BIDDING FOR JUSTICE: THE INFLUENCE OF ATTORNEYS' CONTRIBUTIONS ON STATE SUPREME COURTS*

The need to launch a successful campaign raises issues concerning the influence if campaign contributions-contributions that become increasingly necessary as state judicial campaigns become more expensive. Campaign contributions then become a potential influence on the decision making of state court judges. State judges not only are beholden to the preferences of the voters they face at reelection, but also must satisfy the attorneys and interest groups contributing to their campaigns (Abbe and Hernson, 2002). While past studies have uncovered some influence for contributions on the votes of judges in states with highly competitive judicial elections and highly partisan systems (Hood and Emmert, 2001; McCaIl, 2001, 2004; Waltenburg and Lopeman, 2000), states more moderate in terms of both competition and partisanship have been neglected in the body of research (Cann, 2002).

While there are a number of studies that consider the impact of contributions in states that have not reformed their campaign finance laws (see, for example, McCaIl, 2004, and Waltenburg and Lopeman, 2000), an empirical test of the influence of campaign contributions on judicial decisions in states that attempt to curb the influence of money has not been undertaken and, thus, is necessary. Given what many believe to be the undue influence of campaign contributions on the decisions of state court judges, the repeated calls for reform in the literature, and the efforts to reform the system, it is important to consider whether such influence actually exists (Hensler, 1999). It is possible that while money appears to influence judges' actions, once one accounts for factors other than campaign contributions, the apparent influence of money may disappear.

The present study examines the influence of attorneys' campaign contributions on the behavior of the judges of the Wisconsin Supreme Court. We examine the influence of attorneys' contributions on the judges' decision making over a ten-year period, holding constant such traditional influences as ideology and intra-court influence. By examining the link between donations and the votes of the individual judges, we can better understand if money does, indeed, influence decision making. If votes are influenced by donations in a state that has attempted to curb the influence of money and politics, reformers of the judicial system may have to consider more drastic measures to address the problem of contributions to judicial campaigns on the decision making of judges. If we find no relationship between attorneys' contributions and votes, other states might profitably consider adopting reforms similar to those used in Wisconsin. Put differently, if we find money to be an influence on judicial decisions in Wisconsin, we are likely to find such an influence in states that have not adopted reforms.

The effects of state attempts to curb the influence of money in politics have not been considered here. We make no claims about how contribution limits have changed the behavior of people contributing to the campaigns or about the effectiveness of reforms themselves. An examination of the changes in contribution totals before and after reforms would be interesting, but it is beyond the scope of this study.

The broad concern in the literature is the impact of campaign contributions on decision making on the bench. That literature has long been concerned with the effect of attorney contributions specifically (see, for example, Alfini and Brooks, 1989). For the Wisconsin Supreme Court, the campaign contributions that might have an effect on decision making are overwhelmingly attorney contributions. Attorney contributions are common in Wisconsin, and they create the biggest problems for judicial independence. Attorneys have repeated interactions with judges, both in the courtroom and in the larger legal community. If attorneys are likely to contribute to judges, and if they receive a benefit for the contribution, judicial independence and the fairness of the decisions by the court are both called into question. Attorneys would not only have the advantage of repeated access to the members of the Wisconsin Supreme Court, but also might receive the additional benefit of their contributions. This would be most obvious in cases in which the attorneys were not simply representing the parties to a case but were themselves the parties, when, for example, they were sued for legal malpractice or were the subject of attorney discipline proceedings. If attorney contributors gain advantages from their contributions, other actors who come into contact with the members of the Wisconsin Supreme Court, be they members of the general public, business interests, or interest groups, would be unable to overcome the advantages attorneys have, no matter how persuasive the argument. With the other advantages attorneys already experience, it is important to focus on this group of contributors who are the most likely to receive some direct benefit from getting their preferred judge elected, because if influence does exist, it is most likely to occur with the most prevalent contributors. If influence does not exist for attorney contributors, it is unlikely to exist for other, less-frequent contributors.

Given the importance of attorney contributions, in this research we examine only contributions by attorneys. The data we use suggest that the only consistent contributors are political action committees (PACs) and attorneys. Sixty percent of the attorneys in the database we used (discussed below) contributed to the campaign of at least one Wisconsin Supreme Court judge during the period studied; they are the largest single source of contributions to the campaigns of Wisconsin Supreme Court judges. Attorneys, either those serving as the lawyer for a party or those who were the parties themselves, were also a large majority of those contributing to the campaigns of judges who participated in the cases in the database. By contributing frequently and participating more than any other contributor, attorneys have the greatest opportunity to influence the members of the Wisconsin Supreme Court. That 60 percent of the attorneys in the database made at least one contribution to a judge on this court suggests that these attorneys think they are getting something for their contribution. Because attorney contributions are the most common type of influence, we seek to explore what effect attorney contributions are having on judicial behavior in Wisconsin.

Attorneys and PACs were not the only contributors to the campaigns of judges in Wisconsin. The judges themselves contributed to their own campaigns. The remaining contributors who also participated in cases-mostly members of the general public-were too few in number to be analyzed. Other contributors-including friends of the judges, other politicians, and members of the general public-did not participate in cases, and, because they did not participate in the cases in this database, it is impossible to determine whether they had any influence. More specifically, we are unable to determine what side, if any, they would favor, and, thus, if they were successful in swaying the judge to that side with a contribution. We do not preclude the possibility that an individual nonattorney contributor might have some tie to a case, such as owning stock in the company that is a party to the case, but at this time we are unable to determine these types of links, and moreover, we remind the reader this would be a minority of the cases in the database.

THE PUZZLE: THE INFLUENCE OF MONEY

The mere appearance of an influence of money on decision making in state courts has led many in politics and the legal community to call for reform of the system (see, for example, Abramson, 2000). Throughout the late 1990s and in the first few years of this century, at least twelve states that elect judges, including Wisconsin, actively reviewed ways to reform the system to eliminate the influence of campaign contributions on the votes of judges (see, for example, Champagne, 2001). Those calling for reform included members of the bench in several states, including Wisconsin (Abrahamson, 2003; Baschab, 2001; Wintersheimer, 2001). At the same time, a host of legal academics were quite vocal in calling for reform; some objected to private funding of judicial elections (Bowers, 2003; Goldberg, 2003; Schotland, 2003) or donor anonymity (Ayres and Bulow, 1998); still others suggested the existence of civil'rights violations in electing judges with private contributions (Smith, 2001). Regardless of the particular concern expressed, there appears to be a consensus that the influence of campaign contributions in electoral systems creates problems for judicial independence. Both scholars and practitioners who call for reform argue that votes follow contributions; liberal donations will result in liberal votes by the judges, and the same is true for conservative donations and conservative votes.

Proposed solutions range from elimination of judicial elections completely (Geyh, 2003; Tarr, 2003) to the recusal or disqualification of judges from cases when campaign contributors appear before them (Cann, 2002; Nagle, 2000). However, most proposals are said to fall short of addressing the cause of the problem (see, for example, Baum, 2003; Southwick, 2002).

Possible parallels between legislative campaign contributions and those for judicial elections have been explored. Bonneau (2005) notes the similarities between judicial and legislative campaign contributions. Both types of candidates use contributions to gain support, challenge incumbents, and launch effective campaigns. Thus, we should expect the influence of campaign donations on judges to be similar to the influence of contributions on members of Congress (see, for example, Erikson and Palfrey, 1998, and Jacobson, 1990). However, those studying the influence of contributions on judicial decision making are ambivalent as to whether the money is a reward for past decisions, an attempt to influence future decisions, or simply a way to gain access (Cann, 2002; Eakins, 1999; Hood and Emmert, 2001; McCall, 2001, 2003; Waltenburg and Lopeman, 2000). While students of Congress struggle to determine the effect of contributions on representatives' decisions, most studies of Congress suggest that contributions can buy access but do not necessarily result in influence, while other studies of Congress suggest that contributions yield neither (Ainsworth, 1997; Austen-Smith and Wright, 1994; but see Chin, Bond, and Geva, 2000). Past studies of contributions and the judiciary are clear in their expectations of a relation between contributions and influence, but these studies rarely find that contributions have an effect. Moreover, a number of problems related to case selection and factors selected for examination cast doubt on the results of these studies. Additionally, analysis in a number of past studies has been restricted to a single issue area (McCall, 2001, 2003; Waltenburg and Lopeman, 2000), making it difficult to generalize about the influence of contributions on the votes of state court judges beyond such issues.

A number of past studies measure whether either one side, typically the plaintiff or defendant, contributed in a case, or which side contributed the most (Cann, 2002; McCall, 2001, 2003; Waltenburg and Lopeman, 2000). Such measures, while interesting, may not tap the true influence of contributions on vote choice. Given that there are a number of people who contribute to a judicial campaign, it is possible that only the most visible contributions-the largest dollar amounts-or contributions by an ideological opponent catch the attention of a state court judge. Moreover, examinations of the influence of contributions on case outcomes ignore a core issue: contributions are made to individual judges and, thus, should directly affect individual behavior. This could include the writing of concurrences or dissents or changes in the substantive language of precedent, but these direct effects are beyond the scope of this study.

Changes in individual votes may or may not produce a change in case outcomes, depending on how close the division is among the judges. In addition, a judge receiving a contribution may persuade colleagues during conference or during opinion writing, a matter not examined here, but measuring this indirect influence is difficult because it requires access to the private papers and conversations of the judges, which may explain why the assumption is untested in past research. Any direct effect on an individual judge may seem minimal if it does not alter the outcome of a case, but when attorneys contribute, they do not know the potential margin of victory in future cases. Yet the odds of winning might be increased: if moving one judge could be the difference between a loss and a win, attorneys would be more likely to contribute to judicial campaigns.

We expect the effect of attorneys' contributions on votes to hold even after controlling for the ideology of the judge and the influence of the other members of the Wisconsin Supreme Court. That assertion indicates the importance of the measure of ideology used, which has been another problem in past studies. Without an appropriate control for ideology, it becomes difficult to distinguish if the donations are given to a judge with the same ideology as the contributor to reward the judges' past behavior and to reinforce it, so that the judge will continue it in the future, or instead are made in an attempt to persuade a judge to shift position, perhaps from a moderate stance on a question to a more ideologically extreme position. In addition, some measures of ideology ignore the influence of other members of the court (Cann, 2002; McCall, 2001, 2003; Waltenburg and Lopeman, 2000), and, further, some studies use measures of ideology that are not comparable across state supreme courts, making replication and extension of such research difficult to accomplish.

WHY WISCONSIN

Our study examines the effect of contributions by attorneys on the votes of the members of the Wisconsin Supreme Court between 1989 and 1999. The similarities of Wisconsin to other states and its differences from them make it an appropriate site for an empirical test of the influence of private campaign contributions on decisions. Given that past studies have focused on Texas, Ohio, and Michigan, all of which are heavily partisan states with contentious elections, Wisconsin, with nonpartisan judicial elections that are in part publicly funded, may seem an unusual choice through which to study the influence of contributions on judicial decisions. Yet if money influences decisions in Wisconsin, a less-partisan state with public financing, it likely also influences decisions of judges in many other states.

Several factors contribute to the appropriateness of Wisconsin for such a study. Even with public money available, Wisconsin experienced an increase in the costs of elections similar to increases in other states. The average total donation in judicial elections in Wisconsin increased significantly over the last few years, from an average of $194,642 in donations for 1989 to an average of $656,202 in donations in 1999 (Sanchez, 2001). In addition to a significant increase in the total dollar amounts for judicial elections, Wisconsin has also experienced an increase in the number of contributions. While contributions were made during all ten years of this study, there was a clear increase over time in contributions per year. In fact, a majority of the contributions included in this study were donated between 1996 and 1999.

While there was a significant increase over time in contributing behavior, because past studies show that the impact of contributions on the votes of judges declines as the time between contributions and votes increases (McCall 2001, 2003), we are interested in how contributions were spread across the terms of judges. No judge received a majority of contributions in an election year. The spread of contributions across the terms of judges suggests the potential for contributions to influence the behavior of judges throughout their tenure on the court.

Because judges in Wisconsin, like those elsewhere, can easily determine, through state records and data collected and publicized by private organizations, who donates to their campaigns, the system of contributions to judicial elections allows for the possibility that those contributions will influence the decisions of individual judges. Given the outspoken opposition of Wisconsin chief justice Shirley Abrahamson (2003) to private contributions, we believe there is sufficient reason to think judges know who is contributing to their campaigns.

Also making Wisconsin appropriate for a study of contributions to judicial campaigns is that the Code of Judicial Conduct in Wisconsin is almost identical to those in previously studied Michigan and Ohio. Under these codes, judicial candidates are limited from directly soliciting a contribution but candidates are not restricted from knowing who is contributing to their campaigns (Wisconsin Code SCR 60.06, Ohio CJC Canon 7C2(a), Michigan CJC Canon 7B2(a)). Moreover, the nonpartisan system used to select the judiciary in Wisconsin is also used in Ohio and Michigan, as well as in thirteen other states. However, while the elections in three previously studied states are formally nonpartisan, Ohio and Michigan use partisan nominating procedures but Wisconsin does not. The absence of partisan primaries may explain why Wisconsin's judicial elections have historically been less contentious than those in Ohio and Michigan. In addition, the judges of Wisconsin serve ten-year terms, longer than those of most state judges, and the judicial elections, while contested, are not as competitive as in some other states (Hall, 2001). The greater length of the judges' terms suggests that any contribution made will have a diminished effect on votes in that case over time.

Wisconsin is different from other states in a number of other ways. For one thing, Wisconsin has a higher campaign contribution limit than states considered in past research; the limit in Wisconsin is $10,000, exceeding limits in Texas and Ohio by $5,000 and Michigan by $6,600 (Goldberg, 2002). Wisconsin is also somewhat unusual in having a system of public financing of campaigns, although such a system is used by North Carolina and is being considered by additional states (Common Cause, 2005). The early adoption of public financing reflects the state's moderate political culture, which includes a noted distaste for the influence of campaign contributions, best exhibited by the careers of the late Senator William Proxmire and of incumbent Senator Russell Feingold. Public finance dollars, in constituting only 2 percent of the money spent by all candidates in a judicial race and only 4 percent of what any individual candidate spent (Sanchez, 2001), account for far fewer dollars than do private contributions to candidates for the Wisconsin Supreme Court. Wisconsin also differs from previously studied states in the sources of contributions: attorney contributions outnumber PAC contributions in Wisconsin. In fact, judges contribute more money to their own reelection campaigns than PACs contribute to them (see Table 1).

THE INFORMATION USED

Between 1989 and 1999, seven judges won elections to the Wisconsin Supreme Court; three of them (Bablitch, Geske, and Steinmetz) retired from the court after the period of study. One of the judges, Chief Justice Shirley Abrahamson, won election twice; she was initially elected in 1989 and was reelected in 1999. All eight elections in the study period were contested. Five judges are excluded from the analysis. Four were excluded because, while they made decisions in cases included in the study, they reached the mandatory retirement age during their last term. Because potential contributors knew well in advance that these judges were not eligible for reelection, they would not make contributions to them, so that there would be no contributions in the period studied to influence their behavior. A fifth judge began her term on the court through an interim appointment in 1999. While she did make decisions on the court, she was not engaged in a campaign during the period studied. It is possible that an individual might have contributed to her campaign, but the overlap between her tenure and our period of study was so short that no such contributions occurred.

Only cases decided after the first contribution to a given judge were included in the study. This was done to avoid problems in interpreting the purposes of the contributions, that is, the motives of those contributing-whether they are used as a reward for past decisions or as a means of influence over a future decision. We are unable to determine, for example, if contributions were given to a judge in a case awaiting review or on its way to the court in order to influence the court to accept the case.

The contribution data for our study come from the National Institute for Money in Politics, a nonpartisan campaign organization, which serves as a watchdog over the influence of campaign contributions on multiple institutions of government. To conduct this analysis, we used two of the institute's data sets. The first chronicles all contributions to the campaigns of the judges on the Wisconsin Supreme Court from 1989 through 1999, and the second contains all cases the court decided during the corresponding time period. From the second data set, we gathered 3,811 judge votes from 870 cases; those votes are the observations used in this study. While direct contributions are not the only potential influences on judges, indirect contributions through "soft money" expenditures are not part of the available data and have not been considered.

CONTRIBUTIONS AND VOTES: HOW EXAMINED

Our analysis of the influence of campaign contributions differs from previous studies in a number of ways. As in some past studies, we consider the individual judge's vote in each case during the period of study. However, instead of treating the vote as favoring either the plaintiff or defendant, we treat the vote as liberal or conservative, because judges on state supreme courts are not necessarily inclined to favor a plaintiff or defendant in the same way they are inclined toward a specific ideological orientation. Such an approach allows us to include a broader array of cases than in other studies and to make more general conclusions about the influence of contributions on the votes of state supreme court judges. In past studies, choice of a court controlled by a single party and of a single issue area, such as torts, created a de facto control for ideology (see, for example, McCaIl, 2001, 2003). However, pro-plaintiff or prodefendant decisions in tort cases have an ideological outcome distinct from the same decision in a criminal case. The use of the ideological direction of a vote, while unique among studies examining the effect of contributions on the votes of state supreme court judges, is consistent with the attitudinal model of judicial decision making, which suggests that judges make their decisions consistent with their policy preferences and not solely with respect to the existing law (Segal and Spaeth, 2002).

Another difference between our study and those conducted in the past is the type of contributions included in the analysis. Only attorney contributions, not those of corporations, interest groups, individuals, or the parties to the case, are considered in the present study. As noted in Table 1, attorneys outnumber PACs both in the number of contributions and the amounts given (Sanchez, 2001). Attorneys' contributions are the largest single source of contributions to judges in Wisconsin. Only a small proportion, 20 percent, of the contributors to judicial campaigns were attorneys, but the proportion of contributors identified as PACs is much smaller. However, attorneys are not only more likely than PACs to contribute to the campaigns of judges, but also more likely than any other contributor both to contribute to the judge and then to participate in a case before the judge. PACs are the second largest source of contributions, with the remaining contributors composed of friends of the judges, other politicians, the judges themselves, and members of the general public, many of whom appear to have no tie to the cases in the database. Even if many of these contributors were tied to cases in the database, the diversity of the third group of contributors is such that the group defies classification; there is no category into which these contributors can fall besides a catchall "other."

It is impossible for us to determine at this time what tie, if any, these other contributors might have to any case in the database, as we know very little about the connections between the third group of contributors and the decisions of the judges. One possible connection would be as an amicus curiae (friend of the court). However, participation by an organization in an amicus curiae role occurs in only 10 percent of the cases in our dataset, and only five cases in 1,000 had an amicus participant who had contributed to one of the judges. A second possible connection exists with the other contributors (friends, individuals, and corporations), but because of the limited participation by corporations and individuals, we have chosen to focus on contributions by attorneys and not to look at the contributions by the parties, with an important exception-if the parties were also attorneys. This took place with some frequency, particularly in lawyer-malpractice cases and attorney-discipline matters, with such attorney involvement occurring in roughly one-sixth of the cases considered in this study. Ideally, a study of the effects of contributions would include all contributors who participated in cases, but the data for Wisconsin contain contributions by nonattorneys in numbers insufficient for analysis. We thus remind the reader that anyone who contributed but did not participate, be they interest groups, PACs, attorneys, parties, or members of the general public, was excluded from the analysis.

What do we mean by "liberal" and "conservative" contributions? We do not use the ideology of the individual attorney, in part because determining the ideology of each of the attorneys participating before the court during this period is not possible. Instead, a contribution coded "liberal" is one made by an attorney representing the liberal side in a case. Thus, it is possible that an attorney whose personal ideology is liberal may be coded as conservative for representing the conservative side in a dispute. We do not believe this coding is problematic because the attorney would still be concerned about winning the case that the contribution might affect. While attorneys have no certainty about their opportunities to win their cases, their repeated experience will provide them a sense of how the court is likely to decide and how to frame arguments to increase their chances of success (see, for example, Parole, 1999). It is possible contributions are seen as another tool to increase the probability of winning.

Some attorneys with repeated experience had more than one case before the Wisconsin Supreme Court during the period studied. Our ability to control for the temporal sequence of cases and contributions means that this fact does not impinge negatively on the study. Attorneys often brought one case in which no contribution was made, while they chose to contribute during another case. With the vote in each case the object of our study, we are able to consider the discrete effect for each contribution while holding constant any other case brought by the same attorney. That some lawyers are repeat players before the court also creates the possibility that some will represent the liberal side in one case and the conservative side in the next. In this situation, if the lawyer contributes in both cases, the attorney would be coded as liberal for the contribution in one case and as conservative in another case. However, of the 1,106 attorneys participating during the period studied, only eleven contributed both as the attorney representing the liberal side in one case and as the attorney representing the conservative side in another. This number is too small to have a noticeable effect on our results.

To consider all possible effects of contributions on the votes of judges, we employ three different measures of contributions: the relative advantage of the highest contributor, the advantage of a contribution on either side in a case, and the advantage of a larger-than-average contribution on either side. Each measure is used separately to analyze the influence of contributions on the votes of Wisconsin Supreme Court judges. For all three measures, we expect that liberal contributions will result in liberal votes, and conservative contributions will lead to conservative votes, by the Wisconsin Supreme Court. To consider the impact of time, we measure the length of time from the contribution to the judge's vote in the case. We expect that the greater the time since the contribution, the less influence the contribution will have on the judge's vote.

The first of the three measures of contributions focuses on the advantage in the total amount of contributions, so we can see if such an advantage provides the contributor's side with a greater likelihood of securing the vote of the judge. In that measure, we take the aggregate dollar amount for the liberal and conservative sides and subtract the conservative from the liberal. Thus, if the liberal side has the advantage, the measure will be positive, and if the conservative side has the advantage, it will be negative; if the two sides contributed equally or there were no contributions by either side, the value is zero. This measure of contributions is similar to that used in other studies (see Waltenburg and Lopeman, 2000, and Hood and Emmert, 2001), in that it is a determination of the relative advantage possessed by a particular sidehere, the liberal side.

To consider the possibility that judges will simply view any contribution on a particular side as the important factor, whether or not the liberal and/or conservative sides contributed is examined. This measure of contributions is consistent with past studies (see Cann, 2002, and McCaIl, 2001) in that it accounts for contributions by both sides, with no contribution by either side as the baseline for comparison.

Because it is possible that judges will only remember, or pay attention to, large contributions or those greater than the average, and because we wish to test for the effect of larger-than-average contributions, in our final measure of contributions, we include two measures of "Above-Average Contributions" to test whether there was a contribution on the given side in excess of the $97 average contribution by all individual contributors. Using a single cut point for big versus small contributions may seem artificial, but given that the majority of the judges' contributions ranged from $1 to $75, we believe it likely that any amount donated over the $97 level will gain a judge's attention.

In addition to the measures of campaign contributions, we have also taken into account a number of other factors that may affect the vote of a judge. These factors, if not included, could raise doubts about whether the influence we find is the result of contributions or of the characteristics of the judge or case. One obvious element for which we must control is the ideology of the judge making the decision. In measuring the judges' ideologies and their votes, we use the same coding as has been used in prior studies of judicial behavior, most notably those by Segal and Spaeth (2002) and Brace, Langer, and Hall (2000). Liberal decisions, for example, would be those upholding the rights of the individual in civil-liberties claims against the government, while liberal decisions in economic cases would uphold government regulation of the economy or uphold the position of the underdog. (For more information on the coding of liberal and conservative votes, see the Spaeth Database, 2004.)

We control for the judge's ideology in two ways, one relating to the behavior of all judges within the same analysis and one relating to the behavior of judges across analyses. In the first set of analyses, we combine all the votes of the judges on the Wisconsin Supreme Court and control for the ideology of each individual judge with their Party Adjusted Ideology (PAJID) Score (Brace, Langer, and Hall, 2000). These scores, in which a higher score indicates a more liberal judge, measure the ideology of the citizens of the state at election time and adjust this ideology by the party of the judge-in Wisconsin, the party endorsing the judge. Thus, the scores account for both ideology and party affiliation, two influences known to affect the behavior of judges. Because these judges do not rely on party labels to get elected, these scores are appropriate for measuring the ideology of members of a nonpartisan judiciary, and they also allow extension of this research to other states, where party labels could have a very different meaning from the meaning they have in Wisconsin.

The second of the two measures is the voting behavior of each judge, derived from an analysis of that judge's individual voting. We ran a separate analysis for each judge. The separate analyses allow us to determine if there are effects of contributions on some judges that would be masked when all judges are included in a single model. Related to this measure is a control we have included for the intra-court influence on the judge (Hall and Brace, 1989). The influence of the rest of the judges on the court is examined through the ideology of the median judge on the court: the higher the score of the median judge, the more liberal that judge is. As the number of liberal judges on the court increases, liberal outcomes become more likely. Not only are there more judges on the court who are liberal and who would uphold the liberal position in the case, but conservative members of the court may vote with their liberal colleagues simply to avoid the costs of writing separately.

In addition to contributions and die ideological dynamics of the Wisconsin Supreme Court, our analysis includes a number of legal factors known to affect the decisions of judges. Past research found that the position of the government in a case is more likely to be upheld by high courts (see, for example, Parole, 1999). We thus included a control for the position of the government in the case. With judges more likely to vote for die position of the government, conservative government positions decrease the likelihood of a liberal vote while liberal government positions increase the likelihood of a liberal vote. cases in which the government did not take a position are the baseline for comparison. We would also expect the lower-court decision to affect the likelihood of a liberal vote by the judge, because of the suggestion that judges may be inclined to take cases that would allow them to broaden a decision with which they agree on ideological terms (see, for example, Boucher and Segal, 1995). Consistent with these findings, we expect that a liberal ruling by the Wisconsin Court of Appeals, that state's intermediate appellate court, will increase the likelihood of a liberal vote.

Finally, we have included a number of areas of law in our analysis. There are seven such issue areas: economic, which includes cases such as tort issues and bankruptcy; criminal; family, which includes divorce and custody; environmental; government powers, which includes public-utility-regulation cases as well as election disputes; professional responsibility, the primary component of which is attorney discipline but which also includes lawyer malpractice; and procedural issues, which is the baseline for comparison. We controlled in two ways for each of the issue areas. We used controls included in the analysis reported here, and we performed separate analyses for each issue area (not reported here). The results did not differ between the two types of controls. Including several areas of law permits greater understanding of the effects of contributions on judges' votes, but it also entails a sacrifice. Because a wide variety of issue areas are included, we are unable to control for all case facts that may affect the judges' votes. Procedural matters were chosen for the baseline because of the lack of an ideological direction to the cases. If there is an influence of contributions on votes in procedural cases, one would expect that the effect would be even greater in more contentious areas of law, like criminal matters, where judges' votes seem more ideologically based.

RESULTS AND DISCUSSION

The impact of contributions on votes of the Wisconsin Supreme Court judges was tested in three sets of analyses. The first is an examination of the impact of dollar amounts of contributors; the second is an examination of the influence of the presence of a contribution from either the liberal or conservative sides; and the third is the impact of larger-than-average contributions. For all three sets of analyses, we find that we have correctly specified the factors influencing the votes of the members of the Wisconsin Supreme Court.

In the examination of the impact of dollar amounts, the results for most of the factors corroborated our expectations (see Table 2).1 While the length of time since the contribution did not have a statistically significant effect, it did have a positive influence as predicted. The judges support the position of the state government; when the government is on the liberal side, the judges are more likely to vote liberally. The results for a number of the issues also were statistically significant, suggesting that the type of case affects judges' willingness to vote in a liberal direction.2 Finally, the findings for individual judges' ideologies suggest that liberal judges vote liberally, but the court's collective ideology may have different effects: as the court as a whole moves in a more liberal direction, the probability of a liberal vote for any individual judge decreases. The finding runs counter to expectation and may be a function of the court becoming more ideologically extreme, which may make it more difficult to achieve consensus.

Another finding that runs counter to expectations is that above-average contributions by the liberal side's attorney decreased the likelihood of liberal votes. When there is no above-average contribution by the liberal side, the probability of a liberal vote is roughly 50 percent, but an above-average contribution by the liberal side decreases the probability of a liberal vote by 10 percent.

What are the effects of contributions on the votes of the judges of the Wisconsin Supreme Court when one side contributes more than the other? The contribution advantage ranges from positive to negative values. A positive value means that the liberal side contributed more and thus has the advantage, while a negative value suggests a conservative advantage. When neither side has a contribution advantage, each member of the Wisconsin Supreme Court votes for the liberal position almost two-fifths of the time (between 36 and 41 percent). For at least two of the judges, contributions or time since contributions affect their likelihood of voting liberally in the case (see Table 3). When there is a liberal contribution advantage, Judge Bradley is more likely to vote liberally. When the conservative side in the case has the greatest contribution advantage, Judge Bradley has a 22 percent probability of voting liberally, but when the liberal side has the greater contribution advantage, Judge Bradley's probability of voting liberally rises to 97 percent. This astounding change in the probability of voting liberally suggests that contributions are influencing the voting behavior of Judge Bradley.

By contrast, Judge Geske appears to be affected by the time since a contribution and not the balance of contributions. The longer the time since a liberal contribution, the less likely Judge Geske is to vote for the liberal side. Using the minimum liberal contribution advantage ($25), if there has been less than a year since the liberal side's contribution, Judge Geske has a 72 percent probability of voting liberally in the case, but with the passage of the longest time between contribution and decision (five years), the probability of Judge Geske voting liberally drops to 51 percent. When the contribution is set to the maximum amount and the time of the contribution is within a year, Judge Geske has a 78 percent probability of voting liberally; with the passage of five years that probability drops to 53 percent.

When the likelihood of a judge on the court voting liberally after each side has made a contribution is examined, once again it appears that contributions affect the behavior of only a small number of the judges (see Table 4). Again, Judge Bradley appears to be less likely to vote liberally when there is a contribution on the conservative side of the case. In the absence of a contribution from either side, Judge Bradley has a 91 percent probability of voting for the liberal outcome, but when the conservative side made a contribution, Judge Bradley's probability of voting liberally drops to 6 percent. Moreover, the time since the contribution affects the likelihood that Judge Bradley will vote liberally. A conservative-side contribution within the last year gives Judge Bradley an 8 percent probability of voting liberally. Five years after the conservative-side contribution, Judge Bradley has returned to being more likely to vote for the liberal side (a probability of 95 percent). Thus, it appears that the closer the contribution is to the decision in the case, the more likely the conservative side will sway the vote of the judge.

When we examine the effects of contributions that exceed the average, we find that two judges appear to be affected by larger-than-average contributions (see Table 5). Judge Steinmetz is less likely to vote for the liberal position in the case when there is a larger-than-average contribution on the conservative side. When neither side made a larger-than-average contribution, Judge Steinmetz has a 76 percent probability of voting liberally, but a larger-than-average contribution by the conservative side decreases that probability to 46 percent. Interestingly, Judge Wilcox appears slightly more likely to choose the liberal side in a case where the conservative side made an above-average contribution to his campaign. When neither side made an above-average contribution, Judge Wilcox has a 96 percent probability of voting liberally. When the conservative side makes an above-average contribution, the probability of Judge Wilcox voting liberally increases slightly to 99 percent. The greater the amount of time since the aboveaverage liberal contribution, the less likely Judge Wilcox is to vote for the liberal position. Within a year of an above-average liberal contribution, Judge Wilcox has a 94 percent probability of voting liberally, but five years after the contribution, the judge's probability of voting for the liberal position drops to 35 percent. Thus, it seems that conservative contributors cannot win Judge Wilcox to their side, but the liberal side gains an advantage for the short period after a contribution is made.

Whatever the measure of contributions used, the judges of the Wisconsin Supreme Court appear to take into account the position taken by the rest of the court. When the judge is more liberal than the court, the judge is more likely to vote liberally. However, when the court is more liberal than the judge, each judge is less likely to vote liberally. Thus, it appears that with greater ideological divergence on the court, the judges are more likely to vote their ideological preference. A number of issue areas are significant, suggesting that judges may be more or less likely to vote liberally depending on the type of case before them.

All remaining factors influence the likelihood of a liberal vote as expected. The judges of the Wisconsin Supreme Court vote in accordance with the position taken by the state of Wisconsin. While the position of the state of Wisconsin was not significant for every member of the court, this factor was in the expected direction for all but one judge. The judges also appear more likely to vote liberally if they are upholding a liberal decision by the intermediate appellate court.

CONCLUSIONS AND IMPLICATIONS

The results of our analysis strongly suggest that, for the most part, contributions by attorneys to a judge of the Wisconsin Supreme Court do not result in a greater likelihood that the vote of that judge will be influenced. This suggests that contributions to state supreme court judicial candidates operate in ways not unlike contributions to congressional candidates. Most research on Congress suggests that, for the most part, contributions serve two primary purposes. First, they buy access once the person is elected (see, for example, Hojnacki and Kimball, 1999). Second, they are designed to put into office the "right" politician for the contributor's interest. In the judiciary, if the contributors are primarily seeking more-conservative judges to favor their positions, then they seem to have succeeded even if their money does not directly translate into votes.

Attorney contributions raise much concern among those calling for reform. Questions about judicial independence are implicated when attorneys contribute to a judicial campaign and then appear before the judge to whom they have contributed. States may not be willing to restrict attorneys from contributing because of federal and state free-speech rights, but states like Wisconsin have implemented other reforms to ensure judicial independence. Given that attorney contributions are the most common of all contributions in Wisconsin and that they occur with some regularity, it appears that Wisconsin has been able to limit somewhat the appearance of inappropriate influence on the behavior of judges. If attorneys are the most-frequent contributors but are generally unsuccessful in persuading judges with these contributions, it appears unlikely that less-frequent contributors would be gaining influence, with judicial independence remaining intact.

Overall, at least in Wisconsin, little conclusive evidence exists to indicate that the judges of the Wisconsin Supreme Court are systematically influenced by contributions to their campaigns. There is, however, some evidence that a few judges may be influenced by contributions under specific conditions. These are exceptions to our conclusion that contributions are not producing votes. One judge (Bradley) appears to be influenced by contributions; her likelihood of voting liberally changes significantly depending on the amount of the contribution or which side has a contribution advantage. Another (Geske) appears to be affected by the time since a contribution. However, a third (Wilcox) appears to be voting in a direction opposite the ideology of those making contributions to his campaign. Liberal contributions, especially when the contribution exceeds the average, make him more likely to vote conservatively; conservative contributions increase the likelihood of a liberal vote. Another judge (Steinmetz) also appears to be affected by contributions; when the liberal side contributes more, Judge Steinmetz is more likely to vote in a liberal direction than would be the case absent any contribution imbalance. These findings provide some limited leverage for the critics who seek reform of judicial elections that rely on private contributions. In addition, they suggest that the effect of campaign contributions on votes is best understood by looking at the individual vote, not the aggregate outcome of the case.

Our findings, combined with the findings as to the influence of contributions in other states, indicate the need for further research comparing states with different attributes to determine whether there are state-specific factors that might explain the differences. It is possible that neither the state culture of Wisconsin, with its tradition of good government, nor the efforts to limit the impact of campaign contributions allow contributions to affect votes across all members of the judiciary. The alternative approach presented in our research could be employed to reexamine the previous research from Alabama, Texas, and Ohio, states with more-contentious judicial elections but without campaign finance reform.

This study is merely a first step in understanding the implications of electing judges when campaign contributions are an essential element of selecting judges. The National Institute for Money in Polities' data on many more states would allow for similar analysis of each of those states individually, as well as analysis across states. The differences in the electoral systems across states would allow us to account for such factors as nonpartisan versus partisan elections, different ideological compositions of the court, and differences in the culture of campaign contributions among attorneys, interest groups, and firms. One could also learn about contribution effects by looking at petitions to the court and determining whether those petitions filed by attorneys, interest groups, or contributing parties are more likely to be accepted (for a first step, see Eakins, 1999).

We believe that we have added a useful step in discussions of the role of contributions on judicial decision making. A look at the votes of individual judges makes it possible to determine if those judges are influenced even where the court as a whole is not. Further, a large subset of cases and the addition of ideological controls across all cases, combined with an extension of the number of terms under review, allows for a broader analysis of the potential influence of contributions on outcomes. Finally, by exploring a less-partisan state like Wisconsin, we have been able to assess whether campaign contributions are a problem even in reformed states, as some critics suggest, or whether culture or financing reforms limit the impact of campaign contributions.

As more data become available, more can be done to determine the impact of campaign contributions in judicial elections on the decisions of the judges as they sit on the bench. Likewise, refinement of measures might assist in future analysis. For example, the reporting requirements in some states call for the occupation of the contributor, allowing us to test whether greater percentages of all contributions coming from attorneys translate into greater likelihood of a vote in favor of the contributing attorney. It would also be helpful to explore the role of PAC contributions on decisions in other states where PAC contributions may make up a larger portion of the total contributions.

In our analysis, the vote of the judge in each individual case is the dependent variable. Thus, what matters is only whether an attorney makes a contribution to Judge A before a case is decided, but if the same attorney did not contribute to Judge B, the vote of Judge B is assumed to be free from the influence of a contribution. We assume that the contribution would, however, continue to affect the behavior of Judge A in all future cases brought by the attorney. Other assumptions, such as that if a judge remembers a contribution by an attorney, is facing an election, and hopes that a favorable decision would result in another contribution at some point in the future, might lead to somewhat different outcomes, and could also be tested by future research.