bush. Ronald Reagan made four appointments to the Court during his presidency. He also had the highest number of appointments in the Regional Court of Appeals. Bill Clinton made the most appointments to the district courts, with Ronald Reagan and George W. Bush following close behind. These facts are interesting, since the divide between the Miranda and the Davis Court, for example, may have political overtones. The Miranda (majority) opinion was delivered by justices who were deemed more liberal, starting with Chief Justice Earl Warren (Dwight D. Eisenhower appointment), whereas the Davis (majority) opinion was led by more conservative justices, such as Sandra Day O’Connor, who wrote the opinion for the Court, Chief Justice, William Rehnquist, and Justices Antonin Scalia (concurring), Anthony Kennedy, and Clarence Thomas. These justices, except for Justice Thomas (George H. W. Bush), were appointed by Ronald Reagan. The concurring, joining, opinion of the Court was delivered by Justice David Souter (George H. W. Bush), joined by justices Harry Blackmun (Richard Nixon), John Paul Stevens (Gerald Ford), and Ruth Bader Ginsburg (Bill Clinton.) These justices, most of which were appointed by Republican presidents, were considered ideologically moderate (e.g., Souter) or liberal (e.g., Ginsburg).

The presidential appointment of the judges in the corpus may also prove to have statistical bearing, yet it is important to note that some past (older) studies that have examined the possible relationship between Republican presidents, such as Reagan, and the conservative makeup of the federal courts, have shown:

More recent studies on the ideological makeup of federal circuit courts also show that although there appear to be “differences between circuits that we can characterize as ideological … , knowing who appointed the judges on a federal appeals panel, and which circuit the panel is located in, improves our ability to predict case outcomes only marginally” (Reference BroscheidBroscheid, 2011, p. 189). The statistical analysis of the corpus will examine the possible effect of the judges’ presidential appointments on their rulings, including more recent appointments to the federal courts.

Table 5.9 also shows other types of appointments that are not presidential. Two of the judicial appointments were by operation of law. Here a judge who was confirmed by the Senate to serve on a circuit court is reassigned without the need of Senate approval to another circuit court. The corpus also included magistrate judges (14 percent of the corpus).

The next section will consist of a statistical analysis of the corpus. This analysis, and the research questions posed, will reveal the characteristics that make it more or less likely an invocation for counsel will be ruled as equivocal.

5.3 Method and Research Questions

The statistical analysis of the corpus consists of a difference in proportion test. A test of proportion will assess whether a sample from a population represents the true proportion from the entire population. In this test, an interval is created by subtracting the population proportion from the sample proportion plus and minus a margin of error based on a 0.95 confidence interval:

${\overline{p}}_1-{\overline{p}}_2\pm z_{\alpha /2}\sqrt{\frac{{\overline{p}}_1\left(1-{\overline{p}}_1\right)}{n_1}+\frac{{\overline{p}}_2\left(1-{\overline{p}}_2\right)}{n_2}}$

Where ${\overline{p}}_1$ is the subgroup proportion, ${\overline{p}}_2$ is the population proportion, $n_1$ is the size of the subgroup, $n_2$ is the population size, and $z_{\alpha /2}=1.96$ standard deviations associated with a 95 percent confidence interval.

This test is appropriate for analyzing a corpus that consists of binomial data indicating membership (1) or no membership (0) in one or two categories. In the corpus, rulings of unequivocal invocations for counsel (0) and equivocal invocations (1) are the two possible outcomes, since all cases fall under one or the other category (the population). The subgroups (characteristics) of the corpus are represented in the tables and include: case rulings, presidential appointments, interrogators’ use of follow-ups, and linguistic formulation of the invocation for counsel.

The difference in proportion test will measure the effect of characteristics on case outcomes (indirectly) through the measure of invocations for counsel ruled as equivocal. The following questions were tested:

1. (1) Do types of court rulings exhibit a different proportion of invocations being ruled as equivocal than the general population of cases?

2. (2) Are cases involving police asking follow-up questions ruled differently than the population’s proportion, in terms of invocations being ruled equivocal? Are cases involving police requesting clarification ruled differently?

3. (3) Does the proportion of invocations ruled equivocal vary by whether follow-ups or requests for clarification were used by police or nonpolice interrogators?

4. (4) Is there a difference in proportion between invocations ruled equivocal and pre-Davis/Davis-post-Davis for (a) all cases, (b) just police interrogators, (c) nonpolice interrogators?

5. (5) Does the linguistic formulation of the invocation affect the proportion of cases with invocations ruled equivocal?

6. (6) Do cases involving specific subgroups of suspects exhibit a different proportion of invocations ruled equivocal than the population?

7. (7) Does the presidential appointment of the lead judge writing the opinion for the court affect the proportion of cases with invocations ruled equivocal pre- and Davis/post-Davis? Does the presidential appointment of the lead concurring judge and the lead dissenting judge provide different results than those for the lead majority judge?

5.4 Results

There are three possible outcomes with the difference in proportion test: higher than the population proportion, statistically equal to the population proportion (the null hypothesis), or lower than the population proportion. If the low end of the interval is negative and the high end of the interval is positive, the subgroup is not significantly different from the population for this characteristic (null is accepted). If the interval is entirely negative, the subgroup has a significantly lower incidence of the characteristic than the population (null is rejected). If the interval is entirely positive, the subgroup has a significantly higher incidence of the characteristic than the population (null is rejected).

For question 1, the difference in proportion test shows that in none of the comparisons of types of rulings against the population proportion (see Tables 5.4 and 5.7), did any type of ruling deviate significantly from the population. This means that each of the following case outcomes is different from the judicial ruling on unequivocal/equivocal invocation, and the null hypothesis is accepted.

The results for question 2 show that the subgroup proportion is different from the population proportion (see Tables 5.4 and 5.6) with regard to police using follow-up moves, hence the null hypothesis is accepted. When the distinction is made for requests for clarification, the test shows that this subgroup is significantly lower than the population proportion. Thus, cases that have requests for clarification have a lower rate of invocations for counsel being ruled equivocal than unequivocal, as reported in Table 5.10.

Question 3 shows that neither subgroup proportion, except for nonpolice requests for clarification, is different from the population proportion (see Tables 5.2 and 5.4). Of note, the subgroup of military interrogators was excluded due to its small sample. The test on nonpolice requests for clarification (excluding military) shows that cases with this characteristic have a significantly lower rate of having invocations for counsel ruled equivocal. The sample size is small (16), so results may not be robust; that is, if another sample were taken of 301 cases this result may not hold. Still, this result is statistically significant, as noted in Table 5.11, hence the null hypothesis is rejected.

In question 4, the subgroup pre- and post-Davis is not statistically different for invocations for counsel ruled equivocal (see Tables 5.3 and 5.4). The test was repeated for: all interrogators (pre and post), police interrogators (pre and post), and nonpolice interrogators (pre and post). No significant differences between subsamples and population proportions were found. Differences were found for some of the subgroups tested in question 5 (see Table 5.4).

The results for question 5 show that some of the linguistic formulations of an invocation for counsel affect the proportion of cases with invocations ruled equivocal. Each of the linguistic formulations (subgroups) that showed differences with the population proportion will be discussed individually, starting with “invoked other than attorney,” in Table 5.12.

“Invoking other than attorney” is statistically above the population proportion. Thus, fewer cases with this subgroup were ruled unequivocal invocations and more were ruled equivocal. For the custodial suspects this translates to having less of a chance of being successful with this type of argument in court.

“Invoked directly,” on the other hand, is statistically lower than the population proportion, as shown in Table 5.13.

As Table 5.13 reports, the subgroup “invoked directly” is statistically lower than the population proportion. For custodial suspects, fewer cases with this subgroup were ruled equivocal invocations and more were ruled unequivocal. Hence, custodial suspects have a higher chance of being successful with this type of argument in court, a finding that is consisting with prior research.

Regarding “hedged” invocations, this subgroup is statistically higher than the population proportion, as shown in Table 5.14.

These results indicate that “hedged” invocations are ruled more often as equivocal than the population proportion. For suspects who employed this linguistic formulation to make their invocations for counsel, this translates to fewer successful outcomes in court.

Invocations for counsel formulated as “pre-requests” were also common in the corpus. This subgroup was ruled as equivocal at a higher rate than the population proportion, as reported in Table 5.15.

Invocations for counsel formulated as “requested with an interrogative (pro forma)” were also ruled as equivocal at a higher rate than the population proportion. The results are shown in Table 5.16.

The next question addresses whether cases involving specific subgroups of suspects (see Table 5.5) exhibit a different proportion of invocations ruled as equivocal than the population. The results are reported in Table 5.17.

The results for question 6 show that for L2 speakers of English invocations for counsel were ruled more often as directly formulated. Of note, the sample size of this subgroup was small and the discursive behavior of the police interrogators was also fairly egregious, as discussed in Chapter 4. For the other subgroups in the category of nontraditional suspects (including and excluding police officers as suspects), the null hypothesis was accepted.

The final question addresses whether the presidential appointment of the lead judge writing the opinion for the court affected the proportion of cases with invocations ruled as equivocal pre- and Davis/post-Davis (see Tables 5.3, 5.4, and 5.9). This test was repeated for lead judge writing concurrent and dissenting opinions. Presidential appointments were grouped as either Democrat or Republican for three categories: Majority, Concurring, and Dissenting opinions. Care was taken to eliminate Magistrate Judge, Military Judge, Per curiam, and Operation of Law appointments and decisions. As noted in Table 5.18, only post-Davis Democrat appointed judges writing concurrent opinions were different from the null. For this subgroup, the invocations for counsel were ruled as equivocal at a significantly higher rate than the population proportion. This subgroup included only six cases, so small sample concerns prevail.

One remaining question arises with the presidential appointment of the lead judges in the corpus: whether judges appointed by a particular president had a different proportion of rulings of invocations ruled equivocal than the population proportion. The results show that judicial rulings did not vary by presidential appointment when each president was tested individually (e.g., Reagan, Clinton, Obama, etc.). This finding seems consistent with prior research that has shown that despite Republican judges being more conservative than their Democratic counterparts overall, the presidential appointment does not appear to be a significant factor.

5.5 Interpreting the Results in the Context of the Invocation Game

The broader interpretation of these results, in the context of the invocation game, is that often suspects’ invocations for counsel are ruled as equivocal. This occurred with a variety of linguistic formulations in the corpus, including pro forma invocations for counsel and pre-requests. The latter formulation is of particular interest since it is observed often in the corpus and requires a follow-up move to achieve mutuality of understanding and move the conversation forward. The law and interrogation training may block this next move to reaching common ground, unbeknownst to the suspects who, despite the opinion of some judges, may view the police interrogator as a fiduciary of the Miranda rights.

The effect of judges’ ruling suspects’ invocations as equivocal/ambiguous has an additional potential legal cost, if the case goes to trial. As noted previously it is not uncommon for judges to argue that even if the trial court erred by admitting a defendant’s statements, the alleged error does not have a substantial and injurious effect or influence in determining the jury’s verdict (see Brecht v. Abrahamson, 1993; Fry v. Pliler, 2007). This argument, which is not uncommon in the corpus, has an opposite side: the guilty verdict is directly attributable to the defendant’s statements, hence the effect on the jury’s verdict may be substantial (see Kyger v. Carlton, 1998).

Given the role of suspects’ statements in the investigation of a crime and subsequently in the conviction of such crime, the “preponderance of evidence” argument needs further consideration within the invocation game of custodial interrogation. Otherwise, a suspect’s attempt at invoking counsel during custodial interrogation may in effect be nullified a second time in court, if the harmless error perspective is not revisited in light of the high proportion of invocations for counsel, even those that are formulated pro forma, being ruled as equivocal in the corpus.