Authors: Banks Miller and Brett Curry
Publication: Justice System Journal 34: 125-151
Original source: copyedit_proof.pdf
Conversion note: this Markdown was generated by text extraction from the hosted file. Review equations, tables, notes, and citations before treating it as a final web copy.
THE EFFECT OF PER SE RECUSAL RULES ON DONOR BEHAVIOR IN JUDICIAL ELECTIONS* BANKS MILLER AND BRETT CURRY Recent judicial decisions and political developments have elevated the issue of impartiality among elected judges as a topic of public and scholarly interest. Using a data set of all donations to candidates for the Supreme Court of Alabama from 1994 through 2010, we explore one potential proposal for limiting the appearance of judicial bias and its effects on the behavior of campaign donors—per se recusal. Our results indicate that the existence of a per se recusal statute significantly decreases the likelihood of observing large donations from several categories of donors. In auxiliary analysis, we find that attorney donors have increasingly funneled contributions through PACs since this statute’s enactment—presumably, because such contributions are exempted from the law. As campaign spending has increased, so too has the perception that interested third parties can sway the court system in their favor through monetary participation in the election process. This perception strikes at the very heart of the judiciary’s role in our society—West Virginia Independent Commission on Judicial Reform, Final Report (qtd. in Sample, 2010). O ver the past four decades, a number of factors have converged to spotlight the precarious relationship between the ideal of judicial impartiality and the realities of contemporary American elections. On the one hand, in decisions such as Buckley v. V aleo(1976) and Citizens United v. FEC (2010), the U.S. Supreme Court has held that political expenditures are a form of speech that merit constitutional protection. The Court has also invalidated state restrictions on campaign speech by judicial candidates (see Republican Party of Minnesota v. White, 2002) on First Amendment grounds. Such decisions, combined with growing competition for state supreme court seats and the increasingly central role of state judicial systems in certain legal areas (Goldberg, 2007; Bonneau, 2005; Baum, 2003; Ware, 2002), have contributed to dramatic increases in the cost of statewide judicial races (e.g., Bonneau, 2007b; Baum, 2003; Cheek and Champagne, 2000). For example, total expenditures on state supreme court elections rose from $83 million during the 1990s to over $200 million in the following decade (Skaggs and Silver, 2011:6). Perhaps even more interesting, “between 1994 and 1998, candidates for state supreme courts raised a total of $73.5 million, and nineteen candidates broke the million-dollar threshold” (Sample and Pozen, 2007:1). In addition to increasing the potential influence of donors and their giving behavior in judicial races, such developments have blurred traditional THE JUSTICE SYSTEM JOURNAL, VOL. 34, NUMBER 2 (2013) * Banks Miller (millerbp@utdallas.edu) is an assistant professor in the Political Science Program at the University of T exas at Dallas. Brett Curry (bcurry@georgiasouthern.edu) is an associate professor in the Department of Political Science at Georgia Southern University in Statesboro.
126 THE JUSTICE SYSTEM JOURNAL distinctions between the character of such elections and those of the more “political” branches. At the same time, this has fostered considerable concern about the ability of elected judges to maintain the appearance (or the reality) of objectivity in certain instances—most notably, those cases in which campaign contributors appear before them. Public opinion surveys indicate that nearly 75 percent of Americans believe that campaign contributions have some influence on judicial decisions (Sample and Pozen, 2007:2), as do 46 percent of state judges (Sample et al., 2010:12). Journalistic (e.g., Liptak and Roberts, 2006) and scholarly (e.g., Ware, 2002) accounts have helped stoke that public concern, leading some to assert that such skepticism threatens perceptions of judicial legitimacy (Sample and Pozen, 2007; see also Gibson and Caldeira, 2012). Having unquestionably enhanced the role of money and politics in judicial elections with the decisions noted above, in 2009 the U.S. Supreme Court held in Caperton v. Massey that the Fourteenth Amendment’s due-process clause required a state supreme court justice’s recusal from a case in which a campaign contributor had a substantial financial interest. 1 The Caperton Court addressed the recusal issue by requiring it when “the probability of actual bias . . . is too high to be constitutionally tolerable.” However, well before that decision both the American Bar Association and numerous state legislatures had been engaged in addressing issues of recusal and disqualification (e.g., Raftery, 2010; Brief of the States, Caperton v. Massey, 2009). As detailed more fully below, in 1999 the American Bar Association revised its traditional approach to recusal to account for concerns about impartiality raised by donations to judicial campaigns. In doing so, the ABA called for states to adopt contribution thresholds at which judicial recusals would be mandatory, with the goal of reducing the number of large contributions from donors and, by extension, the appearance of judicial impropriety those donations might produce. Interestingly, several years before the ABA ’s revision of its Model Code, the Alabama legislature passed just such an act—making it a good case study for the likely policy implications of such a rule as other American states adopt the ABA ’s position. 2 In this article, we examine the consequences of that act—which seemingly requires the recusal of judges if either the parties in a case or their lawyers contribute in excess of a specific monetary 1 T echnically, judicial disqualification is mandatory while recusal is voluntary (e.g., Sample et al., 2008:5). However, we use those terms interchangeably here. In Caperton v. A.T . Massey Coal Corp., 129 S. Ct. 2292, Don Blankenship, the president of Massey Coal, had spent over $3 million to help elect Justice Brent Benjamin to the W est Virginia Supreme Court. Benjamin refused to recuse himself from the case, and he cast the deciding vote overturning a $50 million jury verdict against Massey Coal. 2 In Arizona, California, Iowa, Michigan, New Y ork, Oklahoma, Utah, and Washington, proposals have been adopted. In Georgia, Missouri, and T ennessee, proposals are under consideration, although the Missouri Supreme Court has adopted the rule for itself. In Montana, Nevada, T exas, and Wisconsin, recusal based on campaign contributions has been proposed and dismissed. Amounts triggering recusal in these states range from $50 in Utah (adopted in 2010) to amorphous standards in several states (some adopted and others not) that require disqualification if participating in the case would violate due process or threaten the appearance of impartiality. Most of the adopted statutes were put in place in 2010 or early 2011, and all post-dated Caperton. For a general discussion see Hall and Opsal (2011).
THE EFFECT OF PER SE RECUSAL RULES 127 amount—on the behavior of donors in judicial elections. Put another way, we analyze the impact of this Alabama law to uncover the potential for mandatory recusal thresholds to alter the behavior of campaign contributors and, by extension, their theoretical capacity to restore public faith in the impartiality of judges selected via elections. 3 W e find that several types of donors to judicial campaigns in Alabama appear to have altered their giving patterns in response to Alabama’s recusal rule, that contributions to both Republicans and Democrats have been strongly influenced by the rule, and that, of the groups we analyze, attorneys have been among the most strongly influenced by the specter of per se recusal. In light of those findings, we present evidence to suggest that attorney donors have increasingly shifted their contributions to judicial candidates to PACs—contributions that are not subject to the law’s monetary threshold—since the enactment of the per se recusal statute. This transfer of money to PACs appears to be limited to attorneys, as we find that other groups did not similarly shift large contributions to PACs. It is important to focus particularly on attorneys because, as we show in the article, they are generally much more likely to appear before a judge after giving that judge a contribution than are other kinds of donors. Relying on supplementary data from T exas, presented in the Appendix, we demonstrate that our results are unique to Alabama, thereby increasing our confidence in the efficacy of its recusal rule. Our article proceeds as follows. First, we address the recent history of judicial elections in Alabama and, as part of that discussion, consider several reasons why an intensive study of this single state is appropriate in this instance. W e then provide an overview of the recusal law that has been in effect there since 1996 and describe several issues surrounding its enforcement. Next, we supplement our primary focus on the behavior of donors with a brief discussion on the potential importance of recusal for state supreme court judges who are elected and also consider the state of extant research on the relationship between campaign contributions and judicial decisions. After describing our data and methods, we present our results. W e conclude by discussing the implications of our findings, and speculate about the consequences that the ABA Model’s wider adoption might have on the giving behavior of large contributors in other states. In addition, we present data from T exas in an Appendix to confirm the robustness of our Alabama results. ALABAMA: A CASE STUDY OF JUDICIAL ELECTIONS AND PER SE RECUSAL Single-state case studies have advanced our understanding of numerous political issues in recent years (e.g., Chamberlain, 2012; Ross, Rouse, and Bratton, 2010), 3 Gibson and Caldeira’s (2012:27) study concludes that “recusal rehabilitates legitimacy, but it does not restore it to the level when no conflict of interest exists.” According to their theory, prior citizen attitudes toward the courts act in tandem with contextual factors (including whether or not recusal occurs, the nature of the campaign support in question, and the judge’s influence on the case outcome) to help shape individual perceptions as to whether judicial actions are free from bias (Gibson and Caldeira, 2012:21).
128 THE JUSTICE SYSTEM JOURNAL including state judicial elections (see Rock and Baum, 2010; Cann, 2007; Williams and Ditslear, 2007). Using Alabama as a case study in the present context is especially appropriate because it is the first state to adopt a law requiring per se recusal— indeed, it did so thirteen years before the U.S. Supreme Court’s monumental Caperton decision. As such, an intensive examination of Alabama’s experience represents a potentially fertile pathway for understanding policy change over time. As Glick and Hays (1991:848) describe such an analytical approach, “The earliest adopting state provides important policy leadership by elevating a proposal into law, regardless [of] how tentative or restrictive the policy might be. Therefore, the initial adoption is a crucial political event and probably deserves much more study than it has received.” Gerring (2004:344) aptly describes a case study as “the intensive study of a single unit wherever the aim is to shed light on a question pertaining to a broader class of units.” Beyond notions of policy leadership and diffusion, then, we undertake a case study of Alabama in an effort to draw conclusions about the ways in which subsequently adopted recusal laws may be expected to operate elsewhere. Aside from the above points, there are additional characteristics of Alabama and its judicial elections that make the state especially appropriate as a focus of study. Judicial selection in Alabama occurs through partisan elections, and those elections— particularly elections to the Alabama Supreme Court—have historically been both vicious and expensive (Becker and Reddick, 2003). 4 Between 1990 and 2004, Alabama’s judicial races averaged over $1.4 million per election and were the second most expensive in the country (Bonneau, 2007b). From 2000 to 2009, the state’s supreme court campaigns were the most expensive in the nation with a total in excess of $40 million—roughly twice as large an aggregate sum as the second-place state (Sample, Rouse, and Bratton, 2010:7). In addition to being a perpetual leader in total campaign contributions for judicial races, the state has a reputation for awarding especially high jury awards in civil litigation (e.g., B.M.W . of North America v. Gore, 1996; Moller, Pace, and Carroll, 1999:294). Once elected to the bench, the state’s supreme court justices are well positioned to exert significant influence on many such high-dollar lawsuits. In 1995 the Alabama legislature passed a firm disqualification rule that would trigger the automatic recusal of supreme court justices who received above a set amount from the parties in a case or their attorneys. Specifically, Alabama law requires supreme court justices to recuse themselves if the amount they have received from the parties, lawyers for the parties, or contributions from associated law firms was greater than $4,000 in the previous election cycle (Code of Ala. §12-24-2). The statute indicates that recusal is forfeited if no party to the case files a written appeal 4 According to Cann, Bonneau, and Boyea (2012: 42), “Partisan elections breed contentious campaigns marked by issue content and greater campaign spending. . . . Under these conditions, judicial candidates seeking to raise substantial sums of campaign funds may be uniquely vulnerable to the influence of campaign contributors” (internal citations omitted). In light of this finding, per se recusal may be particularly necessary in states with partisan judicial elections.
THE EFFECT OF PER SE RECUSAL RULES 129 for disqualification, meaning that recusal is only “automatic” when it is requested. Additionally, the structure of Alabama’s recusal statute seeks to guard against the potential for donors to contribute in excess of $4,000 to “unfriendly” judicial candidates so as to provide for their automatic recusal in future cases. Specifically, the law requires that the party in opposition to the large donor be the one to file for per se recusal of the judge. Consequently, any attempt to use the law strategically would require an opponent in litigation who is surpassingly naïve; additionally, there is no indication in our data that donors contribute in this fashion. Finally, the statute does not seek to regulate money given by either parties or attorneys who file amicus briefs.5 It is unclear why the Alabama legislature settled on a cap of $4,000. However, subsequent evidence from other states suggests that legislative compromises probably occurred in an effort to limit the number of large contributions that may threaten judicial impartiality while not unduly handicapping candidates of modest means or particular segments of the bar.6 Paradoxically, although the Alabama act took effect on January 1, 1996, it has not been enforced. The major stumbling block to this enforcement has been preclearance from the U.S. Department of Justice, as required by the V oting Rights Act of 1964. After Alabama’s attorney general submitted the act for preclearance, the Department of Justice expressed concern the law would negatively affect African- American judicial candidates (see Little v. King et al. , 2011). Attorney General Jeff Sessions withdrew the preclearance request after concluding that the statute was not subject to the requirement (Little v. Strange, 2011), and responded that the state would enforce the law without preclearance (Orndorff, 2010). Justices on the Alabama Supreme Court have noted the act’s uncertain status and, in 1999, the Administrative Office of Alabama Courts asserted that, until the act secured preclearance from the Department of Justice, “it is not legally enforceable and our office will not take any action for noncompliance with the Alabama Act’s provisions” (Little v. Strange, 2011:15). While our analysis centers on the behavior of donors as opposed to the recusal rates of state judges themselves, in light of the Alabama law’s lack of formal enforcement, we note several additional points that we believe nevertheless underscore the law’s potential to alter the calculus of campaign contributors. Using data on judicial recusal from the Alabama Supreme Court in 2001 and 2003 we found that judges were 5 Underscoring this omission, one Alabama Supreme Court justice explicitly refused to disqualify herself in a case where a PAC contributing over $50,000 to her campaign participated as an amicus (Bracken v. T rimmier Law Firm, 879 So. 2d 207 [2004]). Note also that §12-24-2(b) also does not require recusal if a contribution comes from a PAC, unless, improbably, the PAC is a party to litigation. 6 The report from the Supreme Court of T exas’s Judicial Campaign Finance Study Committee describes the T exas legislative give-and-take as follows: “The Legislature devised the scheme [of contribution limitations] in an effort to set a sufficiently high individual limit to permit plaintiffs’ lawyers (who typically practice as solo practitioners or in smaller firms) to remain on a level playing field with big-firm defense lawyers, yet set the firm aggregate limits sufficiently high so as to permit individual attorneys within large firms to make contributions and participate in the political process.”
130 THE JUSTICE SYSTEM JOURNAL 450 percent more likely to recuse themselves in cases in which one side (including the attorneys and the party to litigation) contributed an amount in excess of the threshold, as compared to cases in which contributors did not exceed the threshold.7 While this evidence is clearly circumstantial, it does provide reason to suspect that donations exceeding the statute’s threshold may be treated as a norm by judges considering when recusal may be necessary. Moreover, even if the law is not being enforced externally, donors cannot know whether or not a particular judge will—assuming the presence of a contribution in excess of the $4,000 maximum—nevertheless feel bound to recuse his- or herself to honor the behavioral norm and avoid the appearance of impropriety or whether the law might be formally enforced at some future point. In other words, perhaps some contributors are reluctant to breach the $4,000 threshold because, regardless of the law’s formal enforcement, they may be concerned that many judges will abide by its provisions anyway or that they could be subject to a decision to enforce the law in the future. Indeed, at least one federal lawsuit over the uncertain status of the law has asserted that the potential for enforcement is enough to alter a donor’s calculus (see Little v. Strange, 2011). Such speculation, if valid, might well be most pronounced for repeat players in the justice system, such as attorneys (e.g., Galanter, 1974). The tenuous legal status of §12-24 notwithstanding, its existence provides a useful opportunity to examine the practical effects of the most prominent and long-standing version of the American Bar Association’s per se rule on judicial disqualification. If we find that the law affects donor behavior, despite not being enforced, then this is powerful evidence of the efficacy of this type of law. Moreover, as we have noted, Alabama judicial elections are a worthy topic in their own right given the incredibly expensive nature of those campaigns. In the remainder of this article, we look to the potential effects this law may be having on campaign contributions—notwithstanding the fact that it is not being formally enforced. Our expectation is that the Alabama recusal rule has led to a decrease in the number of contributors who will donate in excess of $4,000 to state supreme court candidates. As detailed below, our analysis considers the potential effects of this recusal rule on the behavior of different categories of donors. W e are, therefore, interested in the ways this law has affected giving by each of these donor groups, and it is our a priori expectation that §12-24 has exerted a similar effect on these contributors across the board. However, in light of their central position in the judicial process and the frequency with which they appear as donors before judges, we devote particular attention to the statute’s implications for attorney giving. Subsequent to our examination of the statute’s effect on the probability of a contribution being given above the threshold, we also explore what effect, if any, the law has had on the aggregate amount of money raised by judicial candidates in Alabama. 7 W e excluded from this calculation recusals that appeared to be granted for the traditional reasons cited in most recusal statutes, including being related to a litigant or having previously been a judge in the case.
THE EFFECT OF PER SE RECUSAL RULES 131 As an initial matter, it seems probable that, among the donor categories we examine, attorneys should be especially aware of the recusal rule’s existence, even with its lack of official enforcement. Beyond that, attorneys have long been recognized as quintessential repeat players before the courts (Galanter, 1974; McGuire, 1995). As repeat litigators, attorneys appearing before a court will need to preserve credibility to maintain effectiveness (McGuire, 1995; McAtee and McGuire, 2007), and recusal can be a sensitive issue with a judge before whom an attorney must make repeated appearances (Sample and Pozen, 2007). More generally, as Williams and Ditslear (2007:137) put it in their study of Wisconsin judicial races, “With the other advantages attorneys already experience, it is important to focus on this group of contributors who are most likely to receive some direct benefit from getting their preferred judge elected.” T o summarize, our primary goal is to uncover the ways in which Alabama’s per se recusal rule has altered the behavior of campaign contributors in judicial races. Beyond that, we focus on how the rule has influenced attorneys because of their centrality to the litigation process, as well as the frequency with which they donate and subsequently appear before judges (an issue we address in the discussion section). Additionally, apart from whatever utility the Alabama rule may have in limiting the number of large campaign contributions, we discuss the extent to which limiting the frequency of these substantial donations represents an effective way to reduce the appearance of the influence of money in judicial campaigns. Before exploring these questions about donor behavior, however, we briefly consider several issues surrounding judicial elections, recusal, and the potential influence of campaign contributions on judicial decision making. RECUSAL, CAMPAIGN CONTRIBUTIONS, AND JUDICIAL DECISION MAKING Contemporary efforts to reform recusal guidelines have been concentrated in those states that elect their judges and, not surprisingly, agitation for reforming recusal practice has been most significant in those states where judicial elections are especially expensive. As the cost of those races has ballooned (e.g., Bonneau, 2007b), so too have the risks of real or apparent judicial bias. While these risks are most acute for the parties in any given case, there is also concern that this appearance of bias jeopardizes the effectiveness of courts as institutions. For example, Gibson and Caldeira (2012) have presented experimental evidence demonstrating that the giving of money to judges damages public perceptions of judicial impartiality. What seems important here is the frequency with which someone who has contributed a large amount of money is before a judge and not necessarily the amounts of money contributed. When such a core function of the judiciary is called into question, the institutional legitimacy of courts is necessarily imperiled. However, Gibson and Caldeira’s study also finds that recusal may help counteract such concerns. This is important because, while the public seems to favor judicial
132 THE JUSTICE SYSTEM JOURNAL elections (ABA, 2008; Geyh, 2003; Champagne and Haydel, 1993), it is hardly enamored with the potential effect of monetary donations on future judicial decisions. At the same time, because judicial races are such low information contests (Abrahamson, 2001; Baum, 2003), it may actually be desirable to fund them at high levels to yield both more informed voting and more competitive elections (e.g., Bonneau and Hall, 2009). Consequently, recusal practices may provide a way for states to benefit from expensive judicial races and simultaneously avoid some of their most noticeable costs. As the Conference of Chief Justices has put it, “Disqualification is perhaps the States’ most reliable weapon for maintaining both the reality and the appearance of a ‘fair hearing in a fair tribunal’ for every litigant” (Brief of the Conference of Chief Justices, Caperton v. Massey, 2009; see also Bam, 2010). T raditionally, standards in the United States have instructed judges to recuse themselves in those situations where their “impartiality might reasonably be questioned” (ABA Code of Judicial Conduct, R.2.11(A)). That benchmark, contained in the American Bar Association’s Model Code of Judicial Conduct, has been adopted by almost every state (Sample, Pozen, and Y oung, 2008). While this rule covers a number of concerns related to judicial objectivity, 8 it does not speak to the issues surrounding large financial contributions to judicial campaigns. T o address those concerns, in 1999 the ABA adopted a per se model rule calling for mandatory disqualification of elected judges who had received contributions above a predetermined threshold from a party or the party’s counsel within a specific period of time (ABA Code of Judicial Conduct, Canon 2, R.2.11 (A)(4); Skaggs and Silver, 2011; Abramson, 2000). Rejecting a one-size-fits-all framework, the ABA rule prescribes no set financial limits or specific time frames. Instead, it leaves such matters to the discretion of the states. As we have noted, Alabama was the earliest state to adopt those specific dollar amounts and time frames, although it did so ahead of the ABA ’s embrace of such a rule in the Model Code. In addition to concerns over judicial impartiality and the ultimate effects of campaign contributions on the legitimacy of courts, there are other problems with current recusal practice that might be ameliorated by wider adoption of the ABA ’s per se rule. For one, parties are likely deterred from requesting recusal for fear of offending the judge who, in all probability, will deny the recusal request and decide the case. Moreover, judges recognize a “duty to sit” doctrine, which holds that they have a professional responsibility to hear cases whenever possible. Some observers have even suggested that this “duty to sit” is even more important than maintaining the appearance of impartiality (Frost, 2005). In combination with psychological research indicating that important aspects of bias may be unconscious (e.g., Sample and Pozen, 2007; Bassett 2005, 2002), this professional duty to sit is likely to prompt many judges to 8 For example, the rule holds that recusal should always occur when 1) a judge is biased against one of the parties, 2) a judge has previously served as a lawyer in the matter, 3) the judge has an economic interest in the case, 4) the judge is related to a party or lawyer in the case, 5) has personal knowledge of disputed facts, or 6) has made improper ex parte contact during the proceeding (Sample, Pozen, and Y oung, 2008; Abramson, 2000).
THE EFFECT OF PER SE RECUSAL RULES 133 resist making the choice to recuse.9 However, requiring automatic disqualification with preexisting, objective thresholds would seem to resolve a number of these concerns. While campaign spending may influence the outcomes of judicial elections generally (e.g., Hall and Bonneau, 2006; Bonneau, 2005), there is also evidence to suggest that donors may be given favorable treatment before those judges they support (Cann, Bonneau, and Boyea, 2012; Cann, 2007; McCall, 2003; Waltenburg and Lopeman, 2000; Ware, 1999). Although we investigate the influence of Alabama’s per serecusal statute on the behavior of several types of donors, the law’s effect on attorney giving may be especially important to understand for several reasons. For example, Goldberg et al. (2005) report that, although the percentages vary widely across states, attorneys routinely account for a substantial proportion of contributions to judicial campaigns. Cann, Bonneau, and Boyea (2012) contend that this giving by attorneys is largely designed to facilitate access to state judges, though debate remains about the extent to which that money secures access, represents a reward for past decisions, influences outcomes in future cases, or some combination of the above (Cann, 2002; Waltenburg and Lopeman, 2000). Scholars have also noted that attorneys, as a class of contributors, are especially well positioned to benefit from the election of likeminded jurists (see Williams and Ditslear, 2007:135-37). In this way, one might liken attorney donors to “investors” who are chiefly motivated to give by material incentives. As Francia et al. (2003:48) describe such donors in their influential study of congressional campaigns, “large majorities of investors state that they always consider whether a candidate is friendly to their business or industry and has treated their business fairly prior to making a contribution.” At a minimum, many attorneys who contribute to judicial campaigns will find themselves arguing before judges after those elections (Cann, 2007; McCall, 2003; Champagne, 1988, 1986) and while this by no means forecloses the possibility that other types of donors might be affected by a recusal statute, it does make the effect on attorney behavior particularly important to understand. Though isolating the causal direction of the relationship between contributions and judicial votes has proved elusive in several studies (e.g., McCall and McCall, 2007; Waltenburg and Lopeman, 2000; Ware, 1999; but see Cann, Bonneau, and Boyea, 2012; Cann, 2007), there is substantial evidence of a relationship between contributions and votes. For example, Ware’s (2002:584) analysis of arbitration cases concludes that “contributors to judicial election campaigns buy[] changes in law and policy.” McCall (2003, 2001; McCall and McCall, 2007) reports similar findings in her studies of the T exas courts. In their examination of Alabama, Kentucky, and Ohio tort cases, Waltenburg and Lopeman (2000:255) look at attorney contributions in 9 Though we cannot test it here, mandatory recusal requirements may be especially important for judges serving on appellate courts. In contrast to trial courts where a single judge presides and, in the event of a conflict, additional judges would likely be available for substitution, we would expect the professional duty to sit to be a particularly powerful deterrent to voluntary recusal on appellate courts due to its potential to result in an equally divided judgment.
134 THE JUSTICE SYSTEM JOURNAL particular and find that “there is a relationship between the bar’s contributions and a justice’s decisions in tort cases, at least for these three courts” even as they acknowledge their design is unable to isolate whether “decisions follow dollars or dollars follow decisions.” Using a different methodological approach, Cann’s (2007) study of the Supreme Court of Georgia concludes that campaign contributions influence judicial decision making on that tribunal. In another study, Cann, Bonneau, and Boyea (2012) rely on an alternate strategy in attempting to isolate this causal relationship in T exas, Michigan, and Nevada’s courts of last resort. In sum, this potential relationship between contributions and judicial decision making underscores the importance of gaining a better understanding of how judicial recusal standards may influence donor behavior. DATA AND ANALYTICAL APPROACH In our analysis, we use data made available from the National Institute on Money in State Politics (NIMSP) for all donations to candidates for the state supreme court in all races in the state of Alabama from 1994 through 2010. This time frame reflects the availability of data and, more important, allows us to test whether the potential imposition of an automatic recusal provision affects donor behavior. Statute 12-24 went into effect on January 1, 1996, meaning that we can examine “normal,” pre-statute donor activity in the 1994 cycle and for part of the 1996 cycle. 10 In all we have 30,830 unique donors to 73 candidates. 11 Before §12-24 went into effect, we observe 6,108 contributions, giving us ample opportunity to discern any patterns that may be present before the statute is enacted—within the 1996 election we observe 118 donations before the statute goes into effect and 2,918 after it became effective. In the results that follow we make several distinctions among classes of donors. W e created four donor categories from the sector coding provided by the NIMSP: attorney, general business, health care, and party.12 These four categories comprise 58 percent of the total number of donations and 89 percent of the total amount of 10 In all, 20 percent of our observations occur before the implementation of the statute, giving us ample opportunity to observe an effect for the law (see T able 1). There are some issues with respect to Republican donors before implementation of the law, as there are comparatively few donors making large contributions before implementation of the per se recusal law—something that is apparent in the statistical significance of effects presented for Model 3 in T able 3. 11 There are more individual contributions to judges, but we combined instances in which the same contributor gives more than once to the same judge in the same electoral cycle, as required by §12-24. Therefore, by “unique” donor we mean donors who contributed to a candidate in an election cycle. Thus, if firm A gave in the 1994, 1996, 1998, and 2002 elections that would count as four unique donations. 12 The attorney category includes individual attorneys, law firms, and PACs formed by legal groups; the general business category includes a host of types of businesses, including agriculture, communications and electronics, construction, finance, insurance, real estate, and transportation businesses and business PACs; health care includes health professionals, health-service providers, hospitals, nursing homes, and pharmaceuticals and other health-care-related PACs; party includes party committees and candidate transfers (including leadership PACs). In the models of the number of contributions exceeding the threshold we include PAC contributions, as listed above, as this presents a more difficult test of our belief that the law reduced the number of contributions over the limit. Excluding PACs from the regressions in T able 2 does not alter our results.
THE EFFECT OF PER SE RECUSAL RULES 135 money raised in Alabama between 1994 and 2010, with donations from health care groups easily contributing the least to overall fundraising. The remaining donations fell into a residual category, the bulk of which consist of donors who are “uncoded” in the NIMSP data (95 percent of the residual category), as well as a small set of other types of donors including labor unions and single-issue advocacy groups. 13 W e treat this residual category as a baseline group and omit it from the regressions that follow. Figure 1 tracks the relative contributions of each group to the totals raised by all candidates. W e do not adjust contribution amounts to account for inflation to provide a more stringent test of whether §12-24 is effective: inflationary pressure should make the $4,000 limit set in 1996 more constraining with the passage of time. 14 One notable change in the figure is the relatively steep and long-lived drop in the amount of money 13 The reference category is a catchall meant to include a set of miscellaneous donors, the majority of whom tend to give small amounts infrequently. The vast majority of these individuals are simply interested citizens who have no apparent business or partisan interest to promote before the court, with the small exception of some labor unions and other single-issue advocacy groups (e.g., environmental protection). The mean amount contributed from this group is $267 compared to $3,141 for those not in the reference category. Less than 1 percent of the contributions in this group come from labor unions or single-issue advocacy groups. 14 T o account for inflation, the $4,000 limit in 1996 would need to have increased to $5,559 in 2010.
Contributions to Overall Fundraising by Donor Category
136 THE JUSTICE SYSTEM JOURNAL coming from attorneys and law firms immediately after §12-24’s implementation (represented by the solid vertical line). The figure also clearly demonstrates the relative primacy of business spending from the 2000 election onward. In the regressions that follow we use one dependent variable and a host of independent variables, with the majority of these independent variables motivated by the work of others on patterns of campaign fundraising in state supreme court elections and other state elections (e.g., Bonneau, 2007a). Our dependent variable captures whether or not a donation exceeded the threshold amount set by §12-24, and is equal to 1 if the donation is above the threshold and zero otherwise. The main independent variables of theoretical interest are the four categories of donors (attorney, general business, health care, and party), each of which is set to 1 if the donor is in the category and zero otherwise, and before statute (equal to 1 if the donation is made before January 1, 1996). Additionally, because we are interested in the combined effects of the statute on each class of donor, we incorporate interactions between each of the donor category variables and the before statute variable (see T able 1). T able 1 Descriptive Statistics Variable Mean Std. Dev. Min. Max. Donation over Threshold 0.053 0.224 01 Before Statute 0.198 0.399 01 Attorney 0.365 0.481 01 General Business 0.136 0.342 01 Health Care 0.052 0.222 01 Party 0.023 0.150 01 AttorneyBefore Statute 0.109 0.313 01 BusinessBefore Statute 0.019 0.139 01 HealthBefore Statute 0.012 0.108 01 PartyBefore Statute 0.004 0.060 01 Democrat 0.473 0.499 01 Candidate Characteristics Black 0.083 0.276 01 Woman 0.216 0.411 01 Incumbent 0.263 0.440 01 Quality Challenger 0.491 0.499 01 Close Race 0.098 0.297 01 Contest Characteristics Number of Seats 3.736 1.604 15 Open Seat 0.372 0.483 01
THE EFFECT OF PER SE RECUSAL RULES 137 Aside from these variables of interest, we include a number of independent variables to act as controls. Black is equal to one if the candidate is black, woman is equal to one if the candidate is a woman, incumbent equals one if the candidate is an incumbent, quality challenger is equal to one if the candidate is running as a challenger and has previously been a judge, close race equals one if the incumbent in a race previously won office with less than 60 percent of the vote, and open seat is equal to one if the seat being contested is open; all of these variables equal 0 when the char- T able 2 Regression Results, 1994-2010 Variable Before Statute Attorney General Business Health Care Party AttorneyBefore Statute BusinessBefore Statute HealthBefore Statute PartyBefore Statute Democrat Candidate Characteristics Black Woman Incumbent Quality Challenger Close Race Contest Characteristics Number of Seats Open Seat Constant N Likelihood Ratio Test Clusters Area Under ROC Model 1: All Candidates Coefficient Clustered S.E. 2.024* 0.458 0.795* 0.178 3.135* 0.176 1.007* 0.265 2.189* 0.189 0.768* 0.332 -0.936* 0.306 -2.465* 1.044 0.129 0.522 -0.096 0.153 -0.635* 0.203 -0.179 0.164 0.209 0.231 0.011 0.193 -0.138 0.267 0.006 0.050 0.160 0.153 -717.689 128.686 30,830 2,387.02 (p=0.000) 73 0.81 Model 2: Democratic Candidates Coefficient Clustered S.E. 2.768* 0.663 0.531 0.302 2.927* 0.490 -0.575 1.110 1.913* 0.293 0.670* 0.349 -0.610 0.557 0.223 1.418 -0.544 0.496 — — -0.722* 0.339 -0.660* 0.238 -0.079 0.371 0.749* 0.267 0.812* 0.389 -0.139 0.086 -0.263 0.296 -750.603 235.262 14,574 482.904 (p=0.000) 28 0.74 Model 3: Republican Candidates Coefficient Clustered S.E. 0.856 0.621 0.969* 0.252 3.206* 0.181 1.198* 0.271 2.321* 0.244 1.007* 0.399 -0.611 0.422 — — 1.636* 0.539 —— —— 0.049 0.163 0.210 0.264 0.244 0.212 -0.234 0.252 0.023 0.042 -0.175 0.137 -676.439 155.788 15,980 1,849.701 (p=0.000) 45 0.85 * Significant at p < 0.05 (two-tailed). Restricted cubic spline included in regression but omitted from display.
138 THE JUSTICE SYSTEM JOURNAL acteristic is not present for the candidate. Additionally, we include a control called number of seats that equals the number of seats contested in a given election cycle, which ranges from one to five. Democrat is a variable equal to one if the candidate is a Democrat and zero otherwise. W e include this variable to help capture the potential for the realignment in the South in the 1990s in favor of Republicans to explain declines in giving by certain groups because their preferred candidates were uncompetitive. W e further check that our results are robust to the effects of realignment by splitting the sample along partisan lines and reestimating our regressions, finding that the patterns we uncover seem to affect Democrats and Republicans similarly.15 RESULTS In T able 2 we present the results of three regressions. Model 1 includes data on all candidates, whereas Models 2 and 3 separate Democratic and Republican candidates, respectively. Given that we have a dichotomous dependent variable we estimate our regressions using a logit link function. W e include a restricted cubic spline with two knots (as determined by the data) to account for time-dependence across our sample (Beck et al., 1998). Because we treat these as nuisance parameters, they are not displayed in the models that follow. In general, all three models fit the data well, with highly significant Likelihood Ratio test statistics and areas under the Receiver Operating Curve (ROC) above 0.70. Model 1 contains all 30,830 observations in the data. Before turning to the results of the variables of theoretical interest, we note that the black variable is significant. Being a black candidate reduces one’s likelihood of receiving a contribution over the $4,000 threshold by just over 1 percent. T urning to the variables of theoretical interest, our a priori expectation—that the number of donations over $4,000 has been reduced after the per se rule’s enactment— is largely supported. Keeping in mind the interaction terms and that the reference category of donor is the “uncoded” group noted above, the coefficient on the before statute variable represents the likelihood that a donation over the threshold will be made by a contributor in this reference category only. It is positive and, as we will illustrate shortly, substantively large, meaning that donations exceeding the $4,000 limit 15 As an additional check on the robustness of our findings, we reestimated our models with a variable we call GOP margin, which is the number of Republicans on the court minus the number of Democrats. This variable ranges from -7 (in 1994) to 9 (in 2006), with larger numbers representing greater Republican control. In 1994 and 1996 Democrats controlled the supreme court, but by 1998 Republicans gained the upper hand (although the margin was only one in 1998). W e include this variable to control for the possibility that campaign donors might “invest” in campaigns based on partisan control of the court. For example, it may be that contributors who otherwise might not donate to Republicans give to Republican candidates to protect themselves once Republicans gain firm control of the court; alternatively, donors who would normally contribute to a Democrat might simply withhold their donation in the belief that it is impossible for a single Democrat on a court with eight Republicans to effectively represent the interests of the donor. Because the change in the composition of the Alabama Supreme Court is from strong Democratic control to strong Republican control over time, it may be that such “investment” behavior on the part of contributors confounds any effect we find for §12-24. However, including this variable does not change the results we present.
THE EFFECT OF PER SE RECUSAL RULES 139 were more likely before the statute went into effect than after. Interpreting the interaction terms here requires a bit of care—indeed, one cannot simply judge the statistical significance of these terms by gauging the significance of the coefficients (Brambor, Clark, and Golder, 2006), which is why we also include T able 3. Furthermore, the uninteracted constituent terms in the model (attorney, general business, health care, and party) represent the likelihood of a contributor making a donation over the threshold after the statute is enacted (i.e., when before statute equals zero). Unfortunately, the significant coefficients on the attorney/general business/healthcare/party variables do not convey information of substantive interest because they only represent relative comparisons between these variables and our reference group— these uninterpreted coefficients cannot tell us how the likelihood of making a donation over the threshold changed for each group pre- and post-enactment of §12-24. T o isolate the likelihood of giving above the threshold for any given group we must simulate the probability of a donation above the threshold occurring before and after enactment, holding the group variable constant. W e do this in T able 3, which includes results for all candidates (Model 1) and for Democrats (Model 2) and Republicans separately (Model 3). In T able 3, we have estimated the probability of a donation above the $4,000 threshold for each donor group before and after the statute was enacted. In general, relative to their behavior before enactment each group (including the reference group) T able 3 Predicted Probabilities of Donations over Threshold All Candidates Before Statute After Statute Percent Change Democrats Only Before Statute After Statute Percent Change Republicans Only Before Statute After Statute Percent Change Reference Group 0.09 [.02, .15] 0.01 [.01, .02] -89 Reference Group 0.15 [.04, .27] 0.01 [.01, .02] -93 Reference Group 0.03 [.00, .06] 0.01 [.01, .02] -66 Attorneys 0.16 [.05, .27] 0.03 [.02, .03] -81 Attorneys 0.22 [.04, .41] 0.02 [.02, .03] -91 Attorneys 0.08 [.00, .16] 0.03 [.02, .05] -63 Business 0.59 [.43, .74] 0.21 [.18, .24] -47 Business 0.65 [.46, .83] 0.17 [.09, .25] -74 Business 0.39 [.14 .63] 0.23 [.20, .25] -41 Business 0.59 [.43, .74] 0.21 [.18, .24] -47 Business 0.65 [.46, .83] 0.17 [.09, .25] -74 Business 0.39 [.14 .63] 0.23 [.20, .25] -41 Business 0.59 [.43, .74] 0.21 [.18, .24] -47 Business 0.65 [.46, .83] 0.17 [.09, .25] -74 Business 0.39 [.14 .63] 0.23 [.20, .25] -41 95% confidence intervals are in brackets.
140 THE JUSTICE SYSTEM JOURNAL is significantly less likely to give a donation above the threshold. 16 The reference group decreases from having a 9 percent likelihood of a donation over the threshold to just a 1 percent chance, an 89 percent decrease in the probability of a large donation. Similarly, attorneys are significantly less likely to make a large donation after the statute than they are before it, decreasing from a 16 percent likelihood to just 3 percent, a decrease of 81 percent. General business donors follow a similar, if less dramatic pattern: there is a 59 percent chance of a donation over the threshold occurring before enactment, which decreases to 21 percent after enactment—a decline of 47 percent. Similarly, health-care donors are less likely to make a contribution over the threshold after the enactment of §12-24. Finally, the number of contributions from party committees and candidate committees also decreases significantly after enactment, from a 40 percent likelihood that any given donation is above the threshold to just a 10 percent likelihood after enactment of §12-24, a 75 percent decrease in the chances of a large donation. Given our concern that the results we observe could be confounded by the fact that the enactment of §12-24 coincides with the Republican consolidation of the South in the 1990s, we estimated separate regressions for Democratic candidates (Model 2) and Republican candidates (Model 3). 17 If the direction of our results is consistent for both Democrats and Republicans, then we can be somewhat less concerned about this rival explanation for any effects of §12-24 because our results would not be solely driven by donations from different partisan constituencies participating at different rates based on the probability that their preferred candidate would prevail. T urning to Model 2, for Democratic candidates the probability of a donation above the threshold is affected by candidate and contest characteristics. Black candidates and female candidates are less likely to receive large donations; each is about one percentage point less likely than a white or male candidate who is also a Democrat to receive a large donation. Quality challengers are three percentage points more likely, and those candidates whose previous election was a close race are seven percentage points more likely, to receive a large donation. In addition, increasing the number of seats being contested in an election from its minimum (1) to its maximum (5) decreases the probability of a large contribution by three percentage points. As Model 3 demonstrates, neither candidate nor contest characteristics seem to matter appreciably for the likelihood of a Republican candidate receiving a large donation. As noted in the discussion of Model 1 above, interpretation of the coefficients for the different groups of donors, the before statute variable, and the interaction terms is complicated and is best understood in terms of the predicted probabilities for 16 The statistical significance of these effects can be judged by comparing the standard errors for each effect. T o take one example, there is no overlap in the 95 percent confidence intervals for the “attorneys” category in the model that includes all donations (Model 1); therefore, these effects are distinct statistically at the 95 percent confidence level. 17 W e have also included an Appendix to check the robustness of our findings against the patterns in T exas, which has an electoral environment similar to Alabama’s.
THE EFFECT OF PER SE RECUSAL RULES 141 each group before and after the enactment of §12-24. Those quantities are displayed in T able 3, with 95 percent confidence intervals in the brackets. In general, the pattern of decreasing numbers of large donations across all candidates reappears regardless of whether we are looking at Democratic or Republican candidates, with the exception of health-care donors, for whom there are statistically insignificant results (in Model 2) or for whom there are empty cells (Model 3). 18 None of the effects for Model 3 in T able 3 are statistically significant at conventional levels. That is, we observe no statistically discernible effect for donors to Republicans, although the direction of these effects is consistent with our expectations and they are generally quite large. Attorney contributors are significantly less likely to give to both Democratic candidates (a 91 percent decrease) and Republican candidates (a 63 percent decrease), although the change is somewhat larger for Democrats than for Republicans. As noted, the change for attorney giving to Republicans is not statistically significant, given the overlap of the confidence intervals for the effects, but this is entirely driven by a comparatively small number of large donations from attorneys to Republicans before the statute goes into effect. Of greater importance is that the direction of §12-24’s effect is consistent across the partisanship of the candidate, mitigating some concern over the potential for the Republican realignment of the South to explain our results (the Appendix also addresses these concerns). As we expected, businesses become significantly less likely to give to Democrats, a 74 percent decrease, in the wake of the recusal rule’s adoption. There is also a decrease in the number of large contributions from business groups to Republicans, although the 41 percent decrease is smaller than that experienced by Democrats. Party and candidate contributions seem to follow the same expected pattern for Democrats and Republicans— with the likelihood of a large donation decreasing after the statute goes into effect. In summary, the per se recusal statute seems to have played a role in reducing the number of large donations from attorneys, business, and parties and candidates across party affiliation, and the directional consistency of these findings leads us to believe they are robust to Republican consolidation in the South. Similarly, there is a decrease in the number of health-care donations above the threshold, though we are unable to make party-specific conclusions(see note 19). However, though these results tell us a good deal about the number of large contributions made to judges in Alabama—which may have important ramifications for the appearance of bias in those courts given the findings of Gibson and Caldeira (2012)—they do not tell us whether the actual amounts of money given changed significantly in the wake of §12-24’s enactment. T o explore this issue, we first note that the percentage of money from contributions over the threshold set by §12-24 actually increased considerably after the enactment of the statute. In the 1994 election cycle, 58 percent of the total amount of money raised came from contributions over $4,000—in every election after 1994 the 18 There are 276 observations missing from estimation in Model 3 because not a single donor in the health-care group gave a large donation to a Republican before the enactment of the statute, and there were only two donations greater than $4,000 after the law’s adoption.
142 THE JUSTICE SYSTEM JOURNAL percentage raised from these large donations is 77 percent or higher. T wo things are worth noting about this. First, in terms of raw dollars, contributions over the threshold comprise the majority of all monies raised in every election since 1994, and that total usually exceeds 75 percent. Second, these results are somewhat paradoxical given the decrease in the number of contributions over $4,000 we have just reported: how can the number of large donations decrease while the amount raised from those donations increases? The answer is that donations over the threshold get much larger after the enactment of the per se rule, although there are fewer of them. Adjusting for inflation, the average donation over the threshold before enactment is $17,429 and after enactment it is $34,983. And these large donations are not subject to the per se recusal rule if they come from PACs (see §12-24-2(b)). Thus, it may be that one response to §12-24 has been for contributors to alter their methods of giving—namely, by bundling money in PACs so their donations do not count against the contribution limit, triggering recusal. T o investigate this, T able 4 displays the percentage of large donations that came from PACs for attorneys and business interests.19 If the recusal law affects the manner in which large donations are given, then there should be an increase in the percentage of these large donations moving through PACs to candidates. T wo things are apparent: first, donations over the threshold from attorneys shifted from individuals and firms before enactment (15 percent of the total raised from large donations are from PACs) to PACs after enactment (between 54 percent and 90 percent of the total raised from large donations from PACs); second, business interests evince no clear pattern in shifting their large donations to PACs. Therefore, not only is there evidence that §12-24 reduced the number of large donations from attorneys, there is also evidence that when attorneys gave large sums the statute shifted their giving from individual or firm donations to PACs, but businesses were not similarly affected. The Appendix demonstrates that the pattern we see for Alabama attorneys is unique and is not replicated in T exas, making it more likely that this change in behavior is driven by the per se recusal law and not a broader, more global trend. DISCUSSION The per se recusal statute we investigate here appears to have altered patterns of giving in races for the Alabama Supreme Court. 20 Of particular practical interest, attorneys have been significantly less likely to make a large donation as an individual or through a firm since §12-24’s passage, and it seems that the law made it more likely for 19 W e also analyzed the pattern for the number of large contributions, as opposed to the amount of those contributions, and the pattern of increasing use of PACs by attorneys remains. Comparisons with party contributions reveal generally little money in the form of large donations funneled through PACs, with no more than 15 percent of the total contributed above the threshold from these groups coming from PAC donations. Because of the small number of contributions above the threshold from health-care groups, no meaningful comparison is available with that group. 20 Not only did that basic trend hold across all our donor groups; it was also evidenced by decreasing numbers of large contributions regardless of candidates’ party affiliations, albeit with varying degrees of magnitude.
THE EFFECT OF PER SE RECUSAL RULES 143 large donations from the legal community that might otherwise trigger recusal to be funneled through PACs. What is most remarkable is that all of this is true despite the fact that §12-24 has, to our knowledge, never actually been formally enforced. The mere possibility of enforcement seems to have been sufficient to alter donor behavior. But why should donors worry about triggering recusal when there appears to be no enforcement mechanism? W e believe that such an objection ignores the subtle reality that exists for those who wish to donate to campaigns and how judges in Alabama appear to have reacted to the statute. In light of their comparatively large proportion of contributions to judicial races and the critical role that their compliance would almost certainly play in facilitating the success of per se recusal statutes in other venues, we address this concern by focusing on the motivations of attorney contributors. Despite this concentration, we would underscore that several of these factors could be relevant to the behavior of nonattorney donors as well. As noted earlier, lawyers belong to the class of donors that most closely approximate “repeat players” in the courts (e.g., Galanter, 1974). That is, because attorney donors are the most frequent participants in the courts, they arguably have the most to lose from a judge’s mandatory recusal over the long-term—such a recusal, by definition, would extend beyond the instant case to all other cases in which that particular attorney, or his or her firm, might appear before that particular judge until the next election. In other words, attorneys represent the subset of donors that we would expect to be especially loath to undertake actions triggering the recusal of potentially friendly jurists. Indeed, this caution may help explain the shifting of money by attorneys to PACs we have uncovered in examining §12-24, because attorney donors are generally much more likely to appear in a case than are businesses or other interests. T able 4 Percentage of T otal Amount Raised from Donations over Threshold Coming Through PACs Year of Election Attorneys Business 1994 0.15 0.45 1996 0.36 0.65 1998 0.68 0.76 2000 0.81 0.26 2002 0.58 0.37 2004 0.67 0.27 2006 0.90 0.46 2008 0.54 0.30 2010 0.93 0.05
144 THE JUSTICE SYSTEM JOURNAL For a separate project, we have gathered complete data on the donors to judicial campaigns in Georgia, Illinois, T exas, and W est Virginia in 2005. From this data we know that 18 percent of the time a judge will face at least one attorney or law firm who has contributed to her campaign; there is just a 2 percent chance of observing such a relationship between judge and litigant. Furthermore, though we have not yet been able to gather data on litigant contributions in Alabama, our data on attorney contributions in that state in 2005 indicate that a contributing attorney appears 17 percent of the time, in-line with the estimates from other states. The contrast between attorneys and litigants is stark, and though we cannot say for sure whether this exact pattern holds in Alabama until we can gather data on litigants, we suspect that it does. Indeed, the frequent appearance of attorney contributors compared to litigant contributors is one of the reasons why scholars have tended to focus on attorneys versus all others (e.g., Williams and Ditslear, 2007). CONCLUSION In light of our results, it would be unreasonable to describe Alabama’s version of per se recusal as a panacea for either the influence of money in judicial elections or as a solution to concerns over potential judicial bias. With respect to judicial impartiality, as Gibson and Caldeira (2012) note, while recusal has some restorative value for individual perceptions of judicial legitimacy, it does not fully compensate for the circumstances that imperil that reservoir of goodwill in the first place. Furthermore, although we have uncovered evidence that §12-24 has led to a decrease in the raw number of large campaign donations, the overall amount of money contributed to candidates for the Alabama Supreme Court has continued to grow. When all this is considered, there are legitimate questions as to per se recusal’s utility in guarding against judicial bias or its perception. On the other hand, by reducing the number of large donations and funneling them through PACs, the prospect of a law having the potential to limit the number of instances in which judicial impartiality, and thus legitimacy, might be endangered hardly seems insignificant—particularly when that law is not being scrupulously enforced. While many questions in this area must await further research, several considerations serve to underscore the potential relevance that limiting the number of large contributions in judicial races may have for judicial legitimacy. Several portions of Gibson and Caldeira’s (2012) recent experimental work imply that the threat to judicial legitimacy posed by judicial elections is not a problem of campaigns or funding per se; rather, it is the appearance of corruption, or a “quid pro quo relationship between the donor and recipient” (2012:18) that seems to stoke that skepticism. Further, these authors note that “[t]he American people seem not to be bothered by statements of policy positions and policy promises by judicial candidates, or even by the use of advertisements attacking opponents. But campaign contributions are another matter” (2012:19). Put a bit differently, the public seems to accept the necessity of using money in judicial races for advertising, candidate speech on various
THE EFFECT OF PER SE RECUSAL RULES 145 issues, and other campaign-related expenses; however, it is arguably the act of contributing itself where perceptions of judicial impartiality are most endangered. Thus, the idea that “contributions to candidates for judicial office imply a conflict of interest . . . which undermines perceived impartiality and legitimacy” suggests the raw frequency of large donations may be at least as corrosive to the legitimacy of courts as large amounts of money, and perhaps more so (2012:20). The absence of a relationship between citizens’ perceptions of legitimacy and the actual electoral impact of a given donation (2012:27)—which could be conceptualized as being akin to a proxy for donation amount—would seem to reinforce this possibility. It is at least plausible that limiting the overall frequency of large donations, if not the role of money in judicial campaigns generally, could play a constructive role in influencing citizen perceptions of judicial impartiality. Although it is true that Alabama’s per se rule has not stemmed the overall growth of campaign contributions in the state’s supreme court races, the statute does seem to have altered the mechanisms by which donors participate in such races. In discussing Gibson and Caldeira’s (2012) research above, we suggested that such an alteration represents a potentially important effect of such measures on public confidence in the courts. Indeed, the U.S. Supreme Court itself recognized an important distinction between the mere presence of money in judicial elections and the dispositive role played by large individual contributions in those contests in Caperton: “Not every campaign contribution by a litigant or attorney creates a probability of bias that requires a judge’s recusal” (Caperton v. Massey, 556 U.S. 868, 884). 21 But by limiting the number of large contributions from attorneys and by funneling the large contributions that are made through PACs, §12-24 may be able to mitigate some of the damage to legitimacy caused by the appearance of a quid pro quo. The generalizability of our results must, of course, await further investigation. As additional states adopt these rules, it will be important to study how other monetary thresholds and different state legal cultures shape reactions to such per se rules. W e believe the trends we have uncovered in Alabama are likely to repeat themselves elsewhere, except perhaps where the amount triggering per se recusal is quite high. Moreover, the Alabama experience seems to suggest that per se recusal represents a potentially viable, if imperfect, mechanism for limiting the frequency of large donations to judicial campaigns. The lessons of the Alabama case are important to heed as such policies evolve in other jurisdictions (Glick and Hays, 1991), and recusal requirements are revised with greater urgency in the post-Capertonworld of judicial elections. jsj 21 The Court went on to note the critical role of the amount contributed by an individual in assessing the probability of actual bias on the part of a judge: “The inquiry centers on the contribution’s relative size in comparison to the total amount of money contributed to the campaign, the total amount spent in the election, and the apparent effect such contribution had on the outcome of the election” (Caperton v. Massey 556 U.S. 868, 884).
146 THE JUSTICE SYSTEM JOURNAL
Abrahamson, S. (2001). “Speech: The Ballot and the Bench,” 76 New York University Law Review 973. Abramson, L. (2000). “ Appearance of Impropriety: Deciding When a Judge’s Impartiality ‘Might Reasonably Be Questioned,’” 14 Georgetown Journal of Legal Ethics 55. American Bar Association (2008). “Public Dislikes Partisan Political Influence, Special Interest Money, in Judicial Selection.” Press Release. http://apps.americanbar.org/abanet/media/ release/news_release.cfm?releaseid=480 Bam, D. (2010). “Understanding Caperton: Changing the Role of Appearances in Judicial Recusal Analysis,” 42 McGeorge Law Review 65. Bassett, D. L. (2005). “Recusal and the Supreme Court,” 56 Hastings Law Journal 657. — (2002). “Judicial Disqualification in the Federal Appellate Courts,” 87 Iowa Law Review1213. Baum, L. (2003). “Judicial Elections and Judicial Independence: The V oter’s Perspective,” 64 Ohio State Law Journal 161. Beck, N., J. Katz, and R. T ucker (1998). “T aking Time Seriously: Time-Series-Cross-Section Analysis with a Binary Dependent Variable,” 42 American Journal of Political Science1260. Becker, D., and M. Reddick (2003). Judicial Selection Reform: Examples from Six States. Des Moines, IA: American Judicature Society. Bonneau, C. W . (2007a). “Campaign Fundraising in State Supreme Court Elections,” 88 Social Science Quarterly 68. — (2007b). “The Dynamics of Campaign Spending in State Supreme Court Elections,” in M. Streb (ed.), Running for Judge: The Rising Political, Financial and Legal Stakes of Judicial Elections, pp. 59-72. New Y ork: New Y ork University Press. — (2005). “What Price Justice(s)? Understanding Campaign Spending in State Supreme Court Elections,” 5 State Politics and Policy Quarterly 107. Bonneau, C. W ., and M. G. Hall (2009). In Defense of Judicial Elections. New Y ork: Routledge. Brambor, T ., W . R. Clark, and M. Golder (2006). “Understanding Interaction Models: Improving Empirical Analysis,” 14 Political Analysis 63. Brief of the Conference of Chief Justices, Caperton v. Massey (2009). Brief of the States of Alabama, Colorado, Delaware, Florida, Louisiana, Michigan, and Utah (Brief of the States), Caperton v. Massey (2009). Cann, D. M. (2007). “Justice for Sale? Campaign Contributions and Judicial Decisionmaking,” 7 State Politics and Policy Quarterly 281. — (2002). “Campaign Contributions and Judicial Behavior,” 23 American Review of Politics261. Cann, D. M., C. W . Bonneau, and B. D. Boyea (2012). “Campaign Contributions and Judicial Decisions in Partisan and Nonpartisan Elections.” In K. T . McGuire (ed.), New Directions in Judicial Politics, pp. 38-52. New Y ork: Routledge. Chamberlain, A. (2012). “The Growth of Third-Party V oting: An Empirical Case Study of V ermont, 1840-55,” 12 State Politics and Policy Quarterly 343.
THE EFFECT OF PER SE RECUSAL RULES 147 Champagne, A. (1988). “Judicial Reform in T exas,” 72 Judicature 146. — (1986). “The Selection and Retention of Judges in T exas,” 40 Southwestern Law Journal 66. Champagne, A., and J. Haydel, eds. (1993). Judicial Reform in the States. Lanham, MD: University Press of America. Cheek, K., and A. Champagne (2000). “Money in T exas Supreme Court Elections, 1980- 1998,” 84 Judicature 20. Francia, P . L., J. C. Green, P . S. Herrnson, L. W . Powell, and C. Wilcox (2003). The Financiers of Congressional Elections: Investors, Ideologues, and Intimates. New Y ork: Columbia University Press. Frost, A. (2005). “Keeping Up Appearances: A Process-Oriented Approach to Judicial Recusal,” 53 Kansas Law Review 531. Galanter, M. (1974). “Why the ‘Haves’ Come Out Ahead: Speculations on the Limits of Legal Change,” 9 Law and Society Review 95. Gerring, J. (2004). “What Is a Case Study and What Is It Good For?” 98 American Political Science Review 341. Geyh, C. G. (2003). “Why Judicial Elections Stink,” 64 Ohio State University Law Journal 43. Gibson, J. L., and G. A. Caldeira (2012). “Campaign Support, Conflicts of Interest, and Judicial Impartiality: Can Recusals Rescue the Legitimacy of Courts?” 74 Journal of Politics 18. Glick, H. R., and S. P . Hays (1991). “Innovation and Reinvention in State Policymaking: Theory and Evolution of Living Will Laws,” 53 Journal of Politics 835. Goldberg, D. (2007). “Interest Group Participation in Judicial Elections.” In M. Streb (ed.), Running for Judge: The Rising Political, Financial, and Legal Stakes of Judicial Elections, pp. 73-95. New Y ork: New Y ork University Press. Goldberg, D., S. Samis, E. Bender, and R. W eiss (2005). The New Politics of Judicial Elections 2004. Washington, DC: Justice at Stake Campaign. Hall, C., and E. Opsal (2011). “T wo Y ears After Landmark Ethics Case, Courts Lag on Campaign-Case Rules.” Brennan Center for Justice, Press Release, June 6. http://www.brennancenter.org/content/resource/2_years_after_landmark_ethics_case_c ourts_lag_on_campaign-cash_rules/ Hall, M. G., and C. W . Bonneau (2006). “Does Quality Matter? Challengers in State Supreme Court Elections,” 50 American Journal of Political Science 20. Liptak, A., and J. Roberts (2006). “Tilting the Scales? The Ohio Experience; Campaign Cash Mirrors a High Court’s Rulings,” New York Times, October 1, p. A1. McAtee, A., and K. T . McGuire (2007). “Lawyers, Justices, and Issue Salience: When and How Do Legal Arguments Affect the U.S. Supreme Court?” 41 Law and Society Review 259. McCall, M. M. (2003). “The Politics of Judicial Elections: The Influence of Campaign Contributions on the V oting Patterns of T exas Supreme Court Justices, 1994-1997,” 31 Politics and Policy 314. — (2001). “Buying Justice in T exas: The Influence of Campaign Contributions on the V oting Behavior of T exas Supreme Court Justices,” 22 American Review of Politics 349.
148 THE JUSTICE SYSTEM JOURNAL McCall, M. M., and M. A. McCall (2007). “Campaign Contributions, Judicial Decisions, and the T exas Supreme Court: Assessing the Appearance of Impropriety,” 90 Judicature 214. McGuire, K. T . (1995). “Repeat Players in the Supreme Court: The Role of Experienced Lawyers in Litigation Success,” 57 Journal of Politics 187. Moller, E. K., N. M. Pace, and S. J. Carroll (1994). “Punitive Damages in Financial Injury Jury V erdicts,” 28 Journal of Legal Studies 283. Orndorff, M. (2010). “Lawsuit Seeks Justice Department Review of Unenforced 15-Y ear-Old State Law,” Birmingham News, July 22. http://blog.al.com/sweethome//print.html Raftery, W . E. (2010). “‘The Legislature Must Save the Court From Itself?’ Recusal, Separation of Powers, and the Post-Caperton W orld,” 58 Drake Law Review 765. Rock, E., and L. Baum (2010). “The Impact of High-Visibility Contests for U.S. State Court Judgeships: Partisan V oting in Nonpartisan Elections,” 10 State Politics and Policy Quarterly 368. Ross, A. D., S. M. Rouse, and K. A. Bratton (2010). “Latino Representation and Education: Pathways to Latino Student Performance,” 10 State Politics and Policy Quarterly 69. Sample, J. (2010). “Court Reform Enters the Post-Caperton Era,” 58 Drake Law Review 787. Sample, J., and D. E. Pozen (2007). “Making Judicial Recusal More Rigorous,” 46:1 Judges’ Journal 17. Sample, J., D. Pozen, and M. Y oung (2008). “Fair Courts: Setting Recusal Standards.” Paper, Brennan Center for Justice, New Y ork University School of Law. http://brennancenter. org/content/resource/fair_courts_setting_recusal_standards/ Skaggs, A., and A. Silver (2011). “Promoting Fair and Impartial Courts Through Recusal Reform.” Paper, Brennan Center for Justice, New Y ork University School of Law. http://www.brennancenter.org/content/resource/promoting_fair_courts_through_recusal Sample, J., A. Skaggs, J. Blitzer, and L. Casey (2010). The New Politics of Judicial Elections, 2000- 2009: Decade of Change. Washington, DC: Justice at Stake Campaign. Supreme Court of T exas (1999). “Supreme Court of T exas Judicial Campaign Finance Study Committee: Report and Recommendations,” February 23. Waltenburg, E. N., and C. S. Lopeman (2000). “T ort Decisions and Campaign Dollars,” 28 Southeastern Political Review 241. Ware, S. J. (2002). “Money, Politics, and Judicial Decisions: A Case Study of Arbitration Law in Alabama,” 30 Capital University Law Review 583. — (1999). “Money, Politics, and Judicial Decisions: A Case Study of Arbitration Law in Alabama,” 15 Journal of Law and Politics 645. Williams, M. S., and C. A. Ditslear (2007). “Bidding for Justice: The Influence of Attorneys’ Contributions on State Supreme Courts,” 28 Justice System Journal 135.
THE EFFECT OF PER SE RECUSAL RULES 149 CASES CITED B.M.W . of North America, Inc. v. Gore,517 U.S. 559 (1996). Bracken v. T rimmier Law Firm,879 So.2d 207 (2004). Buckley v. V aleo,424 U.S. 1 (1976). Caperton v. A.T . Massey Coal Corp.,556 US 868 (2009). Citizens United v. FEC, 558 U.S. 8 (2010). Little v. Strange, 796 F . Supp. 2d 1314 (2011). Little v. King et al., 768 F . Supp. 2d 56 (2011). Republican Party of Minnesota v. White, 536 U.S. 765 (2002). STATUTES CITED Code of Ala. §12-24-2 ABA Code of Judicial Conduct, R.2.11(A)
150 THE JUSTICE SYSTEM JOURNAL
COMPARING ALABAMA AND TEXAS Given that ours is a one-state study taking place during a time of considerable partisan change in the South, we thought it prudent to gather some additional data on judicial elections in a state comparable to Alabama to bolster our confidence that §12-24 is producing results unique to Alabama rather than reflecting broader natural trends. W e chose T exas as our comparison due to the similarity of its judicial elections to Alabama’s in several key respects. Alabama and T exas both use partisan, statewide elections with six-year terms; both have experienced marked shifts toward Republican control through the 1990s and early 2000s; and both states have expensive judicial elections. Therefore, we believe that Alabama and T exas are comparable in the ways that are most likely to matter in discerning the effects of the per se recusal statute. W e undertake an initial comparison between the patterns of attorney and business giving in the two states between 1994 and 2010. Figure A1 plots the total amount of money raised from attorneys by judicial candidates. The two states exhibit distinct patterns, despite similarities in the nature and type of judicial elections in them and in their overall partisan environments. In Alabama, attorney giving plummets from 1994 (when it comprises 46 percent of the total raised by candidates) with the enactment of the statute and never recovers. The pattern is exactly the opposite in T exas, where attorney giving generally exceeds the levels seen in 1994 and never drops lower than 37 percent of the total. This suggests that there is a divergence in how attorney donors in these two states approach judicial candidates, and the timing of this divergence in approaches strongly suggests that the per se recusal law plays a role. Finally, to bolster confidence in our findings with respect to the movement of large donations through PACs (see T able 4) we present a comparison of how attorneys in Alabama and T exas gave large donations to judicial candidates, where large is defined in terms of exceeding the $4,000 threshold set by §12-24. If the statute has caused attorneys to shift their large donations into PACs to avoid potentially triggering the recusal law, then we should expect to see a marked increase in the percentage of the large donations funneled through PACs in Alabama after enactment of the statute, but no concomitant change in the behavior of large attorney donors in T exas.
Figure A2 presents the expected pattern. As a percentage of the total, the amount funneled through PACs in Alabama is always at least twice as high as it is in T exas.
Therefore, though it is clear that PACs become more important to attorney giving in both states over time, their importance in Alabama is disproportionate to their importance in T exas.
THE EFFECT OF PER SE RECUSAL RULES 151
Attorney Giving in Alabama and T exas
Percentage of Large Attorney Donations Given through PACs