Home Articles Reader Opinion Editorial Book Reviews Discussion Writers Guide About TCRecord
transparent 13
Topics
Discussion
Announcements
 

The Garcetti Effect and the Erosion of Free Speech Rights of K–12 Public Education Employees: Trends and Implications


by Lewis M. Wasserman & John P. Connolly - 2017

Background/Context: Pickering v. Board of Education, decided by the U.S. Supreme Court in 1968, is considered the high-water mark in the constitutional protection of public employee free speech. Two significant decisions issued by that Court since Pickering have limited public employees’ expressive rights: Connick v. Myers and Garcetti v. Ceballos.

Purpose/Objective: The principal research questions which were the focus of this study are: is the adverse effect on free speech presumed by legal analysts following Connick and Garcetti having real effects in terms of judicial voting behavior and, if so, how has this occurred?

Research Design: We set up legal precedent and the judge’s ideology as predictors of judicial voting behavior. The legal precedent variable delineates three intervals, namely the Pickering era [1968–1983], the Connick era [1983–2006) and the Garcetti era [2006–2014]. Two different measures of ideology are considered: party affiliation and the judge’s DW-NOMINATE score. The dependent variable is the judge’s individual vote in each case. Votes are categorized as pro-employer or pro-employee.

Data Collection and Analysis: Because our interest is specific to K–12 settings we analyze 507 judicial votes drawn from the 169 U.S. Courts of Appeals employer-employee free speech cases brought by school employees between the issuance of Pickering in 1968 and the post-Garcetti period into 2014. Since our dependent variable is dichotomous the statistical estimates are obtained via logistic regression.

Conclusions/Recommendations: We find that employee-plaintiffs are prevailing at lower rates at the U.S. Courts of Appeals in free speech claims against public school districts following the Garcetti decision. Since Garcetti, Courts of Appeals judges—those appointed by both Republican and Democratic presidents—have voted in an increasingly pro-employer direction. We contend this results from “doctrinal signaling,” here, the progressive curtailment in the Supreme Court of public employees’ free speech rights from Connick v. Myers in 1983 to Garcetti. We suggest this “signaling” is used by Courts of Appeals judges as an interpretive tool to give meaning to apparently ambiguous decisions such as Garcetti. Finally, the impact and implications of this negative trend for K–12 employees, as well as possible solutions, are considered.



INTRODUCTION


More than seventy years ago the Supreme Court concluded that freedom of speech enjoys a “preferred position” among basic liberties in the U.S. constitutional scheme.1 Yet the Supreme Court’s decisions during the last forty years indicate that its support for this First Amendment right has eroded in the case of public employees.2 Garcetti v. Ceballos, decided in 2006,3 is an extension and, arguably, the culmination of this longstanding trend.4


To date there has been little research to validate whether this decline in free speech protection posited by others5 also applies in public school settings, often assumed to be the quintessential locus of training for democratic ideals,6 and a hotbed for First Amendment controversies.7 Here we identify three key U.S. Supreme Court decisions and examine their effects on the voting of Courts of Appeals judges in cases specifically concerning K–12 public education. Voting trends at the level of the U.S. Courts of Appeals are of paramount interest, as “the courts of last resort for the vast majority of litigants and, hence for the vast majority of contested legal issues.”8 Judicial choices at this level therefore serve as a bellwether of the force of legal precedent.


Our statistical results provide evidence that the curtailment of free speech rights by the upper court is having a systemic effect on the deliberations of Courts of Appeals judges, Republican and Democratic appointees alike. While it is by now a truism that judges may factor their own policy preferences into their legal deliberations, a focus on such ideological distinctions can be misleading when both Republican and Democratic appointees move in the same direction. That they have here done so here demands inquiry and explanation.


We begin by analyzing the three principal free speech cases, placing special emphasis on the Garcetti opinion and what we see as the most significant arguments made by Garcetti dissenters. We then consider the role of the Courts of Appeals in the federal system to give context to the empirical analysis which follows. After providing statistical evidence that both Republican and Democratic appointees have voted in a more conservative direction since Garcetti, we discuss issues of interpretation concerning these findings and briefly consider Garcetti’s impact, its implications for academic freedom, and possible remedies.


FREE SPEECH IN THE PUBLIC WORKPLACE

THE LEGAL ENVIRONMENT PRIOR TO THE 2006 GARCETTI CASE

In Pickering v. Board of Education (1968), the Supreme Court formally recognized that the Free Speech Clause of the First Amendment protects expression pertaining to public policy concerns uttered by public employees.9


Pickering involved a teacher who was fired for sending a letter to a local newspaper criticizing the manner in which school officials had allocated school funds for sports at the expense of classroom instruction.10 The Court held the firing violated the First Amendment. Justice Marshall, writing for the Court, observed that while “[T]he State has interests as an employer in regulating the speech of its employees,” the Court was obliged “to arrive at a balance between the interests of the teacher, as a citizen, in commenting on matters of public concern and the interest of the State, as an employer, in promoting the efficiency of the public services it performs through its employees.”11


Although the Court found Pickering’s interest to be decidedly greater than the board of education’s, the Court did not explain in much detail how to go about conducting the balancing protocol it decreed. It was not clear when speech was important enough to rise to the level of a “public concern,” or which metrics to apply in weighing the competing interests involved. The fact that Pickering was such an easy case may have contributed to the vague character of the ruling, leading in turn to the exercise of broad discretion by lower court judges.12


The Court refined its Pickering “balancing” approach in Connick v. Myers.13 In Connick, an assistant district attorney, Myers, angry over a transfer to a different section of the office, circulated a memorandum soliciting the views of other employees concerning transfer policy, the level of morale, and the need for establishment of a grievance committee.14 After being fired, Myers sued alleging a violation of the First Amendment. The Supreme Court ruled against Myers, emphasizing that the speech was not protected because it did not involve matters of public concern. Justice White asserted that “[When] employee expression cannot fairly be considered as relating to any matter of political, social, or other concern to the community, officials should enjoy wide latitude in managing their offices, without intrusive oversight by the judiciary in the name of the First Amendment.”15


Connick, then, assigned to the judiciary the job of line-drawing between protected and unprotected speech based on its connection to the public interest. Whether speech was constitutionally protected was determined by the content, form, and context of a given statement,16 hardly definitive criteria. Scholars analyzing numerous lower court decisions applying Connick’s public concern” test have criticized its indeterminacy,17 concluding that “little more emerges from a review of the [Connick-based] decisions than an identification of the variables, often contradictory, which influence whether or not speech is protected.”18 Despite its ambiguities, Connick was still a blow to free speech by limiting the scope of protection established in Pickering, and its impact continues to be felt in First Amendment conflicts,19 including those arising in public education.20


GARCETTI V. CEBALLOS

THE MAJORITY OPINION

Garcetti v. Ceballos, a 5-4 decision, is the most recent case in this trilogy and the most controversial. Richard Ceballos was employed as a deputy district attorney in the capacity of calendar deputy at the Los Angeles District Attorney's office.21 Ceballos found that a sheriff misrepresented facts in a search warrant affidavit,22 and communicated the details to his supervisor in the form of a disposition memorandum—something he was required to do as part of his routine job duties. After these concerns were ignored, Ceballos testified on behalf of the defense in the case.23 In the aftermath, Ceballos claimed he was subjected to a series of retaliatory employment actions and initiated an employment grievance, which was denied.24 He then filed a federal action alleging that the District Attorney and others violated his First and Fourteenth Amendment rights in retaliation for his disposition memo.25


In dismissing Ceballos’s First Amendment complaint, the Supreme Court concluded that Ceballos spoke as an employee and not as a citizen when he wrote the memo to his supervisor recommending dismissal of the case,26 establishing the principle that speech made “pursuant to official [job] duties” is unprotected.27 Here, “official duties” are not necessarily those explicitly enumerated in formal job descriptions but rather the functions employees are expected to perform by their employers as part and parcel of their workplace duties.28


Garcetti therefore created a categorical exemption to free speech so that Courts of Appeals judges are now required to perform a multi-stage analysis when considering free speech claims, reflecting the Pickering, Connick, and Garcetti precedents (Figure 1).


Figure 1. Decisional Process for Lower Court Judges in Public Employee Free Speech Claims Under Current Constitutional Standards

[39_21787.htm_g/00002.jpg]


The majority opinion in Garcetti acknowledged the value of “exposing governmental inefficiency and misconduct”29 but pointed to “the powerful network of legislative enactments–such as whistle–blower protection laws and labor codes–available to those who seek to expose wrongdoing”30 as evidence that mechanisms were in place to protect legitimate whistleblowers. In the Court's view, these whistleblower protection laws, “as well as obligations arising from other applicable constitutional provisions and mandates of the criminal and civil laws,” provide adequate protection to employees in exposing unlawful or inappropriate actions by supervisors.31


DISSENTING VOICES

Justice David Souter authored a vigorous dissent to Garcetti’s majority opinion, in which he was joined by Justices Stevens and Ginsburg.32 Among the concerns raised by Souter were that the phrase “pursuant to official duties” was based on an artificial distinction between public employees who speak as private citizens and those who express themselves as pubic employees about the work they perform,33 and that the Garcetti formula is a disservice to individual government employees and the public they serve.34


Souter proposed that an adjusted Pickering balancing rubric should apply to cases like Garcetti, by which the employee would be protected in raising matters of “unusual importance” while “satisfying high standards of responsibility in the way” such matters were raised.35 Eligible subject matter satisfying the “unusual importance” standard would include only: (1) official dishonesty; (2) deliberately unconstitutional action; (3) other serious wrongdoing; or (4) threats to health and safety.36 This approach would provide advance notice to the government and employees about what kinds of speech are protected.37 Thus, according to Souter, public employees should enjoy presumptive First Amendment protection when they address these and similarly important matters, since it is public employees who “are often in the best position to know what ails the agencies for which they work.”38


In bolstering its argument, the Garcetti majority contended that aggrieved employees could rely on whistleblower statutes for protection, rather than the First Amendment.39 However, Souter criticized this assertion on both legal and factual grounds. He noted the great variation in the protection afforded by state whistleblower statutes, depending on the level and type of governmental involvement defending the claim,40 a view corroborated by analysts examining this issue in public education settings.41 These analysts found that only Illinois has a comprehensive whistleblower statute specifically geared toward public educators.42 Thus, uncertainty about which categories of state and local employees are protected is manifest, despite the optimism expressed by Justice Kennedy in his majority opinion.43


EMPIRICAL ANALYSIS OF JUDICIAL VOTING TRENDS

EMPIRICAL STRATEGY

Although there has been much commentary on the Garcetti ruling,44 to date there is an absence of quantitative analyses to support the key claims. We focus on two questions: first, whether there has indeed been a statistically distinguishable change in voting propensities across the three legal eras demarcated by the key upper court decisions; and second, whether any such changes are systemic in the sense of transcending ideological influences that might be assumed to be at work in the cases under consideration. The latter question is especially relevant in assessing whether changes in voting trends simply reflect a change in the ideological complexion of the courts, or whether legal precedents, Garcetti in particular, have had their own independent effect on judicial voting habits.


Since our concern here is specific to the field of education, we analyze the 507 judicial votes drawn from a total of 169 K–12 U.S. Courts of Appeals employee-employer free speech decisions.45 The data are drawn from all published and indexed cases brought by school employees in the Westlaw data base decided from 1968 into 2014,46 and determined on the merits.47


LEGAL ANALYTIC FRAMEWORK

A core institutional characteristic of the U.S. Courts of Appeals is the random assignment of judges to panels to decide cases.48 The appeals process is specifically structured to promote a collaborative form of decision making.49 Three appellate judges are assigned to decide a case together who typically share their background research before sitting together as a panel to hear oral argument. After then meeting to discuss their views they, usually, issue a single opinion to resolve the appeal.50 Unlike the Supreme Court, the jurisdiction of the Courts of Appeals is mandatory.51 This leads to their deciding many uncontroversial matters, which minimizes disagreement among the judges. The Courts of Appeals are also less likely to alter established legal doctrines since their decisions are subject to Supreme Court review.


The Pickering, Connick, and Garcetti decisions delineate three intervals in which to analyze Courts of Appeals’ voting in free speech cases from 1968 into 2014. The three intervals represent respectively 12%, 60%, and 28% of the total number of judicial votes cast in these cases. Since Connick narrowed public employees’ free speech rights relative to Pickering, and Garcetti further narrowed First Amendment protection after Connick, we expected decisions in the United States Courts of Appeals to respond to these precedents by showing an upward trend in pro-employer voting habits over the three periods and, indeed, our data exhibit this pattern (Figure 2).


Figure 2. Distribution of Votes by Legal Era (1968–2014)

[39_21787.htm_g/00004.jpg]


During the post-Garcetti period approximately 82% of the U.S. Courts of Appeals votes have been against plaintiffs who asserted infringement on their K–12 workplace free speech rights. This contrasts with 61% of the votes during the Connick period and 54% during the Pickering period.  Further, when examined from the perspective of case outcomes, only eight of the forty-eight cases (17%) decided during the Garcetti era were favorable to the plaintiff in contrast to 41% of the cases in the Connick era and 48% of the cases in the Pickering era (Figure 3). We now turn to an inferential analysis of voting patterns in these cases.52


Figure 3. Distribution of Case Outcomes by Legal Era (1968–2014)

[39_21787.htm_g/00006.jpg]


METHODOLOGICAL APPROACH

We begin with a baseline model that tests only for legal precedent effects and then consider the effects of political ideology as a force impacting judicial votes over and above that of legal precedent alone. Accordingly, for the baseline model, we include a categorical indicator of the era in which the Courts of Appeals’ decisions fell; that is, whether between Pickering and Connick, Connick and Garcetti, or subsequent to Garcetti into 2014. The rationale for this baseline model is that legal precedent is the overarching predictor of judicial voting habits. The primary purpose here is to verify that judicial voting habits have indeed moved in an increasingly pro-employer direction since the Garcetti decision. Inclusion of these baseline predictors for legal precedent in the subsequent models then allows one to see to if the effects of legal precedent persist when other predictors are included in the model, especially those pertaining to judicial ideology.


We then proceed to test different measures of political ideology. Our interest in political ideology stems from the well-known contention that Courts of Appeals judges may have different policy preferences and, as such, may react quite differently to the legal precedents set by the U.S. Supreme Court. We rely especially on the findings of Sunstein, who reported significantly greater conservative voting by Republicans than Democrats appointed to U.S. Courts of Appeals between 1925 and 2002 in economic and labor disputes,53 and Clarke who observed a marked and statistically significant decrease in employer victories in labor and employment disputes in cases appealed by employees to the Fourth Circuit following the seating of President Obama’s nominees.54 According to Sunstein’s and Clarke’s analyses, we should expect Republican Party appointees to be relatively less sensitive to the needs of workers and more likely to identify with the institutional employer, and, again, our data exhibit this anticipated pattern (Figure 4). Over the entire period of the cases under consideration, 73% of votes by Republican Party appointees were in favor of the employer while just 58% of votes by Democratic Party appointees were in favor of the employer.


Figure 4. Distribution of Votes by Party Affiliation (1968–2014)


[39_21787.htm_g/00008.jpg]


Other considerations underline, however, why the impact of ideology merits empirical analysis. First, the effect size of political ideology in Courts of Appeals’ voting is expected to be generally less than one might find at the Supreme Court, since Supreme Court justices enjoy greater discretion in deciding cases. Second, because of the centrality of free speech in the U.S. constitutional scheme, such rights may usurp conventional conservative and liberal voting patterns observed in other contexts. Third, and perhaps most germane to our line of inquiry here, is that a focus on coarse ideological distinctions alone could potentially be misleading if both conservatives and liberals have been moving in a more conservative direction over time.


Two alternative measures of political ideology are considered here since both are widely used in judicial behavior research. The first is a categorical indicator of whether the justice was nominated by a Democratic or Republican President. The second is a theoretically more nuanced measure of ideology qua ideology using (DW-NOMINATE) continuous ideological scores assigned to the judge, based on that of the nominating President or the score(s) of the home-state Senator(s) whenever Senatorial courtesy is operative.55 These scores range from -1 (extreme Liberal) to +1 (extreme Conservative). The distribution of scores in our dataset is quite symmetrical. The mean score for Republican appointees is 0.32 while that for Democratic appointees is -0.32, with 53% of scores on the conservative side of the continuum.


Figure 5. Distribution of DW-NOMINATE Scores by Party Affiliation

[39_21787.htm_g/00010.jpg]


Although the Connick and Garcetti opinions may have made it more difficult for employees suffering adverse employment actions to secure First Amendment protection, it does not follow that the proportion of subsequent rulings favoring the employer would necessarily increase, since litigating habits may have in turn changed, so that only stronger cases under then-applicable doctrine were pursued. As such, the overall effect of legal precedent, such as that established in Garcetti is, if anything, understated in the analysis which follows, since we focus here only on observed outcomes.


Our dependent variable is a dichotomous indicator of whether the individual judicial vote was pro-employer, coded as a “1,” or pro-plaintiff, coded as a “0” since, in all the cases, the issue is whether a public employee’s free speech rights were violated by the employer. The coding of the dependent variable therefore entails no issues of subjective judgment. Finally, because our dependent variable is dichotomous, our statistical estimates are obtained via logistic regression with clustered standard errors to address potential correlated data patterns within each of the federal circuits.


EMPIRICAL RESULTS

We ran five models in total (summarized in columns in Table 1), each building from a baseline model, then adding predictors for political ideology as well as for interactions between political ideology and legal era.


Table 1. Logistic Regression Estimates for Predicting a Pro-Employer Vote (1968–2014)

 

(1)

(2)

(3)

(4)

(5)

      

Legal Era:

     

   Connick (CO)

0.27

0.11

-0.21

0.08

 
 

(0.42)

(0.41)

(0.41)

(0.40)

 

   Garcetti (GA)

1.35***

1.22**

1.06**

1.17**

1.10***

 

(0.48)

(0.50)

(0.51)

(0.49)

(0.38)

      

Party Affiliation (PA)

0.68***

-0.08

  
  

(0.11)

(0.36)

  
      

DW-NOMINATE score (DW)

 

0.91***

1.02***

    

(0.20)

(0.25)

      

PA × CO

  

0.95**

  
   

(0.38)

  
      

PA × GA

  

0.63

  
   

(0.64)

  
      

DW × GA

    

-0.40

     

(0.51)

      

Constant

0.16

-0.04

0.18

0.33

0.40***

 

(0.33)

(0.36)

(0.34)

(0.31)

(0.12)

      

N

507

507

507

506

506

      

AIC

631.07

620.80

622.21

620.47

620.17

BIC

643.76

637.71

647.59

637.38

637.07

Note: Standard errors adjusted for twelve clusters in circuits

*** p < 0.01, ** p < 0.05, * p < 0.1


According to our baseline model (Model 1), which tests only for legal precedent effects, the Garcetti era (2006 into 2014) has indeed the strongest effect on voting propensities. The baseline model is then augmented with a predictor for the party affiliation of the judge who cast the vote (Model 2). In this second model, the Garcetti effect is shown to be robust when this additional predictor is included in the model; thus, legal precedent has its own independent effect. However, the effect of political ideology is also significant: over the entire period of the cases under consideration (1968 into 2014), the predicted probability of a Democratic appointee voting in a pro-employer direction is 0.58 as opposed to a predicted probability of 0.73 in the case of Republican appointees.


Our next step is to test for an interaction effect between legal precedent and party affiliation to examine whether the voting propensities of both Republican and Democratic appointees have changed across the three legal eras (Model 3). Here we compute the estimated probabilities of voting in a pro-employer direction based on the logistic regression estimates. These indicate that, since Garcetti, the predicted probability of a Democratic appointee voting in a pro-employer direction has risen to 0.78, while that of a Republican appointee has risen to 0.86 (Figure 6, derived from the model 3 estimates). This result suggests that the increasing tendency to vote in favor of the employer following Garcetti transcends the effect of ideology (as operationalized in terms of party affiliation). Thus the observed trend in voting habits cannot plausibly be explained by a change in the political complexion of the federal courts—since both Republicans and Democrats have voted in a more pro-employer direction since Garcetti.


Figure 6. Predicted Probability of a Pro-employer Vote by Party Affiliation and Legal Era (1968–2014) with 95% Confidence Intervals

[39_21787.htm_g/00012.jpg]


Thus far, ideology has been operationalized in terms of a binary measure of party affiliation. In the fourth model, we substitute a more sensitive measure of ideology, coded in terms of continuous (DW-NOMINATE) scores. Our results illustrate that, in the cases under consideration, the two ideology measures are essentially interchangeable in terms of predictive power, with the continuous measure offering only a slight improvement in model fit (compare Model 4 with Model 2),56 and, again, the Garcetti effect is robust regardless of the measure of ideology.


Our final model tests for an interaction effect between legal precedent and ideology, using the more sensitive measure of the latter (Model 5). Here, to simplify, we focus only on the Garcetti effect, since this is the only legal precedent effect that has been statistically distinguishable throughout all the previous models. Interacting the continuous measure of ideology with the Garcetti precedent tells a consistent story in terms of the significance of Garcetti. Across the full range of political ideology scores, we observe a moderate ideology effect pre-Garcetti (Figure 7, derived from Model 5 estimates). However, the more telling finding is that, post-Garcetti, voting tendencies across the ideological spectrum, from the most liberal of Democrats to the most conservative Republicans, became more favorable to employers.57

Figure 7. Predicted Probability of a Pro-employer Vote by Ideology Pre- and Post-Garcetti (1968–2014) with 95% Confidence Intervals


[39_21787.htm_g/00014.jpg]


In sum, these results indicate that Courts of Appeals judges’ increasing tendency to vote in favor of the employer over employee-plaintiffs transcends political ideology, whether operationalized in terms of party affiliation or in terms of ideology qua ideology. This finding is especially noteworthy. Garcetti’s somewhat ambiguous language might have presented an added incentive for Courts of Appeals judges to interpret the law in terms of their own ideological leanings and policy preferences.58 Instead, judges, irrespective of party affiliation or ideology, are now siding with the employer more often, on average, than before, and, further, have been essentially indistinguishable in this tendency following Garcetti. We now consider plausible interpretations of this result.


ISSUES OF INTERPRETATION

In treating the individual votes as our unit of analysis, we have necessarily abstracted away from all the particularities associated with each case. However, each legal case will differ in certain characteristics that could only be ascertained by performing a case-by case qualitative analysis, in which case we would no longer be in the domain of inferential statistics.


To some extent, such a qualitative analysis of post-Garcetti cases has already been performed by Bauries and Schach albeit on a limited selection of cases. From their analysis, Bauries and Schach concluded that the current “dominant trend” is to interpret the Garcetti ruling in a way that makes it more difficult than necessarily intended by the Supreme Court for plaintiffs to secure First Amendment protection.59 This, they argue, is due to a systematic misinterpretation of the Garcetti ruling that effectively broadens its scope; specifically, by interpreting “the words ‘pursuant to’ [one’s official duties] as meaning either (or both) ‘related to’ or ‘in the course of.’”60 They assert, instead, the correct legal interpretation of Garcetti’s exclusion of speech “pursuant to official duties” is that it applies only if the expressive activity is required by the job.61


Our study draws on this kind of analysis but also differs in crucial respects. The claim of a “dominant trend” yielded, for our purposes, a specific a priori hypothesis for our baseline statistical model (Model 1, Table 1); namely, that the probability of a judge voting in favor of the employer would increase following Garcetti. Our statistical analysis, which steps back and looks at voting trends across the three precedential periods focusing specifically on free speech cases in the field of K–12 education, provides direct statistical support for this claim.


We emphasize that this finding has value in itself, even if there remain open several possible theories of the precise mechanism by which Garcetti has influenced judicial voting patterns. For example, without incorporating the legal principles applied in each case in our analysis, we have no way of knowing if the upward trend in pro-employer outcomes is due to (1) cases specifically failing to meet the Garcetti hurdle, or (2) cases being interpreted more favorably on behalf of the employer even when the case is not ruled specifically on Garcetti grounds. As shown in Figure 1, the Garcetti precedent is merely the first of three obstacles facing a plaintiff, and we do not claim that unsuccessful plaintiffs are all necessarily failing at this first hurdle.


In employing inferential statistics, our purpose is to leverage this tool to help adjudicate between the plausibility of alternative theories. In this respect, our own study seems to immediately rule out one possibility; namely that the increasing success of employers in free speech cases is due to a change in the political complexion of the federal courts. As we have shown, both Democrats and Republicans have been voting in favor of the employer at a greater rate since Garcetti.


However, on the possibility that the upward trend in employer victories following Garcetti is due to certain idiosyncratic legal features of the cases in that period, we wish to emphasize two points. First, as we argued earlier, since Garcetti can only be interpreted as a restriction (rather than an expansion) on free speech, it would seem to, if anything, discourage potential plaintiffs from litigating. To the extent that this is true, the kinds of cases that would be brought before the court would be ones with firmer legal grounds. As such, we think it more likely that we have, if anything, underestimated the Garcetti effect.


Second, to focus on whether the free speech cases in question are failing specifically under the Garcetti test—i.e., whether the speech in question is pursuant to one’s job duties—risks overlooking a larger issue. Even if we assume, for argument’s sake, that these free speech cases were failing, not under the Garcetti test, but under the Connick test, then the natural question to arise would be: Why are cases failing the Connick test at a greater rate than before? We would then have to consider this fact in the context of Bauries and Schach’s findings from their legal analysis, which indicates that many judges are committing the same interpretive error in applying the Garcetti ruling.


In fact, both possibilities can be accounted for with the same explanation. We see no reason to disagree with commentators who assert that the effects of a Supreme Court decision may depend on the degree to which it sets out a clear rule which limits Courts of Appeals’ decision making.62 Where Courts of Appeals justices have such clear rules upon which to rely, they will tend to act within those constraints, voting in accord with those rules,64 and indeed arguably do so in the vast majority of cases.65 However, when faced with ambiguous precedents, they must resort to other tools with which to resolve the controversy.  The question here is why a clear dominant trend ensued in response to such a seemingly ambiguous decision.


Here we contend that the judges observed the same trends in employee free speech as the commentators, and used this information as a guide to their decision making. We think the most plausible explanation lies in the concept of “doctrinal signaling,” a phenomenon that emerges in response to the unclear open-ended characteristics of rulings such as Garcetti. This would in turn explain why judges have read more into the Garcetti decision than was intended by its authors.


SUMMARY AND DISCUSSION

We find that employee-plaintiffs are facing increasing difficulty in securing First Amendment protection in free speech disputes in K–12 education settings. While legal analysts have contended that the 2006 Garcetti precedent has impacted judicial voting decisions in a pro-employer direction, we provide a stronger inferential basis for these contentions through our statistical analysis of cases decided by U.S. Courts of Appeals arising in K–12 settings. Further, we find that this increasing tendency to favor the employer in such disputes transcends both the party affiliation and political ideology of federal appellate judges.


We propose that the most plausible political-psychological theory of this change in voting patterns lies in the concept of “doctrinal signaling;’’ specifically, Courts of Appeals judges have observed successive curtailments of the free speech rights of plaintiffs imposed by the upper court and, as such, have “got the message” and reacted accordingly. An attractive feature of this theory which adds to its plausibility is that it explains both why judicial voting has been moving in a direction favorable to the employer and why judges may be reading beyond the strict meaning of the Garcetti ruling in its application.


The 2006 Garcetti case is in many ways the culmination of a long-standing trend in the history of First Amendment protection following the 1968 Pickering case, the highpoint in such protection. In exempting speech “pursuant to one’s duties,” it added one more categorical exemption to those previously recognized by the Supreme Court, such as “true threats,”66 obscenity,67 “fighting words,”68 “incitement to imminent lawless activity,”69 and “defamation.”70 A natural question that arises is: What does all this mean in terms of implications and possible remedies for K–12 employees?


Examples of Garcetti’s impact on K–12 educational settings include the case of a special education teacher who emailed the chancellor to report that the principal had interfered with his job responsibilities by pressuring him to misrepresent students’ performance and test results (Mpoy v. Rhee71), and a public–school teacher who commented that the classroom was unsanitary (Masarro v. N.Y.C. Dep’t of Education72). In both cases, the court gave weight to the fact that the educator communicated internally rather than going outside the chain-of-command, in turn invoking Garcetti to exclude speech made “pursuant to” one’s job duties from First Amendment protection. As in classroom teacher cases, the protection for administrative speech under Garcetti also remains uncertain. The discretion afforded courts in deciding whether speech falls within or outside an administrator’s duties is quite broad and risks subjecting public employees to idiosyncratic applications of the Garcetti formula.


In 2014 the Supreme Court decided Lane v. Franks.73 The Franks Court held that a community college administrator’s testimony in a federal criminal trial about a state legislator, who received salary for a no-show job, was not categorically excluded from First Amendment protection under Garcetti. Thus, even though the information about which Lane testified was acquired in the ordinary course of performing his auditing duties, the speech itself fell outside the scope of the ordinary functions he performed. Accordingly, the speech was entitled to First Amendment protection.74 Although Lane may have gone some way to help clarify the Garcetti phrase, “pursuant to official duties,” the general picture is still one of legal uncertainty and indeterminacy in an area concerning a basic constitutional liberty.75


The issue of academic freedom, broadly conceived, is yet more complicated. In his majority opinion in Garcetti, Justice Kennedy declined to determine whether the “pursuant to official” duties doctrine applied in challenges to academic freedom since that specific issue was not before the Court.76 Nevertheless, he suggested in dicta that the Court might be open to such an exception, at least in higher education teaching and research.77


However, Kennedy’s musings on this point should probably not be read too broadly or optimistically, given the Court’s historical position on these issues. Justice Frankfurter famously declared that academic freedom, “is an atmosphere in which there prevail ‘the four essential freedoms' of a university--to determine for itself on academic grounds who may teach, what may be taught, how it shall be taught, and who may be admitted to study.”78  Thus, Frankfurter conceived of educational institutions, not individual teachers, as having the right to determine the substance and style of a curriculum.79 According to this formulation, it would seem to follow that such autonomy as exists as a constitutional right in K–12 settings would belong to school boards or their executive officers and not the teachers they supervise.80


In this sense, Garcetti’s exemption from First Amendment protection of speech which employees are hired to perform cements the Court's previous pronouncements about academic freedom, in holding that the school, not the teacher, is the arbiter of classroom speech.81 Cases such as Mayer v. Monroe County Community School Corp.82 and Evans-Marshall v. Board of Education83 are illustrative of this position. In truth, the Supreme Court has never recognized academic freedom as a constitutional right for K–12 teachers while they furnish established curriculum to their students. This fact, coupled with the decline of educators’ expressive rights during the last forty plus years, as evidenced in our statistical analysis,84 gives little reason to believe such rights will expand any time soon.


Yet if the option of litigating cases to secure First Amendment protection now faces seemingly insurmountable hurdles, other possible avenues remain open for those seeking more protection for K–12 employees. It is basic in the U.S. constitutional scheme that the rights and liberties established by the Constitution only set a floor below which no state or local educational agency may go, so that state legislatures and constitutions themselves have discretion to offer greater rights to educators in their states than the federal constitution affords, as long as any new provision is not inconsistent with the United States Constitution.85 This leaves the door open to various alternative remedies to litigation, such as amending state Whistleblower statutes, ensuring adequate protection through contractual provisions, including collective bargaining agreements and, where possible, expanding or litigating the meaning of State Constitutional Free Speech provisions. While these themselves are non-trivial solutions and, indeed, would entail large-scale support and coordination to put into practice, they may for now be the most effective options for those who view the nature of K–12 education quite differently from the federal courts.


METHODOLOGICAL APPENDIX

All of the statistical analysis is coded and executed using Stata. Because the dependent variable is dichotomous we use logistic regression to estimate the effects of our predictors. Clustered standard errors are requested to address potential correlation of outcomes within the various legal circuits.


Two of the models include interaction effects. As interpretation in terms of “statistical significance” is problematic in such cases in the context of logistic regression,86 we focus instead on the estimated marginal probabilities that are generated from the logistic regression estimates (Figures 6 and 7).


The purpose of each of the models is as follows:


Model 1 (baseline model): Here we test if voting propensities differ across the three legal eras: Pickering, Connick, and Garcetti.


Model 2: Here we test if political ideology (operationalized as party affiliation) is a significant predictor of voting propensities, and if legal precedent continues to have an effect when this additional predictor is included in the model.


Model 3: This is the same as Model 2, except that we also include an interaction effect between legal precedent and party affiliation. This allows us to test whether the effect of party affiliation changes across the three legal eras. These estimates are then used to estimate the marginal probabilities, depicted in Figure 6.


Model 4: This is the same as model 2, except that we operationalize ideology in terms of continuous DW-NOMINATE scores rather than party affiliation. This allows us to test whether a more sensitive measure of ideology makes any substantive difference in terms of predictive power and model fit.


Model 5: Here we test whether the effect of ideology (operationalized by the DW-NOMINATE continuous score) changes after Garcetti. Note here, the Connick indicator is excluded from the predictors so that we can compare votes in the Garcetti era against all the votes that came before it, i.e., the reference category in this model includes both the votes in the Pickering era and the votes in the Connick era. These estimates are then used to estimate the marginal probabilities, depicted in Figure 7.


The coding of the variables is as follows:

Dependent Variable

Vote: is coded 1 if the judge voted in a pro-employer direction or zero if the judge voted in a pro-employee direction.


Independent Variables

Legal Era: is a nominal variable coded 0 if the vote was cast in a case during the Pickering era (from 1968 up to Connick in 1983), 1 if cast during the Connick era (from Connick up to Garcetti in 2006), and 2 if cast during the Garcetti era (2006 into 2014). The Pickering era is the reference category.


Party Affiliation: is coded 1 if the judge was appointed by a Republican President, and zero if appointed by a Democratic President.


DW-NOMINATE Score: This is a continuous score ranging from -1 to +1.87


Notes


1. Murdock v. Pennsylvania, 319 U.S. 105, 115 (1943).

2. See, e.g. Adam Shinar, Public Employee Speech and the Privatization of the First Amendment, 46 Conn. L. Rev. 1, 3-4, 70 (2013).

3. 547 U.S. 410 (2006).

4. See generally, Thomas J. Keenan, Circuit Court Interpretations of Garcetti v. Ceballos and the Development of Employee Speech, Note, 87 Notre Dame L. Rev. 841 (2011).

5. See generally, Shinar supra note 2; Nancy M. Modesitt, The Garcetti Virus, 80 U. Cin. L. Rev. 137 (2011-12); Matt Wolfe, Does the First Amendment Protect Testimony by Public Employees?, 77 U. Chi. L. Rev. 1473 (2010); John Q. Mulligan, Huppert, Reilly and the Increasing Futility of Relying on the First Amendment to Protect Employee Speech, 19 Wm. & Mary Bill Rts. J. 449 (2010-11); Sheldon Nahmod, Academic Freedom and the Post–Garcetti Blues, 7 First Amend. L. Rev. 54 (2008); Elizabeth M. Ellis, Garcetti v. Ceballos: Public Employees Left to Decide “Your Conscience or Your Job”, 41 Ind. L. Rev. 187 (2008); Charles W. “Rocky” Rhodes IV, Public Employee Speech Rights Fall Prey to an Emerging Doctrinal Formalism, 15 Wm. & Mary Bill Rts. J.  1173 (2007); Sonya Bice, Tough Talk, 8 J. L. Soc’y 45 (2007); Helen Horton, Constraining Public Employee Speech: Government Control of its Workers’ Speech to Control its Own Expression, 59 Duke L. J. 1 (2009).

6. See, e.g., West Va. Bd. of Educ. v. Barnett, 319 U.S. 624, 637 (1943) (“That they are educating the young for citizenship is reason for scrupulous protection of Constitutional freedoms of the individual, if we are not to strangle the free mind at its source and teach youth to discount important principles of our government as mere platitudes.”); Tinker v. Des Moines Indep. Sch. Dist., 393 U.S. 503, 506 (1969) (“teachers do not shed their constitutional rights to freedom of speech and expression at the schoolhouse gate.”).

7. See, e.g., Rosina E. Mummolo, The First Amendment In the Public School Classroom: A Cognitive Theory Approach, 100 CORNELL L. REV. 243, 256-262 (2014).

8. Frank B. Cross, Decision Making in the U.S. Circuit Courts of Appeals, 91 Ca. L. Rev. 1457, 1459-60 (2003); see, also, Cass R. Sunstein, David Schkade, Lisa M. Ellman & Andres Sawicki, Are Judges Political? (2006); Virginia A. Hettinger, Stefanie Linquist & Wendy L. Martinek, Judging on a Collegial Court (2006); Joan Biskupic, Analysis: Republicans Lead Obama in War for Judicial Dominance, http://www.reuters.com/assets/print?aid=USBRE89405E20121005 (October 2, 2012) (observing that since Supreme Court “justices hear less than 1 percent of appeals, [Courts of Appeals] often have the last word on a case and in establishing legal principles.”).

9. 391 U.S. 563 (1968).

10. Id. at 564, 566, 571.

11. Id. at 568. The Court observed that Pickering’s “statements are in no way directed towards any person with whom appellant would normally be in contact in the course of his daily work as a teacher. Thus no question of maintaining either discipline by immediate superiors or harmony among coworkers is presented here. [Pickering’s] employment relationships with the Board and, to a somewhat lesser extent, with the superintendent are not the kind of close working relationships for which it can persuasively be claimed that personal loyalty and confidence are necessary to their proper functioning. Accordingly, to the extent that the Board's position here can be taken to suggest that even comments on matters of public concern that are substantially correct … may furnish grounds for dismissal if they are sufficiently critical in tone, we unequivocally reject it.” Id. at 569-70.

12. Recent public education cases applying Pickering balancing include: Leslie v. Hancock, 720 F.3d 1338 (11th Cir. 2013), on remand, 994 F.Supp.2d 1339 (M.D. Ga. 2013); Craig v. Rich Twp. High Sch. Dist. 227, 736 F.3d 110 (7th Cir. 2013); Dixon v. Univ. of Toledo, 702 F.3d 269 (6th Cir. 2012), reh’g and reh’g en banc denied (Feb. 27, 2013), cert. denied, 134 S.Ct. 119 (2013); Evans-Marshall v. Bd. of Educ. of Tipp City Exempted Village Sch. Dist., 624 F.3d 332 (6th Cir. 2010).

13. 461 U.S. 138 (1983).

14. Id. at 141.

15. Id. at 146 (footnote deleted) (emphasis added).

16. Id. at 147.

17. See, e.g., Cynthia L. Estlund, Speech on Matters of Public Concern: The Perils Of An Emerging First Amendment Category, 59 Geo. Wash. L. Rev. 1, 34-35, 44 n. 249 (1990).

18. Stephen Allred, From Connick to Confusion: The Struggle to Define Speech on Matters of Public Concern, 64 IND. L.J. 43, 81 (1988); see also, R. George Wright, Speech on Matters of Public Interest and Concern, 37 DE PAUL L. REV. 27, 29 (1981) (reaching similar conclusion).

19. See, e.g., DeCotliis v. Whittmore, 635 F.3d 22 (1st Cir. 2011).

20. See, e.g., Nagle v. Marron, 663 F.3d 100 (2d Cir. 2011) (teacher's reporting of assistant principal's forgery of teacher's signature on teaching observation report of her class, even if such conduct were criminal, was not a matter of public concern); Johnson v. Poway Uni. Sch. Dist., 658 F.3d 954, 965 (9th Cir. 2011) (no First Amendment violation occurred where district required math teacher to remove banners promoting religion from classroom wall, since setting was curricular in nature; content is the most important factor in determining whether employee speaks on a matter of public concern);  Desir v. City of New York, 453 F. App’x 30 (2d Cir. 2011) (even if former home instruction teacher’s speech addressed organizational problems with home instruction program and not just personal matters, it did not address matters of public concern where his speech fundamentally concerned his own entitlement to privileges as home instructor); Lee v. York Cnty. Sch. Div., 484 F.3d 687 (4th Cir. 2007)  (materials posted on classroom bulletin board by public high school Spanish teacher were curricular in nature, and thus, were not speech on matter of public concern so as to be protected under First Amendment; although materials, including National Day of Prayer poster, articles about political figures' religious beliefs and activities and about former student's missionary activity, and rescue mission newsletter, were not directly related to Spanish curriculum, they were school-sponsored speech bearing imprimatur of school, since they were displayed in compulsory classroom setting). But see, Goudeau v. E. Baton Rouge Parish Sch. Dist., 540 F. App’x 428 (5th Cir. 2013) (finding that elementary school teacher's complaints to school officials about school principal's directive to artificially inflate student's grades, in violation of state law and school board policy, addressed a matter of public concern; allegations involved an ongoing practice, not an isolated incident of directing teachers to change students' grades in violation of law, and went beyond teacher's personal employment grievance).

21. 547 U.S. 410 (2006).

22. Id. at 414-15.

23. Id.

24. Id. at 415.

25. Id. at 415-17.

26. Id.at 421-22.  

27. “We hold that, when public employees make statements pursuant to their official duties, the employees are not speaking as citizens for First Amendment purposes, and the Constitution does not insulate their communications from employer discipline.” Id. at 421.

28. Id. at 424-25.

29. Id. at 425.

30. Id.

31. Id.

32. Id. at 427-444.

33. Id. at 429.

34. Id. at 427-429.

35. Id. at 435.

36. Id.

37. Id.

38. Id. at 429 (quoting Waters v. Churchill, 511 U.S. 661, 674 (1994)).

39. Id. at 429-30.  

40. Id. at 439-441.

41. See, e.g., Brenda R. Kallio & Richard T. Geisel, To Speak or Not to Speak: Applying Garcetti and Whistleblower Laws to Public School Employee Speech, 264 Ed. L. Rep. 517,524 (2011) (observing lack of specific protection for educational employees).

42. Id.

43. Garcetti, 547 U.S at 425-26.

44. See, e.g., Scott R. Bauries & Patrick Schach, Coloring Outside the Lines: Garcetti v. Ceballos in the Federal Appellate Courts, 262 Ed. L. Rep. 357 (2011); Keenan supra note 4; Mummolo supra note 7 ; Shinar supra note 2; see, also, Elizabeth Dale, Employee Speech and Management Rights: A Counterintuitive Reading of Garcetti v. Ceballos, 29 BERKELEY J. OF EMP. & LAB. L. 175 (2008); Lawrence Rosenthal, The Emerging First Amendment Law of Managerial Prerogative, 77 FORDHAM L. REV. 33 (2008); Robert C. Post, Subsidized Speech, 106 Yale L.J. 151, 164 (1996).

45. If a three-judge appellate panel issued a decision that later was reheard en banc or became the subject of a dissent from the denial of rehearing en banc, we included only the voting of the original three-judge panel. We made this choice because one of our primary interests was the behavior of three-judge panels and how their voting was influenced by our other independent variables. Inclusion of en banc panel voting would necessarily distort this analysis. Moreover, there were very few en banc decisions to consider.

46. Our dataset has sufficient numbers of votes to study the Courts of Appeals judges’ behavior without having to include district court voting. Although we recognize that published decisions have precedential value within the circuit, while unpublished decisions do not, the fact that a decision was published or not has no bearing on our analysis, since our interest is in Garcetti’s impact on how judges voted.

47. We considered cases decided on the merits where they resolved the adequacy of a complaint, granted or denied summary judgment, or gave judgment to a party after trial based on First Amendment legal criteria. We excluded from our data, dismissals based solely on procedural rules or justiciability grounds.

48. Lee Epstein, William M. Landes & Richard A. Posner, The Behavior of Federal Judges, A Theoretical & Empirical Study of Rational Choice 153 (2013).

49. See, Harry T. Edwards, The Effects of Collegiality on Judicial Decision Making, 151 U. Pa. L. Rev. 1639, 1656 (2003); Pauline T. Kim, Deliberation and Strategy on the United States Courts of Appeals: An Empirical Exploration of Panel Effects, 157 U. Pa. L. Rev. 1319, 1320-21 (2009), citing Jonathan Matthew Cohen, Inside Appellate Courts: The Impact of Court Organization on Judicial Decision Making in the United States Courts of Appeals ch. 2 (2002) (describing in detail the organizational structure of the federal appellate courts).

50. Cohen, supra note 49, at ch. 5.

51. See, 28 U.S.C. • 1291 (2014) (establishing obligatory jurisdiction).

52. See, generally, Epstein et al., supra note 48, 17-24, 153-205; Gregory C. Sisk & Michael Heise, Ideology “All the Way Down”? An Empirical Study of Establishment Clause Decisions in the Federal Courts, 110 Mich. L. Rev. 1201, 1207-11 (2012), (each describing research methodologies applicable to this type of inquiry).

53. Sunstein et al, supra, note 8, at 150-60.

54. Brian S. Clarke, Obama Courts? The Impact of Judicial Nominations on Court Ideology, 30 J. L. & POL. 191, 207-19 (2014) (illustrating the ideological changes in employment cases following the Obama appointments).

55. This is based on the same formula employed in Sisk and Heise, supra note 52 at 1222-26, where judges are assigned the DW-NOMINATE score of the appointing President if there is no Senator of the same party in the judge’s state, the score of the Senator if there is one of the same party as the appointing President for the judge’s state, or the mean of the scores of both Senators if there are two Senators of the same party as the appointing President for the judge’s state. However, whereas Sisk and Heise use “common-space” DW-NOMINATE scores [id.], we use the specific scaling for the Senate, since House congresspersons play no role in the judicial appointment process.

56. We report the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC) for which smaller values indicate better model fit. See, J. SCOTT LONG, REGRESSION MODELS FOR CATEGORICAL AND LIMITED DEPENDENT VARIABLES ch. 4 (1997) (describing these information measures in the context of hypothesis testing and goodness of fit).

57. Figure 7 also suggests that Democratic and Republican appointees have become indistinguishable in voting habits since Garcetti in K–12 free speech cases. However, we estimate that we do not have a sufficient number of votes in the post-Garcetti period (144) to substantiate that claim due to a potential lack of statistical power. In any case, our main interest here is in modeling the changes in voting habits across, rather than within, legal eras. The substantive finding here is that both conservative and liberal judges have shown a marked increase in pro-employer voting decisions since Garcetti.

58. See, e.g. Bauries & Schach, supra note 44; Dale, supra note 44 (discussing interpretive controversies with the Garcetti decision) Sisk & Heise, supra, note 52 (describing the effects of ambiguous Supreme Court Establishment Clause decisions on ideological voting).

59. Bauries & Schach, supra note 44, at 383-86 (discussing Garcetti’s effect on curtailing freedom of speech in public workplace).

60. Id. at 369, 372-73, 379—82, 385.

61. Id. at 364, 368, 370-83 (deriving Garcetti’s holding by the usual rules of construction).

62. See, e.g., Sisk & Heise, supra note 52, at 1251-53.

63. Id. at 1250.

64. See supra note 5 (collecting articles reflecting policy and other concerns post-Garcetti).

65. See, Sunstein et al., supra note 8, at 326-27 (“The difference between Republican and Democratic [Courts of Appeals] appointees is statistically significant. But this apparent difference is only of technical interest, since both groups of judges vote to uphold nearly 100% of the time, and panels vote to uphold at least 97% of the time regardless of which combination of judges sits on a panel …”).

66. Virginia v. Black, 538 U.S. 343 (2003).

67. Miller v. California, 413 U.S. 15 (1973).

68. Chaplinski v. New Hampshire, 315 U.S. 568 (1942).

69. Brandenburg v. Ohio, 395 U.S. 444 (1969).

70. N.Y. Times v. Sullivan, 376 U.S. 254 (1964).

71. 758 F.3d 285 (D.C. Cir. 2014).

72. Masarro v. N.Y.C. Dep’t of Educ., 481 F. App’x 653, 656 (2d Cir. 2012).

73. 573 U.S. __ , 134 S.Ct. 2369 (2014).

74. The Court asserted that sworn testimony in judicial proceedings is a quintessential example of citizen speech for the simple reason that anyone who testifies in court bears an obligation, to the court and society at large, to tell the truth. That obligation, the Court, said was distinct and independent from any separate obligations a testifying public employee might have to his employer. Garcetti said nothing about speech that relates to public employment or concerns information learned in the course of that employment. The critical question under Garcetti is whether the speech at issue is itself ordinarily within the scope of an employee’s duties, not whether it merely concerns those duties. Id. at 9-11.

75. Because Franks was decided so recently an insufficient number of cases have accumulated to determine whether it has had a statistically meaningful effect on judicial voting in U.S. Courts of Appeals.  

76. Garcetti, at 425.

77. Id. at 425.

78. Sweezy v. New Hampshire, 354 U.S. 234, 263 (1957) (Frankfurter, J., concurring) (quoting The Open Universities in South Africa 10-12 (1957))(emphasis added); see also Urofsky, 216 F.3d at 414 (“[C]ases that have referred to a First Amendment right of academic freedom have done so generally in terms of the institution, not the individual.”); Paul Horwitz, Universities as First Amendment Institutions: Some Easy Answers and Hard Questions, 54 UCLA L. Rev. 1497, 1497-98 (2007) (arguing that universities should be considered First Amendment actors distinct from the government and individual students and employees).

79. See Sweezy, 354 U.S. at 262-63. (academic freedoms are purely institutional rights).

80. See, e.g., Philip Lee, A Contract Theory Of Academic Freedom, 49 St. Louis U. L. J. 461 (2015) (asserting that although academic freedom is central to the core role of professors in a free society current First Amendment protections exist to protect academic institutions, not the academics themselves).

81. Urofsky v. Gilmore, 216 F.3d 401, 414-15 (4th Cir. 2000) (collecting cases).

82. 474 F.3d 477, 478 (7th Cir. 2007).

83. 624 F.3d 332 (6th Cir. 2010).

84. See discussion supra pp. 15-17.

85. See, Robert F. Williams, The Law of American State Constitutions 113-19, 133-34 (2009) (describing as an essential feature of the U.S. federal constitutional system the ability of states to establish rights beyond the minimum afforded by the federal document).  

86. See, Chunrong Ai and Edward C. Norton, Interaction Terms in Logit and Probit Models, 80 Economics Letters 123-129 (2003).

87. See supra note 55 (for details on the formula by which this variable is coded).




Cite This Article as: Teachers College Record Volume 119 Number 6, 2017, p. 1-28
https://www.tcrecord.org ID Number: 21787, Date Accessed: 1/19/2022 5:35:56 AM

Purchase Reprint Rights for this article or review
 
Article Tools

Related Media


Related Articles

Related Discussion
 
Post a Comment | Read All

About the Author
  • Lewis Wasserman
    University of Texas at Arlington
    E-mail Author
    LEWIS M. WASSERMAN, J.D., Ph.D. is an Associate Professor of Educational Leadership and Policy Studies at the University of Texas at Arlington. He engaged in the private practice of law for 23 years in New York, specializing in public education and civil rights litigation, before entering the academy. He also teaches Education Law at SMU Dedman College of Law in Dallas. His recent articles include: "United States Supreme Court justices’ voting in systemic racial discrimination cases in education" in The Urban Lawyer (2015) (with John P. Connolly); "Overcoming obstacles to religious exercise in K–12 education" in Journal of Legislation(2011); and "Gun control on college and university campuses in the wake of District of Columbia v. Heller and McDonald v. City of Chicago" in Virginia Journal of Policy & The Law(2011).
  • John Connolly
    University of Texas at Arlington
    E-mail Author
    JOHN P. CONNOLLY is a data scientist with the Office of Information Technology at the University of Texas at Arlington. He holds a Ph.D. in political science and master’s degrees in Liberal Studies and Business Administration, as well as a bachelor’s of Engineering. His recent publications in 2015 include "Ideological constraint: History and current status of the concept" in the International Encyclopedia of Social and Behavioral Studies, and "United States Supreme Court justices’ voting in systemic racial discrimination cases in education" in The Urban Lawyer (2015) (with Lewis M. Wasserman).
 
Member Center
In Print
This Month's Issue

Submit
EMAIL

Twitter

RSS