Recent United States Supreme Court decisions have brought nearly to an end the era of judicially-supervised school desegregation and integration. Over a series of decisions, the Supreme Court has sought recently to disengage the federal judiciary from close oversight of local school districts, especially in the realm of racial segregation.^\1\ But the judiciary?s involvement in educational matters is far from negligible. Undaunted by the federal judiciary?s experience with school desegregation—and perhaps even inspired by it—state courts have over the past twenty years embarked on their own efforts to effect dramatic changes in public education. These efforts have focused not on racial segregation but on the financing disparities among school districts.^\2\ Since the U.S. Supreme Court ruled in San Antonio Independent School District v. Rodriguez (1973) that school financing inequities do not violate the equal protection clause of the Fourteenth Amendment, state supreme courts in 27 states have ruled on school financing suits under provisions of state constitutions. Twelve have ruled in favor of greater equity and fifteen have ruled against it (see table 1). Part of a larger trend of using state courts for the protection of civil rights under state constitutions,^\3\ these school finance decisions have the potential dramatically to alter the fiscal policies of numerous state governments—with enormous consequences for both the amount of resources allocated to public education and the equity of that distribution. Legal scholars have given these state supreme court decisions fairly wide notice,^\4\ but little attention has been paid to the impacts of these decisions. The political science and policy communities have also given scant attention to the impact of these decisions. One exception is Michael Mintrom's report in 1993. Another policy study, Hickrod et al. (1992), directly assesses the effects of state supreme court decisions across a number of states. Unfortunately, this study is marred by some significant methodological problems. First, the school finance figures are not adjusted for inflation; only constant figures are used. Also, as the article compares data over a fairly long time span, 1970?1990, inflation could account for much of the increase in educational expenditures by state and local government. Second, measures of school financing equity are used that precede state supreme court decisions in Kentucky, Montana, Texas, and New Jersey. These data, then, cannot be used to evaluate whether the courts changed the distribution, and hence the equity, of funds.

This paper attempts to rectify this imbalance in both the political science and policy literature on school finance. I do so by examining two ways these decisions affect the policies and politics of state governments. First, I want to examine the effects of these state supreme court decisions on the actual distributions of school funds within four states. In short, this paper assesses the success of efforts by four state supreme courts (Connecticut, New Jersey, Texas, and Kentucky) to increase the equity of school finance within their states. Second, I also want to explore the dimensions of public reaction to these decisions, by analyzing polling results and voting returns. This examination of public reactions to the court decisions and the legislative remedies designed to comply with judicial mandates will, I hope, highlight the opportunities, limitations, and constraints that operate on state supreme courts as they strive to effect significant changes in the ways public schools are financed.

My argument is twofold: First, I argue that there has been important variation in the changes wrought by state supreme courts in the four states I study. Some state supreme courts have achieved a great deal of equalization, while others have been less successful. Second, I argue that public opposition to equalization efforts is often keen, but its determinants are not straightforward. Economists would contend, and have,^\5\ that the opposition to equalization stems from economic self-interest to avoid costs. But my analysis below of polling data from New Jersey and election returns in Texas shows that a significant portion of the opposition in both those states is racial or ideological rather than simply driven by perceived economic self-interest.

This conclusion, I contend, has profound consequences for how both judges and legislators approach the issue of school finance equalization. In short, devising a school finance system that distributes economic costs widely and fairly may not be enough; judges and policy makers will still have to overcome racial and ideological cleavages. The former results from the perception that racial minorities are the beneficiaries of equalization. The latter emerges from those who value localist rather than state-centered approaches to school financing—even if the localist approach generates persistent and significant inequality of resources.

Rather than simply confronting interest groups that articulate economic, class-based arguments against equalization, legislatures (and, in turn, state courts) must battle against a mass public opposition to equalization that is, in significant ways, racially based. This paper examines the quantitative effects of court-ordered school finance equalization in four states, and then turns to an analysis of public opinion concerning school finance equalization in two states where supreme courts have been particularly active: Texas and New Jersey. The New Jersey Supreme Court has achieved significant and important reforms in school finance for that state, but at a fairly high political cost. In Texas, meanwhile, the results have been less favorable for the advocates of increased equity, and opposition has been extremely strong. In both cases, I examine the public opinion surrounding the legislature?s policy response to the school finance decisions.

This paper is organized into three sections. The first section presents a quantitative assessment of the effects state supreme courts in Connecticut, New Jersey, Texas, and Kentucky have had on equality of school finance expenditures. The second section shifts the focus to New Jersey and Texas and examines the determinants of public opposition to the legislative responses to the Supreme Court decisions. Finally, the third section concludes with some observations about the limitations of both judges and legislators as they strive to address the problem of school finance inequities.

How Have State Supreme Courts Affected Educational Financing Equity?

In this section, I provide an overview of the trends in school finance equity in the wake of state supreme court decisions that declared existing methods of financing schools unconstitutional.

At this juncture, it is necessary to provide a quick word about school finance data and the notion of expressing "equality" through quantitative data. There are a number of ways to measure the equity within a school finance system and they all embody certain value choices about what is worthy of measurement. Put simply, different measures reflect different normative commitments. One cannot provide an "objective" notion of equality because there are different types of equality.^\6\ For the sake of simplicity, I have chosen to provide here only one basic measure: the coefficient of variation, which is the standard deviation of a population divided by the mean of the population. This calculation measures the dispersion of expenditures across districts within a state. It is a quantitative representation of what Berne and Stiefel (1984) call "horizontal equality," the notion of providing all similarly situated students with equal amounts of educational resources, measured here by dollars.

Quantitative Analysis

I obtained school financing data for each school district within each state, and calculated the per pupil expenditures in a number of categories. These data generally came from the state departments of education, although Connecticut data was obtained from the Connecticut Public Expenditure Council, a well-regarded fiscal watchdog group. Also, because the New Jersey decision was restricted to only particular socio-economic classes of school districts, I have applied my analysis only to those districts that are the focus of the court's ruling: the 30 so-called "special needs" districts, largely inner-city districts, and the roughly 110 affluent districts that fall into the New Jersey Department of Education's "I" and "J" categories of district wealth (the two most affluent categories). I weighted each district for the number of students within that district in order to obtain a per pupil rather than a per district analysis. After adjusting the figure for inflation, I then calculated the coefficient of variation (dividing the standard deviation by the mean).⁷

Figures 1-4 present the results of the equalization efforts in four states: Connecticut, New Jersey, Texas, and Kentucky. In each figure, we see the trends of school finance equity over time, as measured by the coefficient of variation. The line in each chart shows the equity change in the expenditures or revenues of districts in each state. In order to determine whether the state supreme court decisions had an effect on the equity of these expenditures, we need to determine whether the line slopes downward. In all four states, the lines slope downward after the state supreme court decision.^\8\ But the size of the changes and the permanence of the equalizing trend vary significantly from state to state. In figure 1, Connecticut only saw a slight dip in its overall inequities and then a gradual worsening of the inequalities. Within six years, inequities were actually worse than they were at the time of the court?s decision. Similarly, in figure 2, Texas saw only a gradual and modest decline in the level of school financing inequities—despite the Texas Supreme Court?s deep and repeated involvement in the matter. In contrast, in figure 3, New Jersey saw the equity of funds available to "special needs" and affluent, suburban districts increase rather dramatically since the 1990 decision.\9\ Finally, we see in figure 4 that Kentucky saw its inequities cut almost in half over a four year period. Kentucky?s improvement in equities are clearly the most substantial of the four states examined here.

Implications of Quantitative Findings

We see, then, that school finance decisions in some states ultimately produce much greater equity than they do in other states. What accounts for this difference? What contexts render some decisions more effective than others? Or, to put it in a language that is more fashionable in political science: What are the limitations on state supreme courts? judicial capacities—at least within the policy arena of school finance?

In order to respond to these questions, we need to understand the pressures operating on state supreme courts and the institutional contexts within which they must act. In their work State Supreme Courts in State and Nation, Tarr and Porter present a broad analytical framework for the study of state supreme courts, a framework which sketches the institutional and jurisprudential opportunities and limitations of state supreme courts. In table 2 I have adapted their framework to the concrete legal and policy setting of school finance litigation. Although Tarr and Porter do not present their conception of judicial federalism in a tabular form, their analytical understanding of the vertical, horizontal, and intra-state linkages of state supreme courts, combined with the legal and extra-legal contexts of state courts within the American policy, lends itself to the following two by three matrix. Within each cell, I provide only one example of numerous possible relationships or activities that affect a state supreme courts? foray into school finance reform. Table 2 is not designed to be a comprehensive listing of all possible state supreme court relations and contexts, but rather an analytical schema to help us identify various roles and contexts in which a state supreme court might operate within the policy realm of public school finance.

In column A, we can trace the setting and influence of U.S. Supreme Court and other state supreme court decisions on an individual state supreme court?s decision. In Column B, we see the effects of extra-legal relations and contexts on school finance reform. Not all cells are of equal importance, but a full account of the Connecticut, Texas, New Jersey, and Kentucky case studies would closely examine the interaction between the last two cells in Columns A and B. A full account would, that is, demonstrate the effects that both legal and extra-legal contexts have on school finance reform within each state. It is here that we can begin to identify the sources of relative judicial capacity or incapacity in the realm of public school finance. And it is here that perhaps we can find some answers to the question of why some state supreme court decisions yield dramatic results and others meager results.

Unfortunately, a full reckoning of these case studies is beyond the scope of this paper. Instead, I turn next to merely one of the extra-legal factors that arguably has an influence on state supreme courts? capacity to restructure public school finance: mass public attitudes. My account here focuses on Texas and New Jersey. This examination can provide some insight on how courts can and cannot thwart majority will—especially in a climate where legislatures are particularly attuned to the attitudes of their constituents.

Determinants of Mass Public Opposition to School Finance Reform in Texas and New Jersey

In this section, I examine public attitudes towards school finance reform expressed in New Jersey and Texas at the time of major reforms in their public school finance systems. In New Jersey, I analyze data from a public opinion poll conducted by the Eagleton Institute of Politics in July of 1990, a few weeks after the legislature passed the Quality Education Act (QEA) of 1990, a reform package enacted in response to the New Jersey Supreme Court?s decision in Abbott II. In Texas, I analyze voting returns and demographic data from state legislative districts to discern the demographic characteristics most strongly associated with opposition to Proposition One, a 1993 statewide constitutional referendum on school finance reform. This ballot measure was the consequence of three state supreme court decisions striking down the existing school finance arrangements in Texas. Both analyses show that race—in limited contexts—is salient to the issue of court-ordered school finance reform.

Theories of Opposition to School Finance Reforms

This analysis of public opinion posits three possible explanations for public opposition to school finance reform: economic self-interest, an anti-tax ideology, and racial geography. The three are explicated below.

1. Economic Self-Interest: The first, and probably the most commonly accepted understanding of opposition to school finance reform is simply one of economic self-interest.^\10\ Under this theory, opposition to reforms would emerge from the potential costs that court-ordered school finance equalization might bring: increased taxes and/or lower state aid to a district. Typically, individuals who live in affluent suburbs receive the shorter end of two sticks in school finance reform: their school districts receive less state aid and they have to pay more in state taxes, and possibly increased local taxes as well. Because of the zero-sum nature of school finance equalization, economic self-interest would dictate that the financial losers in school finance reform would oppose the reform effort.

2. Anti-Tax and/or Anti-Government Ideology: A second model that might explain opposition to the court-ordered reforms is an ideological one. Similar to, but distinct from an economic self-interest model, an anti-tax sentiment model could account for much of the opposition—independent of whether one?s own district gains or loses aid or independent of whether one?s own tax bill goes up or down. This could be particularly true if the sentiment is conjoint with an overall conservative ideological bent.

3. Racial Geography: A third model I test is a racial geography model: the effects of race and geographic location on one?s likelihood to support or oppose both the court-ordered school finance reform. The rationale of this model is that anti-urban sentiment in the suburbs and rural areas combines with the perception that non-whites are the sole beneficiaries of school finance equalization to yield a white/non-white and urban/suburban-rural cleavage over the issue of school finance equalization. In some ways, this is a test of the racial politics of entitlements described by Edsall and Edsall (1992), but on a local rather than national level.^\11\

New Jersey Public Opinion and the Quality Education Act of 1990

The citizens of New Jersey in 1990 were of two minds concerning the equity of school financing: the principle of greater equity was largely endorsed by a majority of respondents, but the specific policies designed to achieve that equity were simultaneously opposed by respondents. The purpose of this section is to analyze in detail the dimensions of New Jersey public opinion about the QEA. In early June 1990, The Star Ledger/Eagleton Poll queried 800 residents of New Jersey about their support for the school financing plan enacted by the New Jersey legislature a few weeks earlier. The data are in an SPSS portable file which was analyzed using SPSS/PC+. It contains 800 observations and 157 variables, encompassing not only the usual socio-economic demographics but also responses to questions about the perceived impact of new taxes, the perceived effects of school finance reform on local districts, whether the respondent has school-age children, etc. The Eagleton Poll weights all observations to improve sample selection, ensuring that age and education frequencies correspond to U.S. Census data for New Jersey, by using an iterative raking algorithm. I collapsed the response I use here as a dependent variable: An approval/disapproval question concerning the QEA. I also collapsed a number of independent dummy variables into dichotomous approval/disapproval or polytomous responses. I then performed a logit analysis of dichotomous approval/disapproval responses to the QEA.

By testing three theories of opposition to the QEA and controlling for the influence that the presence of school-age children have on respondents? answers, I conclude that whites and non-whites in New Jersey perceive school financing differently if they have school age children. But among people without children, race does not shape one?s perception of school financing; instead, economic costs are more salient to one?s support, or lack thereof, of school finance reform.

On a descriptive level, it is clear that in 1990 there was significant support for greater equity in school financing and an equally significant lack of support for the Quality Education Act (QEA I). The Star-Ledger/Eagleton Poll—taken between July 2 and 10, 1990—showed that 54 percent of those who had heard of Abbott v. Burke agreed with the decision (either mildly or strongly) and 38 percent of those aware disagreed with the court. (again, either mildly or strongly). The remaining 8 percent did not know their position. In contrast, only 35 percent of those surveyed approved of the recently passed QEA. Fifty-six percent disapproved, and 9 percent indicated they didn?t know. But what accounts for this level of support—or lack of it? By using a logistic regression technique on the original Star-Ledger/Eagleton Poll data set, we can estimate the influence of a number of independent variables on the inclination of a respondent to favor the QEA.

Operationalizing the Models for New Jersey Data

1. Economic Self-Interest

   The poll contains a number of questions directly related to the perceived economic impact of both the school finance reform package and the income and sales taxes levied in part to pay for it. Specifically, respondents were asked whether they thought the tax package would hurt, have no effect or help "people like you" (EFFECT); whether they thought property taxes would go up, stay the same, or go down (PROPTAX), whether they thought their local school district would lose aid or get aid (GETAID) and their income level, broken into four categories (INCOME2). I recoded EFFECT into a dichotomous (hurt vs. help/no difference) variable (EFFECT-R). Together, these four variables (EFFECT-R, PROPTAX, GETAID, INCOME2) comprise the economic rationality model.

2. Anti-Tax & Ideology

   The poll also contains data on whether respondents accept an increase and expansion of the sales tax (SALESTAX), whether they accept an increase and expansion of the state income tax (INCTAX), and a ideological self-identification score, using the terms conservative, moderate and liberal (IDEOLOG). These three variables comprise the anti-tax and ideological model.

Racial Geography

Using a racial self-identification variable, I collapse all non-Caucasian values into a "non-white" value, yielding a dichotomous white/non-white variable (RACE). For the geographic residence of the respondent, I employ the Eagleton?s classification of municipalities into four varieties: center city, city and old suburb, new suburb, and rural, yielding a polytomous variable (TYPE). These two variables comprise the racial geography model.

Attitudes Towards the Quality Education Act of 1990

1. Attitudes of All Respondents

   Table 3 shows the results of a logistic regression on the approval/disapproval responses using the three models described above, for all respondents.^\13\ This table shows that economic rationality heavily influenced attitudes towards the QEA. Only two independent variables—perceived loss or gain of aid to one?s district and the perceived effect of the QEA on one?s local property tax rates—are significant above 0.05; anti-tax sentiment and ideology do not appear to be significant.

   But what is the magnitude of these relationships? In order to discern this, we need to look at the right hand side of table 3. A logistic regression requires a further interpretative step because the parameter estimate B is not equivalent to a regression coefficient. It represents, instead, the change in the log of the odds ratio of approving or disapproving the QEA, given a per unit increase of each particular independent variable. From this measure, however, we can calculate the probability that an individual will approve or disapprove of QEA, holding the values of the other independent variables constant at their means. This calculation for each variable is shown in the "Change in Probability" column of table 3.^\14\

   Upon examination, the change in probability column of table 3 shows that the magnitude of the effect of these two significant variables is rather large: a perception that one?s property taxes will rise results in a 25 point drop in support for the QEA.^\15\ Similarly, a perception that the local school district will receive reduced state aid lowers one?s approval rating by 15 points. Thus, we can say with reasonable assurance that economic rationality appears to most heavily influence the attitudes towards the QEA among all respondents.

2. Attitudes of Parents of Public School Children

   It is important to be aware that all sub-groups might not regard the QEA with the same attitudes as the sample as a whole. Parents of school-age children, for example, might view the matter differently than non-parents or retired New Jerseyans. The section asks whether parents of school-age children make a calculation of economic self-interest when they evaluate the desirability of the QEA. The data shown in table 4 indicates that race and the type of municipality the respondent lives in are far more reliable predictors of support or opposition to the QEA than is economic rationality. Among parents of children enrolled in public schools, the race and the municipality of the respondent are the only statistically significant variables—even when they are controlled for income, perceived loss or gain of state aid, ideology, and the perceived effect on property taxes, among other factors. The result is that a white parent of a child enrolled in public schools is 31 points less likely to approve of the QEA than a non-white, when all other independent values are held constant at their means. Ideology also has a significant influence on parental attitudes toward the QEA, but it is less pronounced than either race or geographic location of the city. As one moves from liberal, to moderate, to conservative (from 1 to 3), the likelihood of supporting the QEA drops 19 percentage points. Thus, although ideology meaningfully influences parental attitudes toward the QEA, race is clearly a more influential factor. And, surprisingly, economic self-interest plays virtually no role at all. Among parents of school age children, economic concerns do not divide them, but ideology and geography do.

   It is of interest to note that the municipality of respondent runs counter to the hypothesized trend: The probability difference between a respondent who lives in the inner city and one who lives in the suburbs is 23 percentage points (again, all other values held constant at their means.) But the direction is positive. That is, from this analysis one could conclude that suburbanites are more in favor of the program than inner city residents. But we need to examine these patterns more carefully to fully understand the relationship. Table 5 is a cross-tabulation of the approval and disapproval rates across municipality types for white parents of children enrolled in public schools. Table 5 also shows the same for non-white parents.

   Whites are almost uniformly opposed to the QEA—at weighted rates ranging from 94.8 percent opposed in the inner city to about 71 percent opposed in new suburbs. Non-whites, in contrast, show a more varied response. Non-whites in the inner city favor the QEA by weighted rates of about 65 percent to 35 percent. Non-whites in the older suburbs, in contrast, oppose the QEA by rates similar to whites, 78 percent to 22 percent. But non-whites in the newer suburbs favor the law at rates of about 68 percent to 32 percent. The consistency of white opposition and the variability of black support across municipality types renders the relationship between municipality type and support for the QEA non-linear. This non-linearity produces misleading results because logistic regression assumes linear relationships. Thus, if we were to interpret the logistic regression equation alone, we would come to a somewhat erroneous conclusion that living in the newer suburbs would lead to greater support for the QEA. Perhaps a better way to explain the relationship is to say that whites in the newer suburbs are less opposed than whites in the inner city.

   Two elements of this analysis merit further discussion: 1) intense white parents' opposition to the QEA in the inner city; and 2) non-white parents' opposition in the older suburbs. Whites whose children attend public schools and who live in the inner city oppose the QEA by a ratio of about nine to one. This finding is remarkable because the QEA was designed to improve inner city education. At least three possible explanations exist for this counter-intuitive finding. First, whites may feel disenfranchised in cities with large minority populations and feel that additional funds will aid minority children rather than white children. Second, whites may feel that the money would be wasted in the inner city schools, despite the fact that their children would receive at least some benefit. Third, whites in the inner city may simply be racists, opposing a program that will benefit them because it will also benefit minorities. Whatever the explanation, the pronounced racial division within a group most likely to directly benefit from the QEA—inner-city parents of children enrolled in public schools—combined with the economic irrationality of white inner-city opposition leads me to conclude that race was an implicit factor in opposing the QEA for some important segments of New Jersey's population.

   A second finding requires further explanation: the opposition of non-whites in older suburbs, among parents of children enrolled in public schools. Here, whites and non-whites express similar opposition to the QEA. But non-whites in both newer suburbs and in central cities largely endorse the plan. Why do non-whites in older suburbs view the matter differently? One possible explanation may be that non-whites moved to these older suburbs—most likely from the central cities—because the educational opportunities were greater for their children there. They may feel, as a result, that the inner city schools are not worth the money, having had a direct experience with them.

   Whatever the emphasis we place on the matter, this finding indicates that racial cleavages are not uniform—simply bifurcating suburb from inner city, rich from poor. Rather, race works multivalently in New Jersey educational politics, at times salient, at times not.

   This quantitative analysis of public opinion towards the Quality Education Act of 1990 is illuminating for a number of reasons. First, we can say that for the population at large, race has little to do with support for the QEA. Instead, pocketbook considerations of how the program will affect individual taxes and how it will affect the aid to one?s local district largely determine attitudes of the population at large. In contrast, the attitudes of parents of children enrolled in public schools towards the QEA are less influenced by economic self-interest concerns, but more influenced by their race. Indeed, race is the strongest determinant of their parents support for or opposition to the QEA. But that racial cleavage is somewhat fluid—intersecting with geography and class in ways that sometimes align minorities and whites but usually divide them.

Texas Public Opinion, School Finance, and Proposition One

In Texas, the concerns of race and class are remarkably similar. After the Texas Supreme Court ruled the existing financing system unconstitutional in 1989, the legislature adopted a modest reform. This program, too, was struck down by the Supreme Court. The legislature then passed a constitutional amendment that would allow the state to recapture local property taxes. This reform required, however, majority approval at the polls. Most observers thought securing a victory for the amendment (dubbed by opponents the "Robin Hood" plan) would be a difficult, but not impossible task. A poll conducted by the University of Houston Center for Public Policy a month before the May 1, 1993 referendum showed that 37 percent of survey respondents opposed Proposition One, 29 percent supported it, and a whopping 34 percent were undecided.\16\ Another poll, conducted by Mason-Dixon Political-Media Research, Inc. for the El Paso Times showed that 53 percent favored the amendment and only 27 percent were unopposed. The remaining 20 percent were undecided.^\17\

The polls, however, were wrong. Proposition One suffered a huge defeat, losing 63 percent to 27 percent. Proposition One detractors contended before the election that the opposition was largely concerned with increasing taxes and a failing educational system. Indeed, Tom Pauken, leader of the major opposition group Texans Against Robin Hood Taxes, explicitly played on taxation fears: Proposition One "is a back-door tax increase, it has nothing to do with education," Pauken told a Houston Chronicle reporter. As Pauken stated before the election, "If we make this a tax issue, then we win. If Ann Richards is able to make it an education issue, she wins."^\18\

Answering the question of what determined the outcome of the Proposition One election is essential if we are to understand how courts can be effective in the realm of school finance reform. As we try to locate the sources of popular opposition to court-ordered school finance reform, it would be useful to examine the reasons why large numbers of Texans voted against Proposition One. Ideally, we would examine statewide exit polls to determine explicit or implicit reasons voters had for casting their ballots. Unfortunately, no such exit polls exist. As a result, we have no state-wide individual level voter surveys that would enable us to precisely identify the sources of opposition to Proposition One on May 1, 1993.^\19\

What we do have, however, are demographic data and election returns from 150 state representative districts. In this section, I analyze these data to determine some of the demographic characteristics of districts that opposed Proposition One. To be sure, uncovering the demographic characteristics of those regions that voted no on Proposition One is not the same as determining the reasons why people who live in those regions voted no. But in the absence of statewide individual level exit polls, I have no recourse but to rely on demographic data to uncover patterns in the opposition to the school finance equalization referendum. To the extent that such patterns exist, they will serve, for my purposes, as explanations of opposition.

1. Operationalizing the Models in Texas

   Demographic data on each of Texas?s 150 House of Representative districts was obtained from the Texas Legislative Council, which is responsible for providing the state legislature with appropriate information for reapportionment. These breakdowns were, in turn, based on the United States Census Bureau data gathered in the 1990 census. Election returns for each of the precincts in Texas from the May 1, 1993 constitutional amendment referendum and the June 1, 1993 U.S. Senate run-off election between Kay Bailey Hutchinson and Robert Kreuger were also obtained from the Texas Legislative Council and were aggregated into 150 district totals. Ideally, we would want to regress the election returns from each precinct in Texas against the demographic data for that precinct. This fine-grained analysis would come closer to an individual level survey, and would provide several thousand more data points. Unfortunately, demographic data is not available from the Texas Legislative Council at the precinct level. As a result, the election returns were aggregated to state representative districts.

   The models used in the Texas analysis are similar to those used in New Jersey. Like the New Jersey respondents to the Eagleton Institute?s poll, Texas voters may have been influenced by arguments of economic self-interest—perceptions that they would have to pay for greater school equity. Demographic data, alone, however, cannot tell us whether individuals perceive a tax increase as imminent in the wake of the reform. (In New Jersey, this data was included in the polling results.) Instead, we must rely on measures of class—on the theory that higher incomes would be more likely to pay for equalization—and on measures of exposure to property taxes, i.e., rates of home ownership. These two variables—per capita income of the state representative district and percent of housing units that are owner-occupied—constitute the economic self-interest model of opposition to Proposition One. The theory here is that those most likely to pay for the costs of equalization—homeowners and those in higher income brackets—would be opposed. Thus, we would expect districts with high percentages of owner-occupied housing units and with high income levels to vote more heavily against Proposition One.

   The second model is ideology and/or party identification. Here, we would expect liberals and Democrats to more heavily favor school finance equalization. In the Texas voting data, however, we do not have a measure of ideology for each district. There are, however, two sources for the party identification variable: 1) the percentage of votes won by Republican Kay Bailey Hutchinson over Democrat Robert Krueger in the run-off election a month after the Proposition One election or 2) the party of the district representative to the Texas State Assembly. Both are used in the analysis below.

   The third model is one based on racial politics. Here, blacks and Hispanics see themselves (or conversely, whites see racial or ethnic minorities) as the primary beneficiaries of school finance reform and, thus, are more likely to vote in favor of it. (Conversely, whites would be more likely to vote against it.) Thus, we would expect districts with high percentages of blacks and/or Hispanics to vote more in favor of Proposition One. Because Hispanics can be of any race, I have subtracted from the total number of blacks in each district those Hispanics who identified themselves as blacks. The result produces, in effect, the number of non-Hispanic blacks within a district.

   One further note: A variable to control for an urban-rural split (percentage of district residents that reside in an urban area) is included here because of the geographic isolation of blacks in urban areas. Hispanics in Texas live in both rural and urban areas.

2. Texas Findings

   The data was analyzed using SPSS for Windows and an ordinary least squares regression technique. Because reapportionment requires districts to be roughly equal in size, the districts were not weighted for population. The results of the OLS are presented in table 6.

   From table 6 we see that Per Capita Income, Percentage of Non-Hispanic Blacks, and Percentage of Hispanics all have a significant and fairly sizeable relationship to the percentage of no votes on Proposition One, with higher incomes related to a higher percentage of no votes and higher percentages of blacks and Hispanics related to lower percentages of no votes. Of these three, the two racial categories show the strongest contributions to the percentage of no votes in a district. Surprisingly, the party of the district representative (a rough measure of the party leanings of the district) is not a significant predictor of no votes, if we rely on the conventional 0.05 threshold. Also, the percentage of home ownership in a district and the percentage of urban residents in the district do not show a statistically significant relationship to the dependent variable. Overall, the model shows a healthy 0.63 adjusted R, demonstrating a reasonably good fit.

   There are some shortcomings to this analysis. First, the measure of party identification does not capture the difference between those districts where the parties are competitive and those where party identification is more one-sided. In an effort to employ a more nuanced sense of party strength, I also ran the regression omitting the dummy party variable and replacing it with the percentage of votes received by U.S. Senate Candidate Republican Kay Bailey Hutchinson in the special run-off election against Democrat Robert Krueger, held a month after the Proposition One election. A continuous variable, this measure enables me to capture the degree of party strength in a way that is impossible with the dichotomous state representative party identification. Of course, factors other than party strength—such as candidate-specific factors, the economy, etc.—may play a significant role in the level of support that Hutchinson received. In this respect, this measure may overstate Republican support within traditionally Democratic Texas. The results of this second regression are shown in table 7.

   Three significant changes emerge from this change in the party identification variable. First, degree of support for Kay Bailey Hutchinson is a much stronger predictor of opposition to Proposition One than the party identification of the state district representative. Part of this is due, no doubt, to the fact that support for Hutchinson is registered continuously, and thus more reliably tracks opposition to Proposition One than the dichotomous Republican/Democrat distinction of the first party identification variable. Nonetheless, it is clear that support for Hutchinson is a better predictor of opposition to school finance equalization than the party affiliation of the district representative. (For one thing, the R of the entire regression equation improves substantially when we replace the state representative?s party with Hutchinson?s vote percentage.) That fact begs the question, however, of why Hutchinson supporters oppose school finance equalization. To answer this, it may be more instructive to view the percentage of votes Hutchinson received less as a strength of party identification and more as an indicator of the ideological leanings of the district. In this light, the meaning of Hutchinson?s candidacy is that it registers a cluster of conservative ideological values. And from the regression it is clear that those values—whatever their constituent components—have a very high degree of salience to the school finance equalization debate.

   Second, class, as measured by per capita income, no longer has a significant relationship to the question of school finance equalization, when we consider the degree of support within the district for Hutchinson. Although there is some degree of co-linearity between income and support for Hutchinson (the simple r between the two is 0.591), it is clear that ideological/party support for Hutchinson is more important than income in determining opposition to Proposition One. It is not the affluent, per se, who are opposed, but conservatives/Republicans who vote for Hutchinson who are opposed to Proposition One.

   A third significant change produced by the shift from the state representatives party affiliation to the percentage of votes won by Hutchinson is a change in the relative important of race and ethnicity. The percentage of non-Hispanic blacks within a district is no longer a sizeable or significant predictor of opposition or support of Proposition One. In addition, the percentage of Hispanics within the district slips from being the largest to the second largest factor in predicting support for Proposition One. It is unclear why the importance of the percentage of non-Hispanic blacks would diminish so dramatically with the substitution of Hutchinson?s vote percentage for the party affiliation of the state representative. Perhaps the best explanation for the reduced salience of race is that the Hutchinson vote percentage is such a good predictor of opposition to Proposition One that there is little variance "left over" for the remaining variables to absorb. Nevertheless, the percentage of Hispanic residents within a district is still a very strong and reliable predictor of support for Proposition One. Thus, even though the class and party variables are more fickle, the ethnic cleavage of Hispanics versus non-Hispanics is an enduring one. With both regressions, racial or ethnic variables are always better and more significant predictors of opposition or support of Proposition One than is income.

3. Implications of Texas Findings

   In short, this analysis shows that racial and sometimes class or ideological cleavages divide the supporters and detractors of school finance equalization. The class and ideological divisions are to be expected, but it somewhat surprising to find such strong racial divisions within the electorate over this issue. Within the political and legislative debate over school finance, race or ethnicity was not directly broached as the underlying conflict; from a reading of the newspapers, one could not discern a racial conflict. Instead, the issue was usually debated in terms of burdensome taxes on the middle class, or a wasteful, inefficient educational establishment. While those issues may be have been salient and persuasive to a number of individuals, in the aggregate, blacks and Hispanics in Texas view this issue much differently than whites—even taking into account their respective economic and ideological positions. In significant ways, school finance equalization in Texas is not about taxes and economic issues, but about racial cleavages over educational opportunities.

Conclusion: What Can Courts Do About School Finance?

State supreme courts can have substantive effects on the equity of school finance. Figures 1-4 illustrate this. Their efforts to do so, however, will engender equally substantive mass political opposition—some of which will be racially based. This public opposition is in many ways a constant to school finance reform. As a result, the success or failure of courts? efforts to improve the equity of school funding in primary and secondary education depends ultimately on the capacity of the legislature to withstand this heated political opposition. Courts can act decisively in the face of legislative recalcitrance, by threatening to use injunctions to enforce compliance with the courts decrees—and court's have relied on this threat to ensure passage of politically unpalatable school finance reforms.^\20\ But other than the negative sanction of a threatened school shutdown, there is little a court can do to compel a legislature to act. Instead, it must rely on the legislature?s capacity to forge political coalitions to enact reform legislation. These coalitions of interest groups and key legislators are highly susceptible to public opinion. Thus, a full account of court?s capacity to alter significantly the equity of school financing would link public opinion on school finance with interest group pressure on legislators, who must also confront supreme court justices with injunctions on their minds. Although it is beyond the scope of this paper to provide such links, further examinations into the relative strengths and weaknesses of state supreme courts to accomplish their school finance reform goals must engage this interplay of mass public opinion and interest group coalition building. Courts can achieve impressive results, and they can do so by altering the political calculus legislators and interests groups must make as they decide to support meaningful equalization or oppose it.

Footnotes

List of Cases

References

[Race, Poverty... Curriculum...] [Principles and Practices in Resource Allocation...]