Rich Braunstein University of South Dakota APSA Small Research Grant, 2000 Practicing Democracy: Initiative and Referendum

Voting in the Late 20th Century Introduction American politics is full of examples of the tension between special interests and the broader public interest. From the time Madison wrote about the ratification of the Constitution there have been fears of the effects of interest groups -- or what Madison called factions -- on the political process. In a democratic context, perhaps the greatest tension exists when the desires of a majority of citizens are inhibited by a small, politically powerful minority. Such was the case at the turn of this century. The 20th century brought with it great economic and social transformation. The result of the growing complexity of society was a proliferation of interest groups and ultimately their control of legislative politics. In response to the perceived injustice of special interest dominance, progressive reformers attempted to institute new procedures to cure the evils of faction. During the progressive era, reforms such as the direct primary, direct election of US Senators, womens suffrage and direct democracy (i.e., the initiative and referendum voting) were pursued as means to increase the popular control of governmental processes, thus strengthening the character of democracy in the United States. The research presented here studies the impact of direct democracy in our nations politics. The question of the current research is: Have initiative and referendum elections functioned to increase the democratic character of American politics, as intended by their earliest proponents. More specifically, it questions whether direct democracy, understood as initiative and referendum voting only, produces public policy that is (1) controlled by the public, and (2) offers widely accessible benefits to the community. Much of what has been written about direct democracy in the latter part of this century suggests that initiative and referendum elections are subject to the same abuses they were intended to reform. An often cited conclusion is that the resource bias of well funded interests inhibits citizens from benefiting from ballot issue elections (Smith, 1998; Magleby, 1994; Cronin, 1989). Unfortunately, existing research considering the significance of this institution often fails to provide the systematic analysis necessary to support the general conclusions offered. The current project attempts to resolve some of the controversy by evaluating 34 years of ballot issue elections in California, Colorado, and South Dakota; three states with considerable variation in the use of these procedures, state population demographics, political culture, and economic resources. The project employs an original dataset constructed for this project, including data for each initiative and referendum election in California, Colorado, and South Dakota for the last third of this century (1964-1998), as well as campaign contribution records for the 1992-1998 elections. The current grant proposal is focused on completing this research, which began as a dissertation project at the University of Colorado, Boulder, in 1997. To complete the project, it will be necessary to collect and code campaign finance records for South Dakota ballot measures from 1990-1998. These records are to be added to the campaign finance records of California and Colorado already collected for the final analysis and publication of this study. What follows is a brief description of the research and the costs associated with completing data collection, coding, and analysis.

Research Agenda As noted above, the question remains whether ballot issue elections produce policy that benefits the public. The political science literature suggests, by and large, that the public does not benefit from these elections (Smith, 1998; Magleby, 1994; Cronin, 1989). The overwhelming conclusion has been that the resource bias of well funded groups is too great to allow citizens to use these procedures to their advantage. Unfortunately, much of the contemporary research fails to look beyond aggregate campaign spending as a measure of who participates, from which an inference of who benefits is quickly proffered. The hope for the current research is to provide a systematic analysis of the source and direction of campaign spending, rather than focusing on total spending -as has been done in the majority of previous studies (Lowenstein, 1982; Berg and Holoman, 1987; Zisk, 1987; California Commission, 1992; Magleby, 1994). Although campaign spending has been disagreegated in the past, these studies tend to focus on the more controversial cases (e.g., Smith, 1998 on tax reform; Smith and Herrington, 1997 on parental rights; Tolbert, 1996 on governance policy). This approach ignores the broader concerns of institutional capacity at the heart of the current research. In addition to studying campaign finance, a major part of the current research is focused on the distribution of benefits resulting from ballot issue elections. Here, the goal is to avoid mistaken inference by studying the substance of propositions that qualified and succeeded at the ballot. While it seems essential to include a thorough analysis of the actual type of legislation produced by direct democracy to identify who ultimately benefits from resulting policy rather than who sponsored or blocked it this has not been adequately addressed in the literature. Introducing a systematic analysis of both legislative type and campaign finance will increase our understanding of what this institution contributes to American politics. To this end, the current proposal seeks funding to complete this research. All research costs are associated with the collection and manipulation of campaign finance records. To begin, it will be necessary to acquire committee disclosures for all ballot issue committees active in South Dakota ballot issue elections from 1990-1998. The cost of which is estimated at $900.00. The $900.00 estimate is based on the costs associated with collecting the same records for California and Colorado, adjusted for the number of initiatives and referenda that qualified for the ballot in South Dakota during this period (please see attached budget for individual costs). Once the South Dakota records are collected, it will be necessary to have research assistance to code and enter the data into two separate databases for this research. Coding involves going through the individual contribution records for all ballot issue committees, entering one of 10 possible values1 for each record. Next, the records are tabulated, first by committee, then by proposition. These records are manipulated by a spreadsheet program to develop values for analysis, and entered into a separate statistics package used to conduct the final analysis. Additional support for research assistance is essential to keep this project on schedule for completion by July 1, 2000. There is already one assistant working on this project funded by other grants and awards. To date, his efforts have been concentrated on coding the 1998 California and Colorado elections data, which reflect the final data collection and coding for these two states. The RA, along

The values for the contributor source variable are (1) Corporate, (2) PAC, (3) Political Party, (4) Union, (5) General Business non corporate, (6) Unknown, (7) Candidate Committee, (8) Citizen Group or Ballot Issue Committee, (9) Individual, (10) Public Agency 2

with data collection for 1998 California and Colorado records, have been supported by a $1500.00 Ken and Diane Meier Research Award (2000), and recent Chiesman Fund Grant of $3000.00 for the completion of the California and Colorado components of the project part of that $3000 has been allocated to the purchase of computer equipment for use by research assistants on this and future projects. Almost the entire amount of the Meier Award has been allocated to the purchase of data from the California Secretary of State's office for 1998 records. I am seeking an additional $800.000 for RA support for coding and analysis of the complete set of South Dakota campaign finance records. The research assistant will be responsible for coding approximately 50% of the contribution records. The principal investigator will code the remaining 50%. The cost for research assistance is estimated at $800.00 (100 hours at $8.00 per hour). Thus, the cost of completing the South Dakota component of this research should not exceed $1700 see attached budget. Research Design The main research questions examine the impact campaign spending has on ballot issue outcomes. The research tests the validity of major conclusions in the literature, that contemporary use of direct democracy has been dominated by well funded organizational interests. Again, the concern of this research is that these conclusions have been offered without systematic analysis of disaggregated data. In this study, we examine (1) whether the total amount of campaign contributions for (and against) ballot measures determines the success or failure of ballot measures, (2) if organizational contributions are allocated to special interest ballot issues types2 alone, and (3) the comparative impact of organizational and individual/citizen contributions on ballot issue success. Measuring Dependent and Independent Variables: The dependent variable, PASSFAIL, is a dummy variable coded 1 if a ballot measure passed and 0 if the measure failed. Using PASSFAIL will allow us to study the impact of several independent variables on a propositions success at the ballot. This analysis begins with several bi-variate correlations designed to consider the independent effect of the campaign spending and legislative typology variables. Because some of these variables are introduced for the first time, a wide range of correlations are tested to insure the relevance of these variables to the current analysis. These correlations have the advantage of adding both simplicity and detail to this research. The project continues with a series of multivariate logistic regressions to estimate the effect of independent variables on the probability of a measures success. The list of independent variables includes measures of campaign spending and legislative type3. FUNDSOUR focuses on the main source of monetary contributions for the entire campaign. FUNDSOUR has three values coding the source of campaign contributions: (1) individual, (2) mixed -- when there was less than a 20% difference between funding sources, and (3) organizational. FUNDSUPP, measures the funding source of supporting campaign committees, and FUNDOPP measures the funding source of opposing committees. Values for both FUNDSUPP and FUNDOPP are the same as the values for FUNDSOUR. DOLFOR identifies the total dollar amount spent supporting the issue and DOLAGNST did the same for money spent to oppose the issue. SPENDMOR measures the type of

The legislative type variable is considered in greater depth in earlier stages of this research and is only tangentially related to the study of campaign finance. The variable measures the distribution of costs and benefits of proposed legislation, operationalizing the distinction between policy with public v. private benefits. 3 See footnote 2 above. 3

committee (supporting or opposing) that spent more aggregate dollars. ONESIDED tests the impact of substantial imbalances in the financial resources of competing issue committees, and OSFORAGN measures the object of one-sided spending, either for or against the measures success. Finally, CONTYPE examines the relative impact of various contributor sources beyond the general individual and organizational categories. CONTYPE provides a greater disaggregation of the source of campaign finance for this study and allows us to track the source of contributions to the different legislative types. See footnote one above for a list of the 10 contributor types. Conclusion The ultimate goals for this research are to acquire an understanding of the procedural openness of the initiative and referendum processes, and determine whether inclusive legislation can be successfully pursued through these procedures. This understanding is essential to comprehend our capacity as citizens to participate in the political process and to live under laws we truly consent to, which is perhaps the greatest promise of a democratic society. The task of this research is to locate the potential for this type of governance in the most explicitly democratic procedures practiced in contemporary American politics. Ultimately, this research will contribute to our understanding of the democratic character of this nations politics.

BUDGET PROPOSED BUDGET PROJECT COSTS EXPENSES: Data Collection Travel and Lodging to SD Secretary of State $250.00 Purchase and Copying of Contribution Records $600.00 Purchase and Copying of Supporting Documents $50.00

EXPENSES: Research Assistance 100 hours @ $8.00 per hour $800.00

TOTAL EXPENSES " $1,700.00 "