ELECTRONIC VOTING MACHINES AND CITIZEN PARTICIPATION
IN STATE JUDICIAL ELECTIONS:
TOWARD GREATER JUDICIAL ACCOUNTABILITY?
Stephen M. Nichols
Assistant Professor of Political Science
California State University, San Marcos
ABSTRACT
The role of elections in promoting accountability to popular sentiment is a central tenet of democratic theory. The extent to which such accountability is enhanced in judicial elections in the American states, however, is threatened by low levels of citizen involvement in contests for the state judiciary -- due in part to ballot roll-off: many voters, though already at the polls, simply ignore judicial contests. Critics of elections as a means of selecting state judiciaries contend, then, that high ballot roll-off renders citizen participation too low for such elections to fulfill their intended role of enhancing popular government through greater judicial accountability.
This research -- an aggregate analysis of 35 Kentucky counties voting in that state's 1992 and 1995 Supreme Court elections --suggests that an emerging electoral technology, the electronic voting machine, substantially reduces ballot roll-off in judicial elections, and thus increases the level of judicial electoral participation at least among citizens who turn out to vote. These findings have significant implications for the debate over judicial selection by popular ballot.
____________________________________________________________________________________
INTRODUCTION
As instruments to secure governmental responsibility and accountability to the governed, popular elections occupy a central place in democratic theory. The American judiciary, then, more so than the other branches of government, has long been assailed as elitist and unaccountable to popular sentiment -- a criticism stemming directly from the fact that all federal judges, and many at the state level, are appointed by executives and legislators, not elected by the citizenry.
Indeed, a desire to harness state judiciaries more tightly to the will of the people spurred reformers of the Jacksonian and Progressive eras to push for popular elections, rather than legislative and/or gubernatorial appointments, as a means of selecting members of the judicial branch in the American states. There is little doubt of the reformers' success: election is now the most common method of judicial selection, as judges in twenty-nine states gain office via a vote of the citizenry. There is considerable debate, however, over whether judicial elections have achieved the intended result. Critics contend that American electorate is both uninformed about and uninvolved in state judicial contests -- and that elections, therefore, do little to enhance popular control over state judiciaries.
This study pertains to the second of these two criticisms of judicial elections -- that of low citizen involvement. After a review of previous scholarship on judicial elections, I will present evidence that an emerging electoral technology, the electronic voting machine, substantially increases citizen participation in state judicial elections by reducing ballot roll-off in these contests (roll-off, also known as voter fatigue, is the well-documented decline in the number of votes cast for offices below the "top" contest -- e.g., president, governor, etc. -- on the ballot). I conclude with a discussion of the implications of this research for the controversy surrounding elections as a means of enhancing judicial accountability.
PRIOR STUDIES OF JUDICIAL ELECTIONS AND BALLOT ROLL-OFF
Proponents of judicial elections argue that the vote subjects judges (and judicial policy-making, presumably) to the sort of public accountability consistent with the democratic ideal of popular government. For this to be so, however, the judicial electorate must, minimally, participate in judicial contests and be aware of judicial candidates and issues. But the very nature of judicial campaigns -- in which concerns about their subsequent impartiality on the bench preclude judicial hopefuls from enunciating issue positions and discussing philosophical differences -- works against both criteria. As Dubois (1979b, 867) notes, "Judicial elections rarely feature the colorful candidates, controversial issues, and spirited campaigns which draw voters to the polls to fill the major federal and state offices." Rather, all candidates make essentially the same pledge of impartiality, put forth often-comparable professional resumes, and discuss matters of court administration and reform (Baum 1994, 119; Dubois 1980, 32) . The upshot -- bland, issueless elections -- elicits scant citizen involvement in judicial contests, and leaves those who do go to the polls with little in the way of substantive information by which to differentiate between judicial candidates.
Lacking this substantive knowledge, citizens look elsewhere for judicial voting cues. In states with partisan elections, of course, voters rely heavily on party affiliation (Dubois 1979a; Champagne 1986). Non-partisan judicial races, though -- the state of affairs in eighteen of the twenty-nine judicial election states -- force citizens to rely on other choice heuristics. These include incumbency and the related advantages of name recognition (Baum 1983; Dubois 1980; 1979a), and whatever information may be gleaned from the ballot -- candidate surnames, gender, occupation, ballot position, and the like (Dubois 1984; Nagel 1973; Byrne and Pueschel 1974; Bain and Hecock 1957). Many judicial voters seemingly choose on the basis of no information: a study of Texas trial court elections found that even the winners enjoyed little name or office recognition (Champagne and Thielemann 1991). Similarly, Wyoming voters apparently construed the absence of negative information as sufficient reason for an affirmative vote in judicial retention elections in that state (Griffin and Horan 1983).
This, remember, is the portrait of citizens who do vote in judicial elections; many Americans never make it even that far. Judicial contests -- "the tail on the electoral kite" (Dubois 1979b, 870) -- elicit especially low turnout, particularly when held separately from elections for high-profile partisan offices. Even when judicial races are contested concurrently with major partisan contests, ballot roll-off provides yet another avenue of citizen non-participation: many voters simply fail to register a preference in judicial races, presumably owing to the above-noted lack of substantive information about the candidates and issues.
Analysts suspect that roll-off is a function of the actions of "core" versus "peripheral" voters. Core voters, more politically attentive and active than others, vote in most elections and in most races on the ballot; those on the periphery, in contrast, are generally attracted to the polls by only the highest-profile contests. The latter know little about lower-visibility races -- and are thus more likely to forego voting in them, thereby producing roll-off. Scholars have long known of voter fatigue's relationship to various features of the ballot: certain ballot formats (e.g., the party column ballot) produce noticeably less roll-off than other (e.g., the office block ballot) arrangements (Walker 1966; Mather 1964; Asher et al. 1982). Relationships of this sort have been noted in judicial elections specifically (Dubois 1980, 52-59; 1979b; Adamany and Dubois 1976).
Recent work has revisited the topic of roll-off in light of the gradual disappearance of older, manual (lever-pull) voting devices and their subsequent replacement with new electoral technology. Nichols and Strizek (1995) documented a substantial reduction in roll-off in Franklin County (Columbus), Ohio, associated with the use of a popular new voting instrument, the Shouptronic electronic voting machine. The Shouptronic device (along with a competing machine built by Microvote) reduces voter fatigue apparently as a consequence of a particularly salient stimulus designed to assist voters in keeping track of the voting process: both types of machine feature a flashing red light above each contest on the ballot, and the light atop a given race continues to flash until the voter records a choice in that contest. Few voters seem able to ignore this nagging, visual reminder that portions of the voting task remain: Nichols and Strizek documented roll-off reductions of up to twenty percentage points attributable to the use of the new electronic voting machines (1995, 306-12). While their study examined the effects of new machines on roll-off across the ballot, Nichols and Strizek hinted at a pronounced impact in judicial elections -- roll-off declined by roughly twelve and thirteen percentage points in Ohio State Court and lower court races, respectively, with the use of the electronic device (1995, 312).
Clearly, judicial elections are fertile soil for the roll-off effects seen in other electoral contexts. This study offers a more focused look at the relationship between voter fatigue in judicial elections and emerging electoral technology. I do so by examining roll-off in recent Kentucky State Supreme Court elections. Kentucky presents an ideal case study on several counts. First, judicial contests there are non-partisan, thus providing (a) the setting most often criticized for dampening citizen involvement in judicial contests, and (b) the arena most conducive to the impact of electronic voting machines.
Second, Kentucky law requires that all voting devices in the state "permit each voter ...to vote a straight political party ticket by one or more marks or acts" (Kentucky Election Laws 1994, 103) -- i.e., Kentucky citizens can, in one fell swoop, cast a straight party vote. This fact has important implications for the non-partisan contests on the Kentucky ballot: as one county clerk put it, "Many people think they're finished once they pull the party lever, so they leave the voting booth without voting in the non-partisan contests." Judicial contests are, of course, among the non-partisan offices often ignored by voters who use the party vote lever.
Boards of Election in many of Kentucky's counties, when faced with the plethora of new voting machines available to replace the ailing manual devices used in most of the United States for decades, opted for either the Shouptronic or Microvote electronic machines in large part because these devices would remind voters that their election day task entailed more than voting in partisan races. In short, many Kentucky county election officials have replaced their manual voting machines with electronic devices of the Shouptronic or Microvote "flashing red light" variety, with the specific intent of reducing roll-off in judicial and other non-partisan ballot contests.
This leads to the third appealing feature of Kentucky as a case study. For most of the 1980s, virtually all Kentucky counties employed manual, lever-pull voting instruments. By 1992, roughly half of the state's 120 counties had switched to Shouptronic or Microvote machines; by the 1995 general election , the balance was 70-50 in favor of the electronic devices. The use of both old and new voting instruments, then, provides a nice quasi-experimental setting in which to explore roll-off differences between manual and electronic voting machines.
In the ensuing section I will lay out the evidence indicating that the change to electronic voting machines has had the desired effect on voter fatigue in Kentucky state judicial elections. I will test a multivariate model of ballot roll-off in the 1992 and 1995 contests for the KY Supreme Court; this will show that electronic voting machines sharply attenuate judicial roll-off, even after taking into consideration other correlates of voter fatigue.
DATA AND METHOD
The central question addressed here is simple: is judicial ballot roll-off lower in the Kentucky counties that employ electronic voting machines, compared to those counties still using lever-pull devices that lack the flashing reminder of races not yet voted in? And, does this relationship hold after taking into account a number of other potential causes of voter fatigue? To address these questions I have compiled a dataset containing county-level measures of aggregate voter participation in the 1992 and 1995 Kentucky general elections. These data were collected from records obtained at the Kentucky State Board of Elections, and through telephone interviews with the County Clerks of the thirty-five counties included in this analysis. Finally, the dataset includes county-level sociodemographic information taken from the 1990 U.S. Census, furnished by the Kentucky State Data Center.
The dependent variable is ballot roll-off in the thirty-five counties voting in the 1992 (Fifth District) and 1995 (First District) Kentucky State Supreme Court contests. Following others (Walker 1966; Dubois 1979b; 1980; Nichols and Strizek 1995), I have operationalized roll-off by calculating the percentage of votes cast by county in the Supreme Court contest, relative to the number recorded in the "major partisan office" -- i.e., the high-profile contest drawing the greatest number of votes -- on the ballot. In 1992, of course, this was the presidential contest; the corresponding race in 1995 was the Kentucky gubernatorial general election. The result, then, reflects the percentage of a county's citizens who voted in the contest at the top of the ballot but did not record a preference in the election for State Supreme Court justice.
The primary hypothesis tested here is that electronic machines will lessen roll-off, and thus the main independent variable of interest is a dichotomous measure of whether a given county was one of the twenty-six voting in this election to use a new Shouptronic or Microvote electronic voting machine (coded as one), or one of the nine to use an older variety of lever-pull machine (scored as zero).
A stringent test of the impact of voting machines on roll-off, however, requires consideration of possible correlates of voter fatigue. Previous studies point to the following:
Education. Walker (1966) found a strong link between education and roll-off: citizens with less formal education are presumably less informed about politics, and are thus less likely to record choices in low-salience ballot contests. Education is measured here as the percentage of county residents with at least a high school diploma.
Income. Affluent citizens are, on the whole, more engaged in the political process (Verba and Nie 1972), and perhaps their greater participation extends to lower-ballot races. I have operationalized income as a county's median household income.
Age. Young citizens lag behind others in their participatory proclivities (see Rosenstone and Hansen 1993, 136-140, for a review of the evidence on this point), and this lack of involvement may also be reflected in greater ballot roll-off. This possibility is captured in the model with a measure of the percentage of county residents between the ages of eighteen and twenty-four.
Race. Non-whites participate less in American elections (Verba and Nie 1972) and evince higher levels of voter fatigue (Nichols and Strizek 1995) than their white counterparts. Race is captured here as the percentage of African American residents in a county.
Residential mobility. Those new to an area may be less in tune with state politics, and therefore less inclined to record a preference in a state-specific contest (Nichols and Strizek 1995). Residential mobility is represented in this analysis as the proportion of county residents who moved into Kentucky from outside the state in the past five years.
Finally, the model contains a dummy variable representing whether a county's Supreme Court roll-off is measured in the context of a presidential versus gubernatorial election (the presidential contest is coded as one, the gubernatorial race as zero). Inclusion of this term is necessary to account for probable differences in the nature of the electorate produced by the two types of elections. In keeping with Campbell's (1962) notion of core and peripheral voters, roll-off was probably greater during the 1992 presidential election simply because a presidential contest -- having the highest profile of any American election -- likely draws more peripheral voters than any other race.
Here, then, is the model of judicial ballot roll-off used in the ensuing multivariate estimation:
Roll-off = a + b1 (voting machine dummy) + b2 (education) + b3 (income) +
b4 (age) + b5 (race) + b6 (residential mobility) + b7 (presidential election dummy) + e.
RESULTS
A preliminary glimpse at the data (Table 1) provides strong support for the hypothesized difference in judicial roll-off in Kentucky counties using new versus old voting instruments. Indeed, voter fatigue in electronic machine areas was, on average, fully twenty-six percentage points below that in manual voting machines counties. The range of values producing this mean difference is equally revealing: the highest roll-off figure recorded in any county using an electronic voting device (61%) was only slightly above the average figure for counties voting on manual instruments (60%). Note that roll-off levels and machine differences of the magnitude seen here are consistent with findings in recent work (Nichols and Strizek 1995, 308).
TABLE 1Ballot Roll-Off in Kentucky State Supreme Court Elections
by Voting Machine Used in County
Voting Machine Used in County Roll-Off Difference
(Manual - Electronic)
Manual (N=9) Electronic (N=26)
Mean Roll-Off 60% 34% 26%*
Range
High 85% 61%
Low 31% 1%
*Difference is statistically significant at .01 (one-tailed test).
The bivariate picture, then, is as expected: judicial voter fatigue seems sharply reduced in counties employing the new electoral technology. The remaining task is to ascertain whether the apparent impact of electronic voting machines on judicial roll-off withstands simultaneous consideration of the aforementioned other contributors to voter fatigue. This requires a multivariate assessment of the relationship between roll-off, voting machines, and the several control variables. The evidence relevant to this question, obtained through a multivariate (ordinary least squares) regression analysis, is presented in Table 2.
TABLE 2Effects of Voting Machine and Sociodemographic Factors on Kentucky State Supreme Court Roll-Off
Independent Variable B Beta
Voting machine dummy -.20 (.03) *** -.41
Education -1.55 (.33)** -.55
Income -.0002 (.0001)* -.38
Age 2.07 (.60)** .32
Race .39 (.31) .10
Residential mobility -.63 (.43) .16
Presidential election .18 (.05)** .39 dummy
Constant .82 (.15)***
N=35
R2=.63 ___________________________________________________
* = significant at .10 ** = significant at .05 *** = significant at .01 (one-tailed tests)
Table entries are unstandardized (B) and standardized (Beta) OLS regression coefficients; standard errors are shown in parentheses.
Here, too, is substantial support for the hypothesized impact of new voting technology on voter fatigue in Kentucky's Supreme Court contest: the presence of an electronic voting instrument lowers judicial ballot roll-off by roughly twenty percentage points, even after taking into account other forces affecting the dependent phenomenon.
The overall fit of the model is quite good, accounting for nearly two-thirds of the variation in Supreme Court roll-off. The control variables performed largely as expected, with the exceptions of race and residential mobility (the effects of which were not statistically discernible). Counties with larger concentrations of young residents evinced correspondingly higher levels of roll-off; education and income had the expected attenuating impact on voter fatigue. As anticipated, too, roll-off differed according to the type of election, with the presidential contest producing greater voter fatigue than the gubernatorial race.
These other factors are present in the analysis merely as controls, however: the central point to be culled from these results is that even with allowances for the impact of other forces, counties voting with the Shouptronic or Microvote electronic voting machines experienced substantially less ballot roll-off in the Supreme Court contest than did counties still employing manual voting instruments. The standardized regression coefficients suggests that the relative effect of voting machine type on roll-off is second, and a close second, to that of education only. To further underscore the magnitude of the electronic voting device's impact, omitting the voting machine variable from the regression analysis produces a sizable reduction in the model's ability to explain variation in roll-off: the R-squared falls to .49, a decline of fourteen percentage points (analysis not shown).
DISCUSSION
These findings lend considerable weight to the premise that voters using recent electoral technology will be unlikely to pass over judicial (and other) lower-ballot contests. Returning to the themes that framed this investigation, these results should now be placed back into the context of the debate over elections as a means of enhancing judicial accountability.
Clearly, new voting devices of the sort studied here increase the level of citizen participation in judicial elections. Electoral technology will not likely draw voters to the polls in the first place, of course -- but once there, electronic voting instruments make them less likely to ignore judicial races. This research has not established a direct link between lessened ballot roll-off and heightened judicial accountability; nonethless, if elections are to fulfill their intended function of increasing judicial accountability, increased citizen involvement in those elections -- which this research does link to electronic voting machines -- is obviously a necessary (though not sufficient) first step in that direction. These machines, then, seem a potent rebuttal for one aspect of the two-pronged critique of the value of judicial elections.
But what of the remaining criticism, that voters who do record a preference in races for the judiciary are guided by little or no meaningful information? Flashing lights may prompt a vote -- but what should one make of preferences ostensibly registered to extinguish an electronic annoyance, by persons who, presumably, would otherwise have ignored the race? The pessimistic conclusion is that because the machine provides these reluctant voters with no meaningful information, the additional votes generated in judicial elections by the new devices are therefore even more devoid of substance than is normally the case in such contests.
A somewhat more optimistic conclusion is possible, however. Once prompted to vote in a contest, it seems doubtful that a citizen then randomly selects a candidate; more plausibly, the voter likely searches for some cue to guide the choice (see, among many others, Popkin 1994 for a review of the literature on this point). These cues may be the shallow ones discussed earlier; alternatively, some contend that citizens voting in judicial elections make reasonably good use of the relevant, available information -- the problem being that, as noted above, judicial contests by nature offer little such knowledge (Hojnacki and Baum 1992). This, though, may be starting to change: scholars now note the gradual emergence of a "new style" of judicial campaign, wherein candidates are more willing to enunciate issue positions and stake out philosophical differences between themselves and their competitors (Baum 1994, 119). This trend, then, may provide judicial voters with more substantively helpful information to assist them in selecting judges for the state bench.
The link between that possibility and this study, of course, is that the extra votes stimulated by electronic voting machines are perhaps not be as meaningless as they might appear. It is an obvious leap from that point to a conclusion that the judicial electorate is sufficiently involved and informed such that elections now fulfill their intended role in securing judicial accountability. Nonetheless, the evidence presented here underscores the effectiveness of new electoral technology in stimulating citizen participation in judicial elections, and other signs suggests that the quality of information guiding such choices may be improving. At the very least, these findings provide new fodder for defenders of elections as a means of judicial selection.
REFERENCES
Adamany, David, and Philip L. Dubois. 1976. "Electing State Judges." Wisconsin Law Review No.3:731-79.
Asher, Herbert B., Russell Schussler, and Peg Rosenfield. 1982. "The Effect of Voting Systems on Voter Participation." Paper presented at the annual meeting of the Midwest Political Science Association, April 28-May 1, Milwaukee.
Bain, Henry M. Jr., and Donald S. Hecock. 1957. Ballot Position and Voter's Choice: The Arrangement of Names on the Ballot and Its Effect on the Voter. Westport, CT: Greenwood Press.
Baum, Lawrence. 1994. American Courts. Boston, MA: Houghton Mifflin.
Byrne, Gary C., and J. Kristian Pueschel. 1974. "But Who Should I Vote For for County Coroner?" Journal of Politics 36:778-84.
Campbell, Angus. 1966. "Surge and Decline: A Study of Electoral Change." In Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes (eds.), Elections and the Political Order. New York: Wiley and Sons.
Champagne, Anthony. 1986. "The Selection and Retention of Judges in Texas." Southwest Law Journal 53:95-99.
Champagne, Anthony, and Greg Thielemann. 1991. "Awareness of Trial Court Judges."
Judicature 74:271-76.
Dubois, Philip L. 1979a. "The Significance of Voting Cues in State Supreme Court Elections." Law and Society Review 13:757-79.
Dubois, Philip L. 1979b. "Voter Turnout in State Judicial Elections: An Analysis of the Tail on the Electoral Kite." Journal of Politics 41:865-87.
Dubois, Philip L. 1980. From Ballot to Bench: Judicial Elections and the Quest for Accountability. Austin, TX: University of Texas Press.
Dubois, Philip L. 1984. "Voting Cues in Nonpartisan Trial Court Elections: A Multivariate Assessment." Law and Society Review 18:395-436.
Griffin, Kenyon N., and Michael J. Horan. 1983. "Patterns of Voting Behavior in Judicial Retention Elections for Supreme Court Justices in Wyoming." Judicature 67:68-77.
Hojnacki, Marie, and Lawrence Baum. 1992. "Choosing Judicial Candidates: How Voters Explain Their Decisions." Judicature 75:300-09.
Kentucky Election Laws. 1994 Edition. Frankfort, KY: Kentucky State Board of Elections and Kentucky Registry of Election Finance.
Mather, George B. 1964. Effects of the Use of Voting Machines on Total Votes Cast: Iowa -- 1920-1960. Iowa City: University of Iowa, Institute of Public Affairs.
Nagel, Stuart. 1973. Comparing Elected and Appointed Judicial Systems. Beverly Hills: Sage Professional Papers in American Politics.
Nichols, Stephen M., and Gregory A. Strizek. 1995. "Electronic Voting Machines and Ballot Roll-Off." American Politics Quarterly 23:300-18.
Popkin, Samuel L. 1994. The Reasoning Voter. Second edition. Chicago: University of Chicago Press.
Rosenstone, Steven J., and Mark Hansen. 1993. Mobilization, Participation, and Democracy in America. New York: Macmillan Publishing Company.
Verba, Sidney, and Norman H. Nie. 1972. Participation In America: Political Democracy and Social Equality. New York: Harper and Row.
Walker, Jack L. 1966. "Ballot Forms and Voter Fatigue: An Analysis of the Office Block and Party Column Ballots." Midwest Journal of Political Science 10:448-63.
NOTES
1. This, of course, is a highly simplified summary of the waves of reform that produced the varied means of judicial selection employed in the United States. Note, too, that while doubt about their ability to ensure judicial accountability is the most fundamental concern about judicial elections, there are others (pertaining to the impartiality and general quality of judges chosen via elections). See the first chapter of Dubois 1980 for a thorough discussion of both points. 2. See Dubois 1980, especially pp. 43-46, for a review of turnout in judicial elections. 3. The core/periphery distinction was introduced by Angus Campbell (1962) as part of his "surge and decline" explanation of turnout and results in mid-term versus presidential elections. 4. This quote is from a telephone interview with Donald W. Blevins, Clerk of Fayette County, KY. 5. The Federal Election Commission lists more than twenty-five different manufacturers of electronic, computerized voting machines. 6. All data and documentation necessary to replicate this analysis are available upon request from the author. 7. The data were collected from June 5-12, 1996; telephone interviews with the County Clerks were conducted between June 17 and 28, 1996. The Kentucky State Data Center, located at the University of Louisville, is a repository for Kentucky census information. 8. There is no other sizable non-white population in the Kentucky counties included in this analysis.