Proposals to reform the initiative process are motivated by one or more of the following criticisms: there is too much money and/or too much interest group influence in the process; voters are incompetent; laws are poorly written; initiatives undermine the legislative process; and initiatives infringe upon minority rights. However, few reformers base their proposals on sound empirical evidence of the nature and extent of these problems. In this paper, I describe a number of reforms aimed at addressing each of the five common criticisms of the initiative process. I next review the recent empirical political science literature to assess the validity of each criticism. I then assess each reform in light of the empirical evidence. I end by discussing a number of recent US Supreme Court decisions that pertain to initiative reform and consider the prospects for reform in the future.
Whether one generally supports or opposes the use of initiatives, virtually all observers have ideas about how to improve the initiative process. Over the last two decades, an abundance of proposals to reform the initiative process have been suggested or instituted in a number of states. Some of these attempts at reform are quite modest in scope, dealing with such apparently mundane issues as the numbering of propositions or the formatting of the ballot pamphlet. Others call for a complete overhaul of a state’s direct legislation institutions.1 Together, these reforms seek to change virtually every aspect of the initiative process, from who can propose initiatives and how signatures are gathered, to how ballots are organized, how campaigns are run, how information is disclosed, how state legislatures participate, how initiatives are amended, and how courts behave.
As different as these many individual reform attempts appear, they share a common logic. Reformers believe that current direct legislation institutions lead to a particular set of outcomes (or at least make some outcomes more likely); that these outcomes diverge from the reformers’ view of some “ideal;” and that changing some aspects of direct legislation institutions would lead to different, more preferred outcomes. In some cases, this logic is made clear as reformers identify problems with the initiative process and propose clear (institutional) solutions to those problems. In other cases, the logic is more vague. In any case, reformers view institutional change as the key to achieving more preferred outcomes.
In this chapter, I seek to “unpack” the logic underlying some of the major recent proposals to reform the initiative process. I argue that most recent reforms are designed, either explicitly or implicitly, to address one or more of the following common criticisms of the initiative process:
- Too much money/ interest group influence in the process
- Voters are incompetent
- Laws are poorly written
- Initiatives undermine the legislative process
- Initiatives infringe upon minority rights
To illustrate this logic, consider the second criticism. Some critics believe that because there are few restrictions on content or format, direct legislation propositions tend to be long, complex, and technical. These long and complex measures are too much for voters to comprehend. As a result, voters are unable to make competent decisions in the voting booth and their aggregate decisions may not reflect the “informed” decisions of a fully competent electorate. A proposed solution to the problem of incompetent voters is to limit the length of propositions and/or require their language to be simple and readily accessible. Voters, then, would have a better chance of understanding the propositions and making good decisions.
The problem with most reform proposals is that while they may be designed to address what their proponents perceive as real problems with the initiative process, they typically lack sound empirical foundations. Specifically, reform proposals are rarely based on sound empirical evidence of either the existence of a problem (i.e., do initiative outcomes in fact fail to reflect the informed decisions of fully competent voters) or of the impact of direct legislation institutions on outcomes (i.e., are uninformed decisions caused by long or complex propositions). Such empirical foundations are critical to knowing whether proposed reforms will solve the problems they are intended to address. They also allow us to predict, more generally, what the consequences of the proposed reforms might be.
Unlike the reformers, political scientists have much to say about how the initiative process works and, to some extent, about how direct legislation institutions affect political outcomes. A number of recent studies have analyzed the initiative process, providing empirical evidence to help us answer questions such as: How much influence do interest groups have? What is the role of money? Are voters competent? Are laws poorly written? How do legislatures respond? How do minorities fare?
The purpose of this chapter is to bring this empirical evidence to bear upon an assessment of recent reform proposals. I will do so by reviewing the body of political science research that pertains to each of the major criticisms motivating recent reform proposals. For each criticism, I will discuss a number of specific reform proposals that explicitly or implicitly attempt to address the problem. I draw these examples from recent commission reports, proposed legislation, and law review articles. I will then survey the empirical evidence to establish the extent to which the criticism is valid. I will then evaluate each reform proposal in light of the empirical evidence. I will end by discussing recent US Supreme Court cases that create the constitutional boundaries within which successful reform must take place. In the end, I hope to provide readers with a firm basis for evaluating the potential for improving the initiative process.
Too much money/ too much interest group influence
Perhaps the most common criticism of the initiative process is that there is too much money involved. And, indeed, critics apparently have much to complain about. In California alone, proponents and opponents of nine direct legislation measures spent over $54 million in the 1998 primary election and are estimated to have spent over $200 million on twelve measures in the general election (California Secretary of State 1998a). This spending represents the continuation of a trend towards more and more expensive direct legislation campaigns that began in the mid-1980s (California Commission on Campaign Financing 1992). As a comparison, in the same year, all general election candidates for the US House of Representatives together spent $88 million, while candidates for the US Senate spent $49 million (Federal Election Commission 1998). Certainly, claim the critics, this money must be having some effect on direct legislation outcomes.
Closely related to, but analytically distinct from the concern about money is the more general concern about interest group influence in the initiative process. Most modern observers interpret the Populist and Progressive vision for the initiative process as a way for grassroots movements to pass laws without the participation or agreement of state legislators, political parties, or organized interest groups. In modern times, however, the initiative process seems to bear little resemblance to this vision. It is practically impossible for unorganized citizen movements to mobilize the vast resources required to run a successful statewide initiative campaign. Interest groups play a key role in virtually every successful initiative campaign. Their role is critical in every phase of the initiative process, from drafting initiatives, to qualifying them for the ballot, to financing and running campaigns, to defending the legislation after the election. Organized interest groups provide a vast array of resources, including manpower, expertise, and perhaps most importantly, money.
Of course, interest groups and money are not synonymous. Many interest groups that participate in the initiative process are mass membership citizen organizations that lack the ability to mobilize monetary resources. These groups often have severely limited budgets. They depend, instead, on their large numbers and organizational resources as their main source of political power. Likewise, a vast majority of campaign contributions to direct legislation campaigns come not from organized interests groups, but rather from single businesses and individual citizens. One recent study estimates that these two categories of contributors accounted for nearly 72% of contributions to initiative and referendum campaigns in eight states in the late 1980s and early 1990s (Gerber forthcoming). Still, there is an important sense in which interest groups and money are linked. Many of the largest and most important players in the initiative process are interest groups. Because they are already organized, interest groups may have an advantage in mobilizing resources – monetary and other – which individual businesses and citizens do not enjoy. Their ability to coordinate and target these resources may provide an advantage in influencing the course of direct legislation.
Concerns about organized interest groups and their money are hardly limited to the initiative process. Since the earliest days of the American republic, money has played an important role in politics, and observers have expressed their concerns (for a review, see Thayer 1973). Many reforms have been proposed or implemented to try to limit the role of money in politics and campaigns, such as limiting campaign contributions and requiring extensive disclosure of campaign activities. Whether these reforms have succeeded in limiting either the amount or the effects of money in politics remains open to debate, however, and many observers argue that moneyed interests now play an even greater role in politics than ever before (for a recent account, see West and Loomis 1999. But see Sorauf 1992 for an opposing perspective).
While money in politics is a general concern, four major factors make the initiative process especially vulnerable to the influence of interest groups and their money. First, due to several US Supreme Court decisions, contributions to direct legislation campaigns are constitutionally unlimited, allowing groups with money to spend as much as they wish to try to influence direct legislation outcomes. In its landmark decision in Buckley v. Valeo (1976), the Court established basic principles with regard to the regulation of campaign financing. Although the decision considered the legality of the 1974 Amendments to the Federal Election Campaign Act and therefore strictly applied only to federal elections, many of the arguments made in Buckley laid the foundation for future court decisions that directly or indirectly addressed campaign financing in direct legislation campaigns.
In Buckley, the Court upheld limits on contributions to candidate campaigns. The Court argued that contributions to candidates were given to secure a political quid pro quo. Unlimited contributions could lead to corruption on the part of elected officials, whose campaign debts would lead them to compromise their principles in order to secure such contributions. Limits on contributions would therefore serve to reduce corruption, or at least the appearance of corruption, resulting from over-reliance on large individual contributors. Thus, the Court ruled that the state’s interest in reducing corruption justified the abridgment of contributors’ First Amendment rights. However, the Court struck down all forms of expenditure limits on the grounds that they impose direct and substantial restraints on the quantity of political speech without the benefit of reducing corruption.
With respect to spending limits in direct legislation campaigns, such expenditures are subject to the same reasoning that led the Court to invalidate limits on spending in candidate campaigns in Buckley. These expenditures are considered expressions of speech protected by the First Amendment. With respect to contribution limits, however, the Supreme Court views candidate campaigns and direct legislation campaigns differently. In First National Bank of Boston v. Bellotti (1978), the Court established the right of corporations to make contributions to and expenditures on behalf of ballot measure campaigns. In its majority decision, the Court argued that contrary to prohibitions on contributions to candidate campaigns, where the potential for actual or perceived corruption associated with campaign contributions is great, the same corruptive potential does not exist in direct legislation. Following this line of argument, the Court in Citizens Against Rent Control v. City of Berkeley (1981) reaffirmed prior decisions by further invalidating any limits on contributions to direct legislation campaign committees. Again, the Court deemed there was no potential for corruption, that is, no quid pro quo associated with these contributions that would justify the curtailment of contributors’ First Amendment rights.
A second factor that exposes the initiative process to the influence of interest group money is that voters in direct legislation elections choose policy directly rather than electing representatives who choose policy on their behalf. As a result, direct legislation, especially the initiative, lacks the deliberative quality of legislative policy making. While legislative deliberation may generate problems of its own, it is still the case that no such formal deliberative medium, however imperfect, exists in the initiative process. Therefore, to the extent that interest groups are able to manipulate or mislead voters, the lack of deliberation means that an important external check on groups’ influence is absent.
A third factor that opens the initiative process to the influence of interest group money is that many direct legislation propositions are complex, technical, and unfamiliar to voters. Voters therefore tend to have very little prior information about, or understanding of, the propositions they are asked to evaluate. Few actors besides organized interest groups themselves are likely to have such substantive information. As a result, voters who desire information about the content of propositions have few alternatives but to rely on interest groups for that information. When interest groups can influence the channels of political communication, they may be able to use their informational advantages to mislead voters.
A fourth factor is that many of the short cuts and low information cues that voters rely on when they lack information in candidate elections are absent in direct legislation elections. Perhaps the most important missing cue is partisanship. In candidate elections, voters can infer a great deal about a candidate’s policy positions and likely voting behavior simply by learning his or her partisanship (Popkin 1991). In such cases, it may be both convenient and rational to vote on the basis of party. In direct legislation elections, by contrast, partisan cues are generally absent. Although partisan candidates and officials may endorse particular initiatives and referendums, political parties rarely do. Therefore, party cues – so important to voters in other settings – are absent or ambiguous in direct legislation campaigns. Candidates also have past histories from which voters may learn about a candidate’s likely future actions (Fiorina 1979). New policy initiatives have no such history, making the voters’ problems all the more complicated. The absence of useful cues makes direct legislation voters especially reliant on substantive information disseminated by interest groups during the campaign.
Many recent reform proposals aim to limit the role and influence of interest groups and their money in the initiative process. In describing these reforms, it is useful to note that money enters the initiative process at three key points. First, during the drafting and qualifying stages, interest groups pay political consultants to help them write their propositions and obtain signatures for ballot qualification. Second, during the campaign stage, proponents and opponents of a measure establish campaign committees and spend financial and non-financial resources to promote their side. Third, during the post-election stage, proponents and opponents finance legal challenges to successful initiatives and lobby state legislators to amend initiative legislation. Proposed reforms seek to limit money and its influence at each of these three stages.
Drafting and Qualifying Stages
At the drafting and qualifying stages, proponents of initiative legislation can draw upon internal expertise and manpower to draft and qualify their propositions, or they can expend monetary resources to hire consultants to draft and qualify their measures. The use of consultants and paid signature gatherers has been one of the favorite targets of initiative reformers. Indeed, since one of the primary justifications for the signature requirement is to measure popular support for a proposition, many argue that this purpose is defeated when a proponent can simply pay professional petition circulators who care little about the initiative legislation (Garrett 1999). Furthermore, given the tactics of paid signature gatherers, who have little incentive to engage potential signers and fully explain the proposed legislation, it is unclear that even the people who sign their petitions necessarily support the initiative (Collins and Oesterle 1995).
Despite these concerns, the US Supreme Court has upheld the use of paid signature gatherers. In Meyer v. Grant (1988), the Court struck down a Colorado law prohibiting payment of petition circulators. In its decision, the Court argued that prohibiting such payment “restricts political expression in two ways: First, it limits the number of voices who will convey [the proponents’] message and the hours they can speak and, therefore, limits the size of the audience they can reach. Second, it makes it less likely that [the proponents] will garner the number of signatures necessary to place the matter on the ballot, thus limiting their ability to make the matter the focus of statewide discussion….” Since the forms of expression limited by the prohibition on paid signature gatherers take place in an “area in which the importance of First Amendment protections is ‘at its zenith,’” they require an equally compelling justification by the state, which the Court concluded was not present.
Despite the Court’s decision in Meyer, numerous attempts have been made to restrict the use, if not the existence, of paid signature gatherers. Colorado legislation passed in the wake of Meyer attempted to regulate paid circulators in a number of ways. One provision required petition circulators to wear nametags that identified them as paid or volunteer. This provision – identification of the circulator’s status – was based on the assumption that citizens would be less likely to sign petitions if they knew the circulators were professionals rather than volunteers (Garrett 1999, p. 39). By providing citizens with information that would dissuade them from signing petitions, paid circulators would have more difficulty obtaining their signatures and the advantage of the proponents who paid them would be reduced. However, in its decision in Buckley v. American Constitutional Law Foundation (1999), the Court struck down these requirements, ruling that the state lacked sufficient justification to warrant restricting the potential for political speech in this manner.
A second approach to reducing the impact of paid signature gatherers is to increase the length of the circulation period. Circulation periods currently range from several months (e.g., 150 days in California) to two years. A number of reform proposals recommend lengthening the circulation period (see Collins and Oesterle 1995, California Commission on Campaign Financing 1992). The argument behind this reform proposal is that by lengthening the period during which proponents can obtain the required number of signatures, groups who rely on volunteers will be under less pressure to also use paid signature gatherers. Groups who rely on professional circulators will therefore have less of an advantage over their volunteer counterparts. Other observers (i.e., DuBois and Feeney 1992) argue that this reform is likely to have little impact.
A third approach to reducing the impact of paid signature gatherers is to employ a two-tiered signature requirement (Lowenstein and Stern 1989, California Commission on Campaign Financing 1992). Under this proposal, states would have two separate signature requirements, one (higher than the current threshold) for signatures gathered by paid circulators and a second lower one for signatures gathered by volunteers. The logic behind this reform is to provide initiative sponsors with an incentive to rely on volunteers. In doing so, the proponents argue that some groups who now rely on paid circulators will be priced off the ballot. Other groups who now cannot mobilize enough volunteers will be able to meet the lower threshold. The state of Nebraska is currently considering such a proposal (Boellstorff 1995).
A fourth proposed reform is to allow alternative means for gathering signatures. The California Secretary of State is currently considering allowing initiative sponsors to obtain signatures over the Internet (California Secretary of State 1998bc). This reform would reduce the impact of paid signature gatherers by providing groups with a low cost alternative to human signature gatherers. Opponents of this reform argue that signers with Internet access are unrepresentative of the general public, and that the Internet medium discourages debate and public discourse (Garrett 1999).
A fifth proposed reform is to eliminate the signature requirement altogether, relying instead on a simple cash payment to the state (referred to as the “Cynic’s Choice” in California Commission on Campaign Financing 1992, pp. 168-169) or a lottery system to determine which initiatives will make it to the ballot (California Commission on Campaign Financing 1992, Garrett 1999). Some argue that such a lottery should be preceded by a state-run public opinion poll to assess the level and intensity of popular support (Garrett 1999).
A sixth reform proposal is to adopt some form of public financing. This could be in the form of state subsidization of the cost of signature collection, especially when signatures are collected primarily through electronic means (Garrett 1999). One could also envision more traditional forms of public financing, such as direct payments to ballot measure committees, perhaps tied to voluntary expenditure restrictions.
As I explained at the beginning of this section, the US Supreme Court has ruled that contributions to and spending on direct legislation campaigns is constitutionally unlimited. In First National Bank of Boston v. Bellotti (1978), the Court argued that in contrast to prohibitions on contributions to candidate campaigns, where the potential for actual or perceived corruption associated with campaign contributions is great, the same corruptive potential does not exist in direct legislation. In Citizens Against Rent Control v. City of Berkeley (1981), the Court further invalidated any limits on contributions to sponsoring committees using direct legislation. Again, the Court deemed there was insufficient evidence of a potential for corruption associated with these contributions that would justify the curtailment of First Amendment rights.
Since these decisions invalidate attempts to limit the amount of campaign contributions and spending at the campaign stage, most reform attempts in this area focus instead on limiting the influence of money. The primary way reformers seek to limit the influence of money in campaigns is through disclosing the sources of contributions. Most states require contributors to ballot measure campaigns to form campaign committees. These committees are then required to file detailed periodic reports documenting the contributions they receive and the expenditures they make. Some states require committees to report all campaign activities; others require disclosure of contributions only over some amount (DuBois and Feeney 1992).
While campaign committee disclosure reports are publicly available, they are rarely easily accessible. Thus, additional attempts to disclose the sources of campaign contributions include requirements to state the main financial supporter of a ballot measure on paid campaign advertisements (California Commission on Campaign Financing 1992). California’s Proposition 105 required that the primary source of campaign contributions (not simply their committee’s name) be identified on all print, radio, and television advertisements. However, the State Supreme Court struck down the measure on the grounds that it violated the state’s single subject law. Other states such as Michigan and Oklahoma also require initiative campaigns to disclose their primary sources of support.
Another reform intended to improve the disclosure process is to require on-line filing of disclosure statements (California Secretary of State 1998b). Since final summaries of campaign finance activities are now available only months after the election, they are of little use to voters who are trying to discover the supporters and opponents of a measure. Internet technology makes instantaneous on-line access a real possibility.
After the election, proponents and opponents continue to battle over initiative legislation. Most successful initiatives are now challenged in state or federal courts. Millions of dollars are spent in these legal challenges, and observers are concerned that wealthy interest groups are able to thwart the will of the majority by dominating the legal process (Holman and Stern 1998). In addition, state legislatures have the power to amend initiative legislation in most states, although often only after some waiting period or by a supermajority vote (DuBois and Feeney 1992). Initiative proponents and opponents spend substantial resources to influence post-election legislative decisions regarding successful initiatives.
One reform proposal designed to limit the potential for interest group influence over (post-election) judicial review is made by Holman and Stern (1998). They advocate replacing a single judge with a three-judge panel to hear challenges to initiative legislation. Increasing the number of judges would arguably reduce the ability of interest groups to influence legislation by influencing or corrupting a single judge.
Many reforms have been proposed to reduce interest group influence over post-election legislative amendments. These reforms include proposals to limit campaign contributions to state legislators (for recent examples, see California’s Proposition 208 and 212). State legislators respond to interest group pressure because they desire something the interest groups have – technical information, political information, constituency support, organizational resources, or money. Restrictions on campaign contributions or other transfers of resources from interest groups to elected representatives may limit their influence in all phases of the legislative process, including the post-election amendment stage.
All of these proposed reforms assume that interest groups influence direct legislation outcomes with their money and other resources, and that limiting the amount and influence of money would change outcomes in a significant and favorable way. What is the evidence on this front? First, I will consider evidence regarding the role of money at the qualifying stage.
It is clear that money is sufficient to qualify a measure for the ballot (see Garrett 1999). Campaign consultants in most direct legislation states will provide professionals to circulate petitions. Proponents pay per signature (anywhere from $.30 to $1.50), and consultants “guarantee” the required number of signatures, plus up to 10% extra to compensate for error or fraud. Although there is no solid data on this topic, it appears from anecdotal evidence that professional petition circulators can qualify just about any measure, no matter how narrow, obscure, extreme, or obnoxious, for a price. In a large state like California, that price is between $1 million and $2 million.
Not all proponents have access to a million dollars to qualify their measure. Those who lack sufficient monetary resources rely either entirely or partially on volunteers. Other proponents are priced off the ballot altogether. Further, even proponents who appear to represent broad-based, grassroots citizen interests tend to rely on organized interest groups to offset the steep cost of qualifying and later campaigning for their initiatives. “Citizen” groups such as Common Cause relied on paid circulators to qualify their recent initiatives (California Secretary of State 1996a). In a recent study, Smith (1998) found that behind the allegedly “populist” tax reform movements in several states, and their associated tax limitation initiatives, were highly professional campaign organizations and substantial corporate or economic interest group backing. While I disagree with Smith’s ultimate conclusion that involvement by organized interest groups necessarily implies that the tax reform movement was not a truly populist movement, I agree that his case studies highlight the importance of organized interest groups and their resources in the initiative process.
Given the great advantage that money provides at the qualifying stage, we would expect the set of measures that make it to the ballot to reflect this bias towards wealthy interest groups. Surprisingly, however, the bias, if one exists at all, is not overwhelming. Examining ballots in a number of states, we find that some measures are supported by narrow economic interests with vast financial resources, but a surprisingly large number are supported by broad-based citizen groups. For example, of the 87 initiatives that qualified for the ballot in eight states between 1988 and 1992, 29 received majority support from economic interests, 44 from citizen interests, and 14 from other contributors (Gerber forthcoming). There are at least two explanations for the lack of bias towards wealthy economic interest groups. One is that groups who have the money to qualify their measures with paid signature gatherers can also use those resources for other purposes, and so don’t bother pursuing their interests through the initiative process. A second possible explanation is that groups know that qualification is just the first step in a long and costly process. Many additional resources will be required to run a successful campaign and pass an initiative, and the group might not have (or choose to spend) those resources.
Still, even if groups know that the probability of passing an initiative is slim, it might still benefit them to qualify their measure for the ballot. Garrett (1999) argues that qualifying a measure can have important indirect effects, most importantly by placing the issue on the legislative agenda. She notes that there are always many groups competing for a place on the political agenda, too many to all receive consideration from legislative policy makers. Groups have many ways of attracting legislator attention to their issues – playing off critical events, making campaign contributions, engaging in traditional lobbying, mobilizing grassroots pressure, and organizing an initiative petition drive. A petition drive may have advantages over these other methods of affecting the legislative agenda, since it both signals the level and intensity of interest group preferences and mobilizes broad-based public support. Thus, groups with sufficient monetary resources to qualify an initiative gain an advantage in influencing the legislative agenda, and ultimately policy outcomes, even if their measure enjoys little support from the broader public.
Whereas money is sufficient to qualify a measure for the ballot, recent research indicates that it is far from sufficient to pass initiative legislation. A number of studies have addressed the role of money in the initiative process, particularly at the campaign stage. The basic finding of this research is that while some amount of money is necessary, allowing proponents to purchase advertisements, mass mailings, and other forms of campaign information, groups with vast financial resources are regularly stymied at the ballot box. Something important is lacking from their expensive campaigns.
A recent study by Gerber (1998 and forthcoming) provides insight into why money is insufficient to guarantee success at the polls. Gerber distinguishes between several forms of influence that groups might try to pursue through the direct legislation process. Influence can be direct, in which the effect on policy results directly from an initiative or referendum, or indirect, in which a group uses direct legislation to affect policy in some other arena (i.e., the legislative process). Influence can also modify or preserve the status quo policy. Thus, the four forms of influence are direct modifying (e.g., passing new initiatives), direct preserving (e.g., blocking initiatives), indirect modifying (e.g., pressuring state legislators to pass new laws), and indirect preserving (e.g., pressuring state legislators to block new laws).
Gerber’s basic argument is that each form of influence requires different resources. To achieve direct preserving influence (e.g., blocking initiatives), opponents of an initiative require monetary resources (i.e., money) or personnel resources (i.e., manpower, volunteers, or expertise) to campaign against a measure. Often, a large campaign directed against a ballot measure is sufficient to turn voters against it. Lowenstein (1982) and Owens and Wade (1986) each show that when opponents waged “big no” campaigns, defined as campaigns in which opponents outspent proponents by at least 2 to 1 and spent at least $250,000 (adjusted for inflation in the Owens and Wade study), the measures they targeted were defeated approximately 90% of the time. While neither study provides a micro-level model to explain why voters respond to “big no” campaigns, this ability to defeat measures by spending large sums on an opposition campaign probably derives from voter risk aversion. When opponents spend enough to create doubt in the minds of voters, voters may prefer to stick with the status quo policy – flawed though it may be – rather than risk moving to a new policy that just might have very bad consequences.
To achieve direct modifying influence (e.g., passing new initiatives), by contrast, groups require both monetary and personnel resources. This is due to the types of information and cues that voters use when deciding how to vote on ballot measures. Studies by Lupia (1994) and Bowler and Donovan (1998) illustrate how voters overcome vast informational shortcomings in initiative campaigns. One of the main findings of both studies is that voters rely heavily upon endorsements to link their own interests to a position on an initiative. When “citizens like them” support an initiative, voters are more likely to support the measure; when narrow, wealthy economic interests are known to support a measure, voters see less in common with the endorsers and are less likely to support a measure. Therefore, to mobilize voter support, groups must have broad-based citizen constituencies with whom regular voters affiliate. Money alone does not buy this affiliation.
Finally, to achieve indirect modifying or indirect preserving influence, groups require either monetary or personnel resources. Groups with monetary resources can use (unsuccessful) initiatives to provide legislators with a costly signal about their preferences on a given issue. If legislators value the groups’ support or desire their resources in the future, they may pass or block legislation on the group’s behalf. Groups with personnel resources can credibly threaten to pass an adverse initiative if the legislature does not act. Legislators may prefer to pass moderate compromise legislation themselves rather than relinquish agenda control to initiative proponents. In either case, groups need not expend the full amount of resources required to achieve direct modifying or direct preserving influence.
Given the different resources required to achieve each form of influence, we expect different types of groups to pursue them. Specifically, economic interest groups with vast financial resources tend to lack the personnel resources required to achieve direct modifying influence. We therefore expect them to expend most of their direct legislation resources on achieving direct preserving or indirect influence. Citizen interest groups can more easily mobilize personnel resources; for them, the difficulty comes in mobilizing monetary resources. When they are successful in extracting monetary resources from their citizen membership, we expect citizen groups to more often pursue direct modifying influence.
These expectations are borne out in the data. Analyzing the detailed disclosure reports for all ballot measure campaign committees in eight states between 1988 and 1992, Gerber (forthcoming) finds that 74% of contributions from citizen interests are made to support initiatives and referendums, while only 32% from economic interests are made to support ballot measures (p. 94). This difference is consistent with the expectation that citizen interests will try to pass new laws by initiative, and economic interests will try to defeat proposed initiative legislation. In the same study, Gerber also finds that these differences in interest group activity translate into important differences in outcomes. Initiatives that receive majority support from citizen interests pass at a much higher rate (50%) than those that receive majority support from economic interests (31%), and the set of successful initiatives reflect greater citizen group support (47% of total contributions) than economic group support (37%, with 16% from other contributors). In other words, Gerber’s analysis shows that economic groups with primarily monetary resources tend not to pursue direct modifying influence, and their interests are not strongly reflected in the initiatives that pass.
What is the role of money at the post-election stage? For analytic purposes, it is useful to separately consider two aspects of the post-election stage: judicial review and legislative amendments. To the author’s knowledge, there are no empirical studies of the role of money in influencing court decisions. Clearly, some money is necessary to run a successful defense or challenge to initiative legislation. Since state officials are charged with defending initiative legislation once it passes and becomes law, the state effectively subsidizes the cost of the proponent’s defense, thereby reducing their financial burden.
The other aspect of the post-election stage regards legislative amendments. Here, the important question is whether groups can use financial resources to influence whether state legislators amend initiative legislation. There is a vast empirical literature on the role of money in legislative politics at both the congressional and state legislative levels (see Smith 1995 for a review of the literature on interest group influence in Congress; see Gray and Lowrey 1996, Thomas and Hrebenar 1996 for work on interest group influence in state legislatures). While specific findings are mixed, the main conclusion from this body of research is that the effect of money, typically in the form of campaign contributions, is important but limited. Many studies show that contributions cannot “buy votes” and, in fact, are not intended to. More likely, they are intended to “buy access.” Access then allows interest groups to influence the legislative agenda, make the groups’ positions heard, and otherwise affect the content of legislation. This research also indicates that while money is important in obtaining access to the legislative process, so might be other resources that legislators value (such as promises to deliver votes, in kind contributions, etc.).
To summarize, the literature on the role of interest groups in the initiative process identifies when and how groups can use their monetary and other resources to achieve various forms of influence. In the qualifying stage, money plays an important role in determining what laws qualify for the ballot and hence what issues voters ultimately decide. However, there are other ways that groups with limited financial resources can qualify their measures, and in fact, the set of issues that qualify for the statewide ballot in most states reflects a mix of support from economic and citizen interest groups. In the campaign stage, money plays a more limited role. Groups can use money to block initiatives they oppose or to pressure state legislators; however, they need much more than money to pass initiative legislation. Finally, in the post-election stage, groups can use either monetary or personnel resources to defend or challenge initiative legislation, and to lobby legislators to pursue or block amendments to initiative legislation.
This empirical research has important implications for prospects for reforming the initiative process. The research suggests that money may be most influential at the agenda setting stage. To a large extent, money affects what gets on the ballot. If reformers wish to open the initiative process to interest groups who lack the resources to qualify their initiatives under current institutions, proposals such as extending the qualification period or instituting a two-tiered system may help level the playing field. The effects of such efforts will be severely limited, however, if those groups are subsequently unable to mobilize the resources required to run a successful initiative campaign.
Money may also help groups pressure legislators to pass or block amendments to initiative legislation, or even to pursue legislation in policy areas where initiatives fail. To limit this sort of influence, reformers ought to look not to the initiative process itself, but to regulating the role of money in state legislative politics. A number of states already have limits on campaign contributions to state legislative candidates, on the model of congressional campaign finance laws. Strengthening the role of political parties in state legislative elections, perhaps through public financing of party activities, is one way of reducing the reliance of individual candidates on wealthy economic groups.
A second common criticism of the initiative process is that voters are not competent to decide complex matters of public policy. As evidence, critics often point to the low levels of information that voters report having about the details of initiative legislation immediately before elections. For example, the Field Poll periodically surveys California voters about their positions on upcoming statewide initiatives. In their survey immediately preceding the November 1998 general election, between 30 and 60% of likely voters in their sample reported not being aware of three of the four major upcoming initiatives (DiCamillo and Field 1998). Granted, numerous factors probably depress self-reported levels of information – length of time before the election, fear of being asked follow-up questions that probe the respondent’s factual knowledge, and apathy or disinterest in the survey. Nevertheless, anecdotal evidence of this same lack of voter information abounds.
Some of the same features that make the initiative process especially vulnerable to interest group influence also exacerbate the information problems that voters face. First, initiative legislation is often technical and complex. Initiatives cover a multitude of issues regulating social, economic, and political behavior. Some initiatives are written in complex legal language and are difficult to discern on their face; others appear quite simple but have vast implications that are difficult for even policy experts to comprehend. Voters, who typically have very little experience with the policies they are evaluating, decide these issues directly. By contrast, voters in candidate elections choose representatives who decide for them. These representatives either have expertise themselves or can rely on professional staffs and other policy experts. Second, many of the important cues that voters rely on in candidate elections are absent in direct legislation elections. Most importantly, party cues, which are so important in guiding voter decisions in candidate elections, are either absent altogether or are difficult to infer. Other characteristics that aid voters in candidate elections but that are missing in direct legislation elections include candidate incumbency, experience, and personal characteristics.
A number of reform proposals seek to make initiatives easier to understand. Several of these reforms are based on the assumption that voter incompetence derives from the complexity of the initiatives themselves. For example, several proposals seek to limit word length or require initiative language to be below some specified reading level (California Commission on Campaign Financing 1992, Collins and Oesterle 1995). A number of states already have such restrictions in place (DuBois and Feeney 1992). Other proposals seek to lower voter informational demands by limiting the number of initiatives on the ballot (California Commission on Campaign Financing 1992, DuBois and Feeney 1992, Gais and Benjamin 1995) or by grouping together on the ballot multiple measures dealing with a single issue (Dubois and Feeney 1992). These reforms assume that with fewer initiatives or distinct issues on the ballot, voters can devote more of their time and energy concentrating to each individual measure. Still others seek to institute or more vigorously enforce a single subject law (Collins and Oesterle 1995). Single subject laws reduce complexity by limiting the number of subjects that can be contained in a single measure.
A second set of reform proposals seeks to enhance voter competence by changing the campaign environment. As with the reforms discussed above, these changes attempt to ease the voter’s informational burden. Unlike the above reforms, however, they do so by providing information that voters can use to better understand initiative propositions. One such reform requires public hearings on all qualified initiatives (California Commission on Campaign Financing 1992). These hearings would provide a mechanism for political and technical experts to directly disseminate information to voters, opinion leaders, and policy makers. They would also provide the media with low cost access to such information. Along similar lines, Gais and Benjamin (1995) call for a standing Constitutional Commission that is empowered to call hearings if the legislature fails to, and Tornquist (1998) advocates a Citizens Initiative Review Committee to assess all qualified initiatives and provide an impartial review to the public. A number of states already allow or require public hearings on qualified initiatives (DuBois and Feeney 1992).
A second reform that aims to enhance the voter’s informational environment is to improve the content and availability of the ballot pamphlet. A number of states now mail an official voter guide (also known as a voter handbook or ballot pamphlet), which describes the propositions in detail, to registered voters in the state. Reformers in other states propose the same (Cronin 1989). However, the quality of the information contained in these pamphlets varies widely. In the late 1980s, California reorganized its ballot pamphlet to include a summary of signed arguments for and against each proposition at the beginning of the pamphlet. This improvement seems to have increased voter reliance on the pamphlet. In 1982, only 27 percent of Californians reported using the ballot pamphlet as a source of information about ballot measures (Magleby 1984, p. 133). In 1990, a majority of California voters reported using the ballot pamphlet, including 62 percent of highly educated voters (Bowler and Donovan 1998, p. 57). Reform proposals recommend further improvements (California Commission on Campaign Financing 1992, DuBois and Feeney 1992).
A third reform that improves the voter’s information environment is to require better disclosure of campaign activities. All initiative states except Nevada require campaign committees to disclose some of their campaign activities, including by-name disclosure of contributions or expenditures above some amount. 19 states require all campaign committees to register with state election officials, while 14 require individual contributors to separately file (DuBois and Feeney 1992). Voters may use this information to determine the sources of support and opposition for a proposition. Reformers propose more stringent reporting requirements, including lower thresholds and more covered activities. Along these lines, California Secretary of State Bill Jones’ proposal to require on-line disclosure would make information about campaign activities immediately available to voters during the campaign, rather than several months after the election (California Secretary of State 1998b).
It is one thing to require committees and individuals to disclose their campaign activities; it is another to disseminate that information to voters. And in many states, information on campaign activities is publicly available but not easily accessible. Voters must rely on interest groups or the media to wade through the raw disclosure reports. Naturally, they are less likely to provide this information on low salience measures. This problem of disseminating information on campaign activities has generated a number of reform proposals. 20 states require initiative advertisements to identify the sponsors of the ads. Some, such as California, provide detailed requirements such as the size and color of print for these disclosure statements on mass mailings. Numerous reform proposals recommend more extensive requirements for identifying an advertisement’s sponsors (California Proposition 105, California Commission on Campaign Financing 1992, DuBois and Feeney 1992).
Since the behavioral revolution began in the 1950s, political scientists have studied whether regular citizens are competent to make political decisions in a modern democracy and, more recently, in direct democracy. A number of studies have argued that despite low levels of detailed information about specific political issues, regular citizens are capable of making informed political decisions. Key (1966), and later Fiorina (1979) and others argued that voters use readily available indicators of performance, such as retrospective evaluations of the economy, to reward or punish incumbent politicians. Popkin (1991) generalized this argument to contend that voters rationally employ simple cues about candidates, most importantly partisanship, to infer the candidates’ policy positions. Indeed, according to Popkin, obtaining detailed policy-specific information is irrational – busy voters with extreme informational demands can do just as well with easily accessible information such as party.
To what extent do these conclusions about low information voter rationality apply to direct legislation voters? Despite their low levels of reported substantive information, are direct legislation voters able to make informed decisions about initiatives? A number of empirical studies have considered voter rationality in direct legislation elections. In his study of voting on the 1988 California insurance initiatives, Lupia (1994) conducted an exit poll to assess what voters knew and how they voted on the five initiatives. Information demands in this election were especially severe, for several reasons. First, each of the five insurance initiatives was long, technical, and complex pieces of economic regulation legislation, together totaling over 30,000 words. Given the complexity of the measures, most experts were unaware of important provisions in the initiatives; one would expect that voters were even less informed. Second, the campaigns for and against the several measures were deliberately intended to mislead and confuse voters. Trial lawyers and the insurance industry placed four of the measures on the ballot to compete with a pro-consumer measure opposed by both industry groups. Statements by important actors in the campaigns suggest that one of the primary motivations was to lead voters to vote “no” by swamping them with confusing and contradictory information. The campaign committees formed by the industry groups to support and oppose the various measures also adopted misleading names (e.g., Californians Against Unfair Rate Increases) that masked the true identities of the sponsors.
Despite these severe informational demands, Lupia found that when voters possessed just one key piece of information – which measures were supported and opposed by the insurance industry – they were able to link their personal economic self-interests to the measures and behave “as if” they possessed detailed substantive information. In other words, voters used key endorsements to successfully overcome their informational shortcomings.
Building upon Lupia’s basic logic, Bowler and Donovan (1998) examine in detail the various ways direct legislation voters cope with their informational demands. They argue that voters can employ at least three strategies: abstaining or voting “no”; using low-cost information sources; and relying on elite endorsements. They find significant empirical support for each of these strategies. Under the first strategy, voters who lack sufficient information to make informed decisions simply stay home altogether, skip the measures they are unsure about, or vote “no.” In all three cases, voters avoid supporting measures about which they lack information, letting more informed voters decide on their behalf. Under the second strategy, voters obtain information from the ballot pamphlet, newspaper editorials, television ads, and friends and neighbors. These information sources may provide less than complete details about the ballot propositions, but they may provide enough information to allow voters to make reasonable decisions. Similarly, under the third strategy, voters rely on elite endorsements. These endorsements allow voters to link their preferences, and hence their policy positions, to the known preferences or interests of the endorsers. Bowler and Donovan find that the use of these strategies varies predictably across groups of voters with different characteristics. Most importantly, highly educated voters are less likely to abstain and vote no, and are more likely to report reliance on the ballot pamphlet and newspaper editorials. Bowler and Donovan also find evidence that voters use these strategies to make decisions that correspond to the voters’ interests. They hypothesize that voters will be more risk-averse when economic conditions are poor and, indeed, find that negative voting increases in bad economic times. Aggregate voting patterns also correspond to regional interests. Regions that benefit from certain initiatives are, on average, more favorable than regions that suffer from these initiatives.
While Bowler and Donovan find differences across voters in their ability to make informed decisions, Gerber and Lupia (1996) find important differences across elections. Analyzing 36 California statewide initiatives, Gerber and Lupia find that when spending by both the proponents and the opponents of a measure is high, election outcomes more closely reflect their estimates of “informed votes,” as inferred from the preferences of well informed survey respondents. When spending is one-sided or absent altogether, voters lack information about the identity and intensity of a measure’s supporters and opponents, and initiative outcomes diverge from estimated “informed votes.” Gerber and Lupia attribute these differences to the nature of information available in each competitive scenario: in two-sided competitive races, each side has an incentive to reveal the identity, and hence the interests, of the other side. In one-sided campaigns, campaigners can provide false or misleading information without an adversary to expose them. In low-spending campaigns, no such information is available to voters.
Empirical research shows that voters who participate in direct legislation elections can and do overcome their informational shortcomings and make good decisions. They are best able to do so when they have easy access to valuable low-cost information sources. This implies that to enhance voter competence, states should strive to provide all voters with useful information. This includes ballot pamphlets that contain summaries of the measures and identification of (or preferably signed arguments by) the measure’s supporters and opponents. Such efforts seem to have paid off in California, where voters now rely heavily on the redesigned ballot pamphlets. It also includes ready access to disclosure reports in a format that facilitates usage by voters.
Empirical research also shows that voters make better decisions when campaigns are more competitive. Reforms that aim to limit or restrict campaign activity and money make matters worse for voters, as they distort this important information about how intensely groups support and oppose a given measure. Likewise, public funding similarly masks the intensity of interest group preferences. Instead, more extensive and easily accessible disclosure of campaign activities would allow voters to better discern the interests of an initiative’s supporters and opponents and hence to link their own interests to positions on the initiative.
Poorly written laws
A third common criticism of the initiative process is that it produces poorly written laws. Critics contend that in the traditional legislative process, professional lawmakers draft legislation with the input of knowledgeable staff and interest group experts. Bills go through numerous iterations before a final version is produced, and numerous policy actors have opportunities to influence the language of the bill. By contrast, initiative legislation is written by “outsiders” – interest group officials, activists, or regular citizens who may lack lawmaking experience and expertise – and is not subject to the same level of scrutiny as traditional legislation.
Critics contend that the drafting of initiative legislation by non-professionals leads to poorly written initiatives in three different senses. First, they argue that initiative legislation is prone to technical errors, contradictions, lack of clarity, and confusing or ambiguous language (for example, see Magleby 1984 and Cronin 1989). These errors result from mistakes made by non-professionals in writing the legislation.
Second, critics argue that initiative legislation tends to have many unintended consequences (DuBois and Feeney 1992, Schrag 1998). These unintended consequences result from two factors. One factor is that the non-professionals who draft initiative legislation simply lack the experience to anticipate the far-reaching effects of their measures. A second factor is that the initiative process excludes the input of interested parties who might be better able to anticipate consequences. Both factors mean that policy actors who might be better able to anticipate an initiative’s ultimate consequences are excluded from the drafting phase of the initiative process.
Third, critics argue that initiative laws fail to reflect underlying citizen preferences. Initiative proponents are not required to balance the many complex dimensions of citizen preferences. Rather, they must simply offer voters an alternative that a majority prefers to the status quo policy (Ferejohn 1995, Johnston and Lupia 1999). In other words, by allowing initiative proponents monopoly agenda setting power, the process forces voters to choose between the lesser of two evils (Romer and Rosenthal 1978, Lupia 1992).
Of course, all of these criticisms are made relative to the legislative process. In other words, critics contend that the initiative process leads to poorly written legislation relative to the legislative process. Whether or not this is true is an empirical question.
Reforms have been proposed to address each of the three potential problems with the language of initiatives. Reforms intended to reduce technical errors or ambiguity in language include provisions for assistance in drafting from state election officials (California Commission on Campaign Financing 1992, DuBois and Feeney 1992, Collins and Oesterle 1995, Martin 1997). A number of states already make public resources and employees available for these purposes (DuBois and Feeney 1992). These public resources and expertise may be especially valuable to citizen interest groups that lack the financial resources to hire professional campaign consultants. Along similar lines, some states require public hearings in which interested parties can review initiative legislation and offer suggestions for clarifications and improvements (DuBois and Feeney 1992).
Reforms intended to reduce or mitigate unintended consequences include several forms of pre-election review (administrative, legislative, or judicial) and post-election legislative amendments and judicial review. Prior to the election, administrative, legislative, or judicial branch officials can review, analyze, and comment on qualified initiatives to help proponents and voters understand the likely consequences of the legislation. Many states currently allow voluntary pre-election review or require review but do not allow proponents to amend their propositions in response (DuBois and Feeney 1992). Reforms would widen these pre-election review powers (see Michael 1983, Farrell 1985, Martin 1997 on judicial review; see California Commission on Campaign Financing 1992, Citizen’s Commission 1994, Collins and Oesterle 1995, Gerber 1995, Tornquist 1998 on legislative review, see DuBois and Feeney 1992 on administrative review).
After the election, legislators can amend initiative legislation to address consequences that are revealed after the initiative’s implementation. All states except California currently allow some form of post-election legislative amendments, sometimes requiring a supermajority vote or waiting period. A number of reform proposals include provisions to allow easier post-election legislative amendments (California Commission on Campaign Financing 1992, DuBois and Feeney 1992, Citizen’s Commission 1994, Gerber 1995, California Constitution Revision Commission 1996). Some require a legislative supermajority vote or waiting period.
Reforms to reduce the monopoly agenda setting power of interest group proponents include greater involvement by legislators in the initiative process, perhaps through the indirect initiative (California Commission on Campaign Financing 1992, Citizen’s Commission 1994, Gerber 1995, Gais and Benjamin 1995, Collins and Oesterle 1995). Nine states currently allow some form of the indirect initiative. In these states, the legislature is required to consider all qualified initiatives prior to the election. If the legislature passes the measure itself, the proposition is removed from the ballot either automatically or upon the sponsor’s request. If the legislature does not pass the measure, the proposition is placed on the ballot as a regular initiative, sometimes requiring additional signatures for ballot qualification. This reform tends to be very appealing to political elites who desire a greater role in the initiative process. However, voters tend to be reluctant to provide the legislature – who they distrust even more than they distrust their fellow citizens – with significant power over initiative legislation (DiCamillo and Field 1997). A second proposed reform that reduces an interest group proponent’s monopoly agenda power involves allowing the legislature to propose a compromise bill to appear on the ballot with the initiative. Voters would then be given a wider range of policy options – the initiative, the legislature’s compromise, and the status quo – to choose from. This practice is currently used at the national level in Switzerland (Treschel and Kriesi 1996). Ferejohn (1995) also recommends a similar reform.
Unfortunately, there is little empirical research on whether initiative legislation is, in fact, poorly written. We do observe that most successful initiatives are challenged in state or federal courts, and many are invalidated in part or whole. It is also important to note, however, that many measures passed by state legislatures are also challenged in the courts and are overturned, or are amended by the legislature within several years. Some but clearly not all legislative amendments are due to errors in drafting; presumably, some but not all court challenges to initiative legislation are also due to errors in drafting. There is no reason to expect that initiatives would have fewer errors than regular legislation.
Progressive reformers made initiatives difficult to amend precisely because their purpose was to circumvent state legislators in order to pass legislation that legislators (or their financial constituents) opposed. This means that, to the extent that critics are correct in asserting that initiative legislation is prone to drafting errors and unintended consequences, the ability to fix such errors is severely limited. In some states, legislators must wait 3 to 5 years after a statutory initiative’s passage before it can amend the legislation. In other states, legislative amendments to statutory initiatives require a supermajority vote. In all states except Delaware, if the initiative is passed in the form of a constitutional amendment, the legislature must also obtain a majority popular vote for any subsequent amendments. This popular vote requirement applies to all initiatives, statutory and constitutional, in California.
Reforms that make it easier for legislatures to amend initiative legislation would reduce the impact of drafting errors by allowing amendments after the initiative is implemented and its consequences are observed. However, these reforms are likely to be met with harsh criticism from direct legislation advocates who resist empowering the legislature vis-a-vis voters. Reforms that involve legislative, administrative, or judicial branch actors in the drafting stage of the initiative process may be more feasible. Pre-election review allows input from experts and professionals in government who have more experience drafting legislation. From an efficiency perspective, pre-election involvement may be less efficient than post-election amendments since the actors must anticipate the consequences of a proposed initiative before passage or implementation. From a feasibility standpoint, however, pre-election involvement may be less offensive to reformers and direct legislation advocates since the initiative’s sponsors retain final agenda control and can simply ignore the reviewers’ recommendations.
Undermines legislative process
From its conception, direct legislation was conceived as a complement to representative government (Magelby 1984, Cronin 1989). Even direct legislation’s strongest proponents never intended for it to replace traditional forms of lawmaking via elected representatives; rather, the initiative (and referendum and recall) were seen as ways to supplement the legislative process when circumstances (i.e., interest group influence) prevented elected representatives from responding to broad-based citizen preferences. In recent years, however, many observers claim that direct legislation no longer fulfills this complementary role (if, in fact, it ever did). They argue that the existence and use of direct democracy undermine, rather than complement, representative government itself (see e.g., Gamble 1997).
Critics argue that direct legislation undermines representative government in at least three ways. The first way is by providing a mechanism for voters to pass laws that directly or indirectly limit the power of the legislature (Tolbert 1998). Many initiatives are aimed at directly limiting the power of legislative actors. Examples include term limits, campaign finance reform, open primaries, and nonpartisan redistricting. Others limit the power of the legislature more indirectly. For example, initiatives that restrict the state’s taxing authority shrink the pool of resources available for the legislature to distribute. Initiatives that mandate spending on particular programs limit legislative power both by reducing their discretion over the target programs, and by reducing the amount of discretionary funds available to spend on other programs (Gerber et al., 1999).
Some argue that a second way direct legislation undermines the legislative process is by providing elected representatives with incentives and opportunities to shirk (Schrag 1998). There are a number of issues on which elected representatives would simply prefer not to act. The most obvious examples are political reform issues – campaign finance reform, term limits, redistricting – that reduce the power or discretion of elected representatives. Legislators may also prefer not to act on highly contentious or divisive issues for fear of alienating important constituencies or colleagues in the legislature. When voters take up these issues through the initiative process, legislators may feel less pressure to act. Indeed, they can justify their inaction in the name of democracy, arguing that they are simply “letting the voters decide.” In states where legislators are prohibited from amending initiative legislation, there are few incentives for legislators to act in areas where future initiatives are likely.
The third way direct legislation allegedly undermines the legislative process is by forcing legislators to place greater weight on the desires of interest groups when they formulate policy (Cronin 1989). Here, the argument is that narrow interest groups may be able to use the initiative process to pressure legislators. Gerber (1998) argues that groups may be able to use the initiative process to pressure legislators in at least two ways. First, groups that have sufficient resources may be able to credibly threaten to pass an initiative that legislators oppose. Legislators may prefer to pass a compromise law itself rather than forfeit authority in this policy area. Note that “sufficient resources” does not necessarily mean lots of money. As I explained in my discussion of interest group influence, sufficient resources means having both the monetary resources and the personnel resources required to pass an initiative. Second, groups that lack sufficient resources to pass a law directly may nevertheless be able to influence legislators by expending resources. Such costly effort can be interpreted by legislators as a signal of the group’s level and intensity of preferences on an issue. If legislators value the group’s support and resources, they may respond to the group’s signal, even if the policy they support (or oppose) fails to amass majority support. Matsusaka and McCarty (1998) suggest a third way that groups can use direct legislation to affect legislative policy. They argue that when the legislature is uncertain about a group’s preferences, the group may be able to effectively pressure legislators to pass non-majoritarian policies.
Note that in the first example of indirectly pressuring the legislature, groups are able to move legislative policy closer to the majority preference. In the latter two examples, groups are able to move the legislature away from the median voter. In all three examples, the initiative process reduces the legislature’s autonomy and empowers interest groups to influence legislative policy.
Note further that not all of the effects described in this section are necessarily and unambiguously detrimental to the legislative process. For example, if legislators can avoid taking sides on contentious issues, they may be able to lower overall levels of partisan conflict and work more closely on bipartisan efforts in other areas. Similarly, if pressuring legislators to be more responsive to majority or minority interests enhances the representation of important groups in society, representative government as a whole may be enhanced. Still, in all of cases described above, the legislative process is weakened in the sense that legislators lose discretion to respond to the interests it wishes.
It is difficult to imagine reforms proposing to limit the ability of voters to pass political reform laws or spending mandates. In fact, the ability to pass laws that legislators oppose is one of the widely perceived benefits of the initiative process. However, to the extent that such laws are seen to unduly weaken the legislative process, several reform proposals may dampen the impact of such policies. For example, reforms that allow the legislature to more easily amend initiative legislation after the election would make such laws less binding on legislators (California Commission on Campaign Financing 1992, DuBois and Feeney 1992, Citizen’s Commission 1994, Gerber 1995, California Constitution Revision Commission 1996). Of course, many reformers oppose such changes for precisely that reason.
Other reforms have been proposed to prevent the legislature from shirking due to initiative activity. These proposals require state legislatures to take an active part in the initiative process. They include mandatory public hearings on all qualified initiatives (California Constitutional Revision Commission 1996), legislative or citizen review and comment (Citizen’s Commission 1994, Gerber 1995, Tornquist 1998), and legislative votes (i.e., the indirect initiative – California Commission on Campaign Financing 1992, Citizen’s Commission 1994, Gerber 1995, Gais and Benjamin 1995). Of these, requiring legislative votes on qualified initiatives is the strongest form of legislative involvement.
Finally, I discussed in the section on interest group influence reforms that limit the ability of interest groups to pressure state legislators. These include limiting campaign contributions and empowering state parties to reduce the importance of interest groups in the electoral process.
It is clear that groups often turn to the initiative process to try to limit the power of legislators and the legislative process. Between 1988 and 1992, 20% of the statewide initiatives considered by voters in 8 states and 23% of those that passed dealt with issues of political and governmental regulation (Gerber forthcoming, pp. 117-118). These issues included term limits, campaign finance reform, redistricting, primary elections, requirements for holding office, tax limits, and spending limits. In addition, it is also clear that having the initiative process makes a difference in some of these policy areas. The starkest example is that of state legislative term limits: Of the 19 states with state legislative term limits in 1998, 18 passed the enacting legislation by initiative. Only Utah’s state legislators voted to limit their own terms. Differences in policy due to the initiative process are evident in other areas such as fiscal policy (Matsusaka 1995, 1998).
Somewhat less clear is the ability of groups to indirectly limit the power of legislators through the passage of tax limitations, spending mandates, and other initiative legislation. Some of the most important attempts have been the tax limitation initiatives in California (Prop 13 of 1978); Massachusetts (Prop 2 1/2 of 1980); and Colorado (Amendment 1 of 1990). While California’s Prop 13 is blamed for many of the state’s woes (see, for example, Schrag 1998), serious studies of the direct fiscal consequences of Prop 13 tell a different story. Shires (1999) finds that the main effect has been a shift of revenue sources, not reduction in overall tax burden. In fact, this shift has largely been from local governments to the state. The implication of this finding is that, if anything, Prop 13 has empowered the state by increasing the pool of resources available to it relative to other government actors.
Evidence on the effects of spending mandates is also mixed. In a 1990 report, the nonpartisan California Legislative Analyst Office claimed that when one accounts for all voter initiatives and federal mandates that make claims on the state’s resources, only 8% of the general fund is available for discretionary spending purposes (California LAO 1990). A recent study by Gerber et al., however, sheds serious doubt on this estimate (Gerber et al., 1999). They show that most mandates are at best partially enforceable and few are fully implemented. In other words, because of agency and implementation problems, legislators in fact have much more discretion than estimates such as the LAO’s suggest.
While the research described above considers how laws passed by initiative limit the power of legislators and the legislative process, and number of recent studies have analyzed, both theoretically and empirically, how the mere presence of the initiative constrains or influences legislative behavior. Indeed, one of the major advances in the study of direct legislation in recent years has been the recognition that its effects reach far beyond the confines of the initiative process itself. Many of these insights come from spatial or game-theoretic models of legislator-interest group interaction in which interest groups can use the threat of initiatives to pressure legislative actors (Steunenberg 1992, Gerber 1996 and forthcoming, Moser 1996, Matsusaka and McCarty 1998). These studies identify the conditions under which the initiative process does and does not empower interests groups vis-a-vis the legislature. Insights from these models are now shared by and recognized in other non-formal analyses of the initiative process (Cronin 1989, Schmidt 1989, Magleby 1995, Donovan and Bowler 1998, and Matsusaka 1998).
A number of studies have attempted to estimate empirically the effects of the initiative process on policy outcomes posited in the theoretical analyses. Of course, estimating indirect effects is extremely difficult, since it requires isolating the effects of the initiative process relative to other factors that influence legislative behavior. The common approach in these studies is to compare policy outcomes in initiative states to policy outcomes in non-initiative states, controlling for other factors that may affect policy outcomes (i.e., political, economic, and social differences across states). Any remaining differences in policy outcomes are attributed to the effects of the initiative process.
Matsusaka (1995 and 1998) analyzes differences in fiscal policies in initiative and non-initiative states. Controlling for differences in state demographics that may translate into differences in preferences for taxing and spending policies, he finds that in the early part of the century, initiative states had systematically higher levels of both state and local taxes and of total state and local spending. In the last thirty years, initiative states have systematically lower levels of both taxes and spending. Lascher, Hagen and Rochlin 1996 and Camobreco 1998 consider whether initiative states pass laws in a number of issue areas that are closer to or further from the estimated policy position of the state’s median voter. Using a composite measure of state ideology constructed by Erikson et al. (1993), both studies conclude that policies in initiative states are either indistinguishable from, or are further from the state ideology measure, than policies in non-initiative states. Employing a different measure of median voter preferences, Gerber and Hug (1999) find evidence to the contrary: they find that gay and lesbian protection policies in initiative states do more closely reflect the estimated median voter’s preference. Finally, Gerber (1996 and forthcoming) isolates the effect of the initiative process on legislative policy outcomes by comparing policies in initiative and non-initiative states in areas where state legislatures were responsible for passing all or most of the relevant legislation. She finds that legislatures in initiative states pass parental consent and death penalty laws that more closely reflect the states estimated median voter’s preference on those policies than legislatures in non-initiative states.
To summarize, the evidence on indirect effects is somewhat mixed. Most studies find systematic differences in policy between initiative and non-initiative states. However, the studies produce different conclusions about whether policy more or less closely reflects the state’s estimated median voter’s preference.
The empirical evidence shows that voters have used the initiative process to successfully regulate the powers of legislative branch actors and, somewhat less successfully, to limit legislative discretion through the passage of tax limitations and mandated spending programs. The evidence also shows that legislatures in initiative states pass different laws than legislatures in non-initiative states, presumably because of the ability of interest groups to threaten to propose adverse initiatives or to signal their preferences through initiative activities. The evidence is mixed, however, on the types of groups that can use direct legislation to threaten or signal the legislature. Some studies conclude that policies in initiative states more closely reflect the preferences of broad-based (majoritarian) interests; others conclude that they more closely reflect the preferences of narrow (non-majoritarian) interests.
These conclusions have important implications for understanding the extent to which the initiative process undermines representative government. On the one hand, they show that the effects of the initiative process extend beyond the immediate policy consequences of successful initiative legislation. On the other hand, however, it is not clear whether these effects translate into a stronger or weaker legislature. Some observers begin with the assumption that state legislators are dangerously detached from voters. Because of the incentives created by institutions such as primary elections and the reliance on interest group money, legislators respond not to broad-based constituents but rather to narrow, wealthy special interests. This leads to unrepresentative policy, alienated voters, and unhealthy democracy. From this perspective, reforms that force legislators to be more responsive to broad-based citizen interests are the answer. One example of this type of reform is the indirect initiative. By requiring legislators to vote on all qualified initiative legislation, the indirect initiative institutionalizes the type of pressure described in the theoretical models above.
Other observers begin with the assumption that modern state legislators are extremely weak already. Federal mandates and other restrictions on their activities severely limit legislative discretion. Making matters worse, many state legislators now face term limitations, preventing them from amassing policy expertise and further reducing their effectiveness. From this perspective, what legislators need are not less discretion, but more. To the extent that interest groups are able to limit legislative power and discretion through the initiative process, appropriate reforms lessen this influence. Examples of such reforms are post-election legislative amendments.
Disregard for minority rights
A fifth major criticism of the initiative process is that it exposes minorities to potentially severe rights violations. Direct legislation is, by definition, a form of lawmaking by majority rule. A majority interest can propose, qualify, and pass legislation without any regard to the interests, preferences, or demands of others. In some cases, these “others” are important groups in society with substantial numbers (but less than a majority) and important claims on societal resources. In other cases, they are “minorities at risk,” groups that society deems worthy of protection. Since the initiative process involves decision making by simple majority rule, affected minorities have few ways of protecting their interests from hostile majorities but to try to defeat the initiative or, if unsuccessful, to take their case to the courts. By contrast, representative government is designed to provide minorities with many opportunities to influence policy. They have formal opportunities to participate in deciding what issues will be considered and how bills will be drafted. Most importantly, minorities have numerous veto points at which they can protect their interests by blocking detrimental legislation (Shepsle and Weingast 1987). No such veto points exist in the initiative process.
Somewhat ironically, this criticism and the first – that moneyed interests can “buy” favorable policy outcomes through the initiative process – are motivated by opposite views of what types of groups can use initiatives to advance their interests. Critics who complain about the influence of money are concerned that policy minorities are able to pass initiatives not because of their electoral strength, but rather because of their superior monetary resources. Critics who complain about undermining minority rights are concerned that majorities, not minorities, can use the process to advance their interests at the expense of minorities. Of course, in reality, those who criticize the process as being captured by wealthy interest groups are concerned about one type of policy minority, namely largely narrow economic interest groups. Those who criticize the process as disregarding minority rights are generally concerned about a different type of policy minority, primarily racial, ethnic, or social minority groups. Still, these two perspectives reflect very different views of whether minorities or majorities can use initiatives to their advantage.
Concerns about the anti-minority biases of direct legislation date to the earliest days of American democracy. In the founding period, few intellectuals advocated large-scale direct democracy, and most dismissed any but the most limited forms of citizen participation in government. Perhaps the most forceful (and influential) articulation of this latter view is found in Madison’s Federalist 10. Madison warns of the potential for tyranny of the majority and makes the case for a republican government to protect the interests of minority factions. The opposing view was well represented by Jefferson, who advocated a more participatory government, perhaps including small-scale self-governing communities. Note, however, that even Jefferson’s vision excluded anything so radical as statewide direct legislation.
In recent years, concerns about the anti-minority bias of the initiative process have heightened. Initiatives are now frequently used to decide social and moral issues with direct implications for racial, ethnic, and social minorities. Indeed, as observers such as Schrag (1998) contend, the politics of immigration reform, affirmative action, gay rights, and bilingual education, embodied in statewide initiatives in several states, have been the most important state-level political battles of the 1990s. Many believe that minorities suffer under the will of the majority, who are quick to sacrifice the rights and policies minorities have achieved through the traditional legislative process.
A number of reforms have been proposed that would allow policy minorities more participation in drafting initiative legislation. Most importantly, mandatory public hearings would provide a forum for minorities to make known their positions on prospective initiatives (California Commission on Campaign Financing 1992, Gais and Benjamin 1995). Of course, the value of public hearings in protecting minority rights depends on whether or not initiative sponsors choose to incorporate other perspectives into their propositions. In some cases, initiative sponsors might accommodate minorities whose positions they had previously not considered. In other cases, initiative sponsors might accommodate minorities to pre-empt legal action that is brought up during the hearings. In still other cases, however, where a sponsor has interests that are directly antithetical to a minority, there is little reason to believe the sponsor will be more accommodating after hearing the minority’s appeal.
Other reforms would better allow minorities to protect their interests through the legal system. Of course, these reforms would be limited to providing a way to protect a minority’s constitutionally granted rights and not other policy advantages they may have gained through the legislative process. Two forms of judicial review enhance the protection of minority rights. The first is pre-election judicial review (see Michael 1983, Farrell 1985, Martin 1997). Under this reform, courts rule on the constitutionality of initiative legislation before it goes to the ballot. If the measure is ruled unconstitutional, it is either removed from the ballot or left on the ballot with a comment regarding its constitutionality. Proponents argue that in addition to protecting minorities against clear cases of rights violations, this reform would lead to substantial cost savings if unconstitutional measures are removed from the ballot and costly post-election legal battles are avoided. Opponents argue that it is extremely difficult for the courts to decide such matters as rights violations until a law is implemented and its consequences are felt.
The second form of judicial review that would enhance the protection of minority rights is a more activist post-election judicial review. Most important successful initiatives are now challenged in state or federal courts. Judges already rule often in favor of minorities, overturning many the initiatives they consider. Holman and Stern (1998) encourage judges to more actively police and protect minority rights, and advocate a three-judge panel to hear such cases. Beyond these recommendations, however, it is not clear that specific institutional reforms would lead to better protection of minorities.
A number of recent articles consider the consequences of direct legislation for minority rights. Several analyze the success of anti-minority initiatives and referendums in various jurisdictions. Gamble (1997) analyzes a number of anti-minority initiatives and referendums in American states and cities. Finding that 78% of the anti-minority measures in her sample passed, she concludes that direct legislation significantly curtails minority rights achieved through the legislative process. Donovan and Bowler (1998) question this result. They demonstrate that, at least at the state level, an analysis based on a more complete dataset of measures on the civil rights of gays and lesbians contradicts part of Gamble’s findings. They find a passage rate of only 18% for the statewide anti-gay and anti-lesbian measures in their sample. Since this passage rate is lower than for other initiatives, they argue that the anti-minority potential of the initiative process is quite limited. Finally, Frey and Goette (1998) show that in Switzerland, comparatively few measures restricting minority rights have passed in popular votes. They observe passage rates of 20% (for national level), 62% (for cantonal level), and 23% (for local level) for anti-minority measures.
These three studies represent major advances in understanding the consequences of the initiative process for minority rights. Rather than relying on philosophical or partisan political arguments, they are based on careful empirical analyses. They consider the effects of direct legislation at various levels of government on a range of social minority groups. And their results are cumulative, each study building upon the past to show the strengths and weaknesses of previous research. Nevertheless, the analyses, and the inferences that can be drawn from them, are all limited in two fundamental ways. Perhaps most importantly, by focusing on passage rates of anti-minority initiatives, they are limited to assessing only some of the potential effects of initiatives on minority rights. In the language presented in the discussion of “undermining the legislative process” above, they consider only direct effects and ignore any possible indirect effects of the process. However, if we are concerned with the ultimate policy consequences of having direct legislation institutions, then clearly both direct and indirect effects are potentially important.
Gerber and Hug (1999) argue that focusing only on direct effects can bias our inferences about the consequences of the initiative process in two ways. If the threat of initiatives makes legislators more responsive to anti-minority views in the population, then analyzing only direct effects will understate the anti-minority potential of the initiative process. If votes on anti-minority initiatives lead legislators (at the same or different levels of government) to compensate affected minorities by passing more protective legislation, then analyzing only direct effects will overstate the anti-minority potential of the initiative process.
The second limitation of the Gamble, Donovan and Bowler, and Frey and Goette articles is that they fail to account for differences in voter preferences across states. Indeed, the patterns they reveal could be caused by differences in preferences that happen to correlate with the existence of the initiative.
Building on these two points, Gerber and Hug (1999) analyze the relationship between institutions, voter preferences, and policy for ten policies dealing with protections against discrimination for gays and lesbians. They find that the critical factor in determining policy is not simply whether a state has direct legislation institutions. Rather, they argue that direct legislation simply aggregates the preferences of the majority. They find that on some policies, the majority has preferences that involve restricting the rights of minorities. In these cases, direct legislation states have more anti-minority policies than non-direct legislation states. On other policies, the majority prefers policies that enhance the rights of minorities (such as hate crime laws), and direct legislation states have less anti-minority policies than non-direct legislation states. In other words, it is not simply direct legislation institutions that may bias policy against minority rights, but rather direct legislation institutions in the hands of an anti-minority majority.
To summarize, the empirical evidence on the alleged anti-minority bias of direct legislation institutions in general, and the initiative process in particular, is mixed. Some studies conclude that voters use direct legislation to strip minorities of the advances they have made through the legislative process. Others claim that the direct effects of initiatives on minorities are quite limited. Still others claim that the total effects of direct legislation – both direct and indirect – are conditional upon the majority’s preferences. In the end, it is clear that to fully appreciate the ultimate impact of direct legislation on policy, we must account for both direct and indirect effects.
Considered on their own, there are clearly some direct anti-minority effects of direct legislation. Donovan and Bowler argue that since the passage rates of anti-minority initiatives are lower than for initiatives in other policy areas, concerns over these effects are misplaced. However, I would argue (and do so in Gerber and Hug 1999) that if any anti-minority initiatives are passing, then some minority groups are being adversely affected by the initiatives process (whether or not legislatures compensate these groups is a separate question, and the empirical evidence suggests that legislatures reinforce, rather than offset, anti-minority preferences). Reforms that may help reduce the potential for anti-minority bias in direct legislation measures are public hearings. Public hearings may help to inform citizens of the potential harms of anti-minority initiatives. To the extent that anti-minority preferences come from ignorance rather than malice, providing such information may reduce support for anti-minority initiatives.
The empirical evidence also shows that indirect effects may be important. Therefore, one way to offset the direct anti-minority effects of the initiative process is to institute reforms that make legislatures more responsive to vulnerable minorities. One approach is to more extensively involve the legislature in conceiving and drafting initiative legislation, perhaps through mandatory pre-election legislative review. A second approach is to encourage more independent legislative involvement in policy areas that initiative proponents pursue, perhaps through requiring legislatures to debate and vote on qualified propositions as in the indirect initiative.
Multiple Effects, Externalities, and Interactions
It is convenient to organize the discussion of proposed reforms as responses to specific perceived problems with the initiative process. However, as is evident from the discussion above, we really should not consider these problems and their proposed solutions in isolation. Some of the reforms have multiple effects and are proposed by reformers to address multiple problems. Some are proposed to address one or more problems, but in so doing, exacerbate other problems with the process or create new ones. Others, if enacted simultaneously, interact with one another and may produce unpredictable results.
A single reform might address more than one problem. Indeed, different reformers concerned with very different aspects of the initiative process have proposed a number of the same reforms discussed above. A prime example is the case of public hearings. Public hearings are proposed as a way of increasing voter competence (by increasing the quality and quantity of information available to the public), reducing the possibility of poorly written laws (by facilitating participation by experts and interested parties), and enhancing involvement by the legislature (by involving them in the hearings). We might think of these multiple effects as “positive externalities” – additional benefits associated with a given reform.
Not all externalities are positive, however, and not all effects of a given reform are necessarily positive. A reform that is design to solve one problem with the initiative process might, in fact, exacerbate another problem. Perhaps the strongest example of negative externalities arises in the context of campaign finance reform. Campaign finance reform has been advocated as a way to reduce the role of money and interest groups in the process. We noted above that the empirical evidence casts serious doubt on whether this reform is likely to achieve its stated purpose, since the effects of money are quite limited. Worse, it is likely to create new problems for voters. Limiting campaign contributions distorts the information available to voters regarding the strength and intensity of preferences of an initiative’s supporters and opponents. In other words, by attempting to reduce the amount of money in the initiative process, reformers advocating campaign finance reform remove or weaken an important electoral cue and make the problem of voter incompetence more severe.
A second example of a reform that may produce negative externalities is restricting competing measures. Proponents argue that restricting competing measures enhances voter competence by making their choice simpler (i.e., voting yes or no on a single issue). While there is little or no empirical evidence that voters make bad decisions because their choices are too complex, simplifying the task voters face may nevertheless have some positive effects such as enhancing voters’ feelings of efficacy and increasing voter turnout. The downside, however, is that restricting competing initiatives limits the opportunities for legislative involvement in the direct legislation process. Legislatures are often the proponents of competing measures. Thus, in the 1990 California general election, the state legislature placed four referendums on the ballot in response to initiatives it opposed. The referendums were widely viewed as moderate alternatives to the extreme interest group measures. Likewise, in Switzerland, most popular initiatives are accompanied on the ballot by a legislative alternative, and often the original initiatives are ultimately removed from the ballot before the election (Butler and Ranney 1994). Placing competing measures on the ballot provides an opportunity for legislatures to respond to interest group initiatives and creates an incentive to actively participate in the direct legislation process. Restricting legislatures’ ability to propose competing measures removes these opportunities and incentives, and effectively sidelines them from the policy process.
A final consideration is the extent that multiple reforms interact. Consider what would happen if a jurisdiction simultaneously enacted reforms requiring more extensive disclosure and public funding. Disclosure is intended to provide the public with information about campaign activities. If those activities are financed or subsidized by the state (or county or city), however, then the information conveyed in the disclosure reports in ambiguous at best. Presumably, reformers would anticipate this interaction and choose one reform or the other. Consider also what the consequence would be of simultaneously increasing the length of circulation period and limiting number of propositions. If increasing the length of the circulation period has its intended effect of increasing the number of (grassroots) initiatives that qualify for the ballot, additional reforms would be necessary to select among those that qualify.
It is not surprising that reform proposals have multiple effects, externalities, and interactions. Many of the criticisms of the process are closely linked, emphasizing different aspects of a single complex process. What this discussion makes clear is that it is critical to recognize the tradeoffs involved in any reform proposal. We cannot consider possible weaknesses with the process in isolation, nor can we expect to improve all or even several aspects of the process at the same time. Improvements on one dimension might necessarily involve impediments on another, and reformers need to be aware of these tradeoffs.
Reform and the Courts
Empirical political science provides guidelines for understanding how the initiative process works and what reforms would change the initiative process in the ways envisioned by their proponents. Of course, if such proposed reforms are unconstitutional, then debating their potential efficacy is of little practical use. In this section, I discuss several of the major US Supreme Court decisions regarding the use of direct legislation and attempts to reform the initiative process. These decisions form the broad parameters in which effective reform must take place.
Validity of the initiative process
In its decision in Pacific States Telephone and Telegraph v. Oregon (1912), the US Supreme Court validated the basic principle of direct legislation and its use in the American states. The case concerned a challenge to an Oregon law, passed by initiative, requiring licensing of telephone and telegraph companies. The litigants (PST&T) argued that the law was unconstitutional because it was passed by initiative. Their contention was that the use of direct legislation was inconsistent with the US Constitution’s “Guaranty Clause.” The Guaranty Clause asserts that each state is guaranteed a republican form of government. The litigants argued that by providing citizens with direct policy making authority, states that allow direct legislation no longer have strictly republican (i.e., representative) government and hence are in violation of the Clause. Any laws passed by this unconstitutional process, they argued, would be themselves unconstitutional.
In its decision, the Court ruled that questions raised under the Guaranty Clause are nonjusticiable. They contended that the issue concerned the “framework and political character of government” and not the content of the legislation, and thus was not subject to judicial review (Hahn and Kamieniecki 1987, p. 14). By its decision not to rule on the constitutionality of the law, the Court effectively established the validity of the institution of direct legislation. Lowenstein (1995) notes that state courts have followed Pacific States and have declined to address whether the initiative process violates the Guaranty Clause (p. 284).
It is unlikely that the Court will revisit the basic question of whether the initiative process is itself constitutional; its position has been further reaffirmed in many subsequent cases. Invalidating the constitutionality of the initiative process would implicate not only future laws, but also the large body of state legislation already passed by initiatives during the last century. Rather than contest the validity of the initiative process as an institution, reformers – legislators, voters, administrators, academics, and activists – take the process as given and seek to regulate how it is used. A number of these reform attempts have been the subject of Supreme Court decisions. These decisions, then, form the regulatory framework in which effective reform must take place. I consider them below.
Decisions regarding reform
Most of the reforms discussed in this paper seek to regulate some aspect(s) of the initiative process in order to empower certain groups or interests relative to others. Such regulation necessarily involves restricting the rights or abilities of the target groups to engage in certain forms of political speech and activities. So, for example, reforms that seek to reduce the amount of money in the process can be interpreted (and have been interpreted by the Court) as attempts to limit the ability of wealthy interests to express their political preferences through campaign spending. Reforms that seek to restrict ballot content (such as through restrictions on a proposition’s length or language, number of subjects, or number of propositions on a ballot) limit the ability of groups who advocate the restricted provisions from expressing themselves by proposing certain initiatives. Reforms that require more extensive disclosure infringe upon an individual’s right to anonymous speech. In numerous decisions, the Supreme Court has established these rights to express oneself through the use of the initiative process as protected by the First Amendment. Hence any attempts to curtail that right must be justified by a compelling state interest. The more severe the abridgement of First Amendment rights, the more compelling the state’s interest must be (Timmons v. Twin Cities Area New Party ).
I will discuss a number of decisions relevant to reform in four main areas: campaign finance, ballot access, content restrictions, and judicial review. These represent the major areas in which the Court has directly addressed the issue of what types of reforms sufficiently advance the state’s interests to justify the associated abridgement of rights. I certainly do not claim that I include in this discussion all Court decisions relevant to the issue of reform. Many decisions are made in areas quite remote from the initiative process that nevertheless have important implications for reform. And many constraints on reform are imposed not by Court decisions, but by the absence of decisions in a given area. I emphasize the decisions below because both political scientists and constitutional law specialists view these cases as among the most important major recent decisions in each area. My contention is that they are directly relevant to many of the reforms discussed in preceding sections of this paper.
Since I discussed the major cases pertaining to campaign finance in previous sections, I will limit my comments here to reviewing that discussion and to considering what the decisions mean for prospects for reform. The Court’s first major decision regarding the regulation of campaign finance was in Buckley v. Valeo (1976). While Buckley concerned a challenge to the 1974 Amendments to the FECA, and therefore applied strictly to federal candidate elections, much of the logic of Buckley has been applied to the question of regulating campaign finance in direct legislation elections as well.
Among the most important provisions in Buckley was the Court’s emphasis on reducing corruption as a compelling state interest. The Court ruled that the act of making contributions to candidate campaigns created the potential for actual or perceived corruption since it involved a direct trade of money for “representation.” This potential was sufficiently strong to justify abridging the First Amendment rights of potential contributors. Campaign spending is a different story. No such potential for corruption is created by campaign spending, since there is no political quid pro quo. Rather, campaign spending is interpreted simply as direct communication of political views. Hence, on this basis, the Court ruled that the potential for corruption was not sufficiently strong to justify abridging the free speech rights of potential campaigners and invalidated the provisions limiting campaign spending.
The implication of the Court’s decision in Buckley for direct legislation campaigns is simple: both contributing to and spending on behalf of direct legislation campaigns are analogous to spending in candidate campaigns. Neither activity generates the same potential for corruption as contributions to candidates, since they do not involve explicit payments to candidates for political services. Thus, neither contributions nor expenditures in direct legislation campaigns can be limited. Two subsequent decisions expanded upon this basic principle. In First National Bank of Boston v. Bellotti (1978), the Court ruled that contributions by banks and corporations are legitimate forms of political speech, even if the proposed initiative legislation does not directly and immediately affect the contributor. The potential for corruption resulting from these corporate contributions was deemed insufficient to curtail the contributor’s rights, and the prohibition was overturned. In Citizens against Rent Control v. City of Berkeley (1981), the Court ruled that contribution limits to ballot measure committees could not be justified because the corruptive potential of these contributions, made to ballot measure committees and not candidates, is absent.
These decisions have important implications for attempts to reform the initiative process. Any attempts to directly restrict the ability of particular groups or interests to spend their resources on initiative-related activities will be considered a violation of the groups’ First Amendment rights. Thus, reforms that attempt to reduce the amount of money in the process or to restrict the influence of wealthy economic interest groups must be justified with strong evidence of serious corruption resulting from money or interest group influence. The empirical evidence I presented in this chapter, and the evidence upon which the Court has based its decisions, indicate that such corruption does not exist. This is not to say that reformers will not continue to advocate contribution and spending limits. Reformers in California, for example, proposed Propositions 208 and 212 in 1998 fully recognizing that they contained provisions that the Court was likely to throw out as unconstitutional. The proponents pursued these initiatives in an attempt to pressure the Court to reconsider its prior decisions.
Of course, there are other ways of reducing interest group money and influence, short of direct contribution and expenditure limits. One way is to more visibly expose the source of campaign contributions. We know from previous research that voters rely heavily on information about a measure’s supporters and opponents to link their own interests with a position on a ballot proposition (Lupia 1994, Bowler and Donovan 1998). Informing voters about who supports and opposes a measure may harm economic interest groups with vast resources but little grassroots support, since voters are less likely to affiliate with them. Thus, more extensive disclosure diminishes the value of interest group contributions and may dissuade them from participating in the initiative process. Disclosure also informs voters and may reduce the ability of wealthy interest groups to manipulate voters (although, again, there is little evidence that such manipulation actually takes place). A number of states have passed laws that require visible and extensive disclosure of the source of financial support in campaign advertisements (e.g., California, Michigan, and Oklahoma) and qualification drives (e.g., Colorado). The Court has upheld some of these reforms and overturned others, depending upon the evidence of corruption and the severity of the imposed burden on political speech.
Closely related to the issue of campaign finance is the issue of ballot access. A number of states have attempted to restrict the amount and influence of money at the qualification stage. The Court’s two major decisions in this area, Meyer v. Grant (1988) and Buckley v. American Constitutional Law Foundation (1999), consider Colorado’s successive attempts to prohibit (in Meyer) and then regulate (in Buckley) the use of paid signature gatherers. The decisions apply to other states’ attempts to regulate ballot access as well.
Meyer v. Grant considered a Colorado statute prohibiting the use of paid signature gatherers. The case was brought by proponents of a state constitutional amendment who wished to employ professional circulators to qualify their proposition. They argued that the statute violated their rights under the First Amendment. The state, in their defense of the law, argued that “even if the statute imposes some limitation on First Amendment expression, the burden is permissible because other avenues of expression remain open to appellees and because the State has the authority to impose limitations on the scope of the state-created right to legislate by initiative.” The Court disagreed, and struck down the statute.
Four components of the Court’s decision are critical for understanding the implications of this case for future reform. First, the Court ruled that the nature of the burden imposed by the statute – limitation of political expression – warranted exacting scrutiny. This is the same standard the court applied in the campaign finance cases. It meant that the burden on groups wishing to engage in a particular form of political speech (here paying signature gatherers) was sufficiently severe that it could only be justified by evidence of equally severe corruption.
Second, as in the campaign finance cases, the Court failed to see such evidence of corruption. The State argued that the prohibition was required to protect the integrity of the initiative process. Paid circulators, they argued, may lack an incentive to verify that potential signers were qualified to sign petitions (i.e., were registered voters). However, the Court ruled that this argument was insufficient. Volunteers, they argued, who feel passionately about an issue, have no less incentive to accept false signatures. Further, there were other laws that guarded against fraud short of prohibiting paid circulators. This line of argument suggests that the Court does and will likely to continue to require proponents of reform to establish more than the potential for corruption. They must establish that such corruption in fact occurred, and that the reform is necessary to alleviate corruption in the future.
Third, the Court ruled that even if other means existed for affected groups to disseminate their ideas, this did not “take their speech through petition circulators outside the bounds of First Amendment protection.” The prohibition of paid signature gatherers eliminates one of the most efficient and effective means for speech. The First Amendment, argued the Court, protects the right “not only to advocate their cause but also to select what they believe to be the most effective means for doing so.” In so ruling, the Court reaffirmed its position that the direct prohibition on avenues for political speech would not be allowed, except perhaps under the most extreme cases of political corruption.
Fourth, the Court ruled that while states may legitimately regulate commercial speech (as in a case cited by the litigants), the same did not apply to the regulation of political speech.
In response to the Court’s decision in Meyer, Colorado amended its code to allow but regulate the use of paid signature gatherers. The American Constitutional Law Foundation challenged three of the main provisions of the statute: (1) requiring petition circulators to be registered to vote; (2) requiring circulators to wear name badges indicating their address and status as “Volunteer” or “Paid;” (3) and requiring initiative proponents to disclose the name, address, and county of registration of all paid circulators, the amount paid per signature, and the total amount paid for signatures. The state argued that these requirements were necessary to reduce the corruption associated with paid petition circulators. In Buckley v. ACLF, the Court struck down provision 1 and parts of 2 and 3.
The Court ruled that requiring circulators to be registered voters and to wear name badges would significantly reduce the quantity and quality of political speech. Name badges violated the principle of anonymous political speech, and the prospect of having to identify oneself may dissuade potential circulators. The Court argued that to the extent that the state had an interest in policing against fraud in the act of collecting signatures, other provisions of the state’s law (such as requiring circulators to be adult state residents and to attach affidavits to their submitted petitions) sufficiently served this interest. Likewise, the Court argued that voter registration requirements limit the pool of potential circulators. This restriction was judged to be especially severe, since many people who might wish to circulate petitions for a fee choose not to register as a matter of principle. Thus, the Court argued that as in Meyer, both of these requirements impose a burden on proponents who wished to employ paid circulators by potentially reducing their numbers. They also burden potential circulators’ First Amendment rights by restricting their ability to engage in anonymous (but paid) political speech. Note that the Court was not asked to rule on the validity of requiring a circulator to identify his or her status on a name tag, although in the majority opinion, Justice Ginsberg speculated that this provisions would probably not stand on its own.
The Court also ruled on Colorado’s disclosure requirements. The Court upheld the requirement that initiative sponsors submit monthly reports disclosing the amounts they have spent to gather signatures. However, it struck down additional requirements to report the names of paid petition circulators and the amounts paid or owed to each (in both monthly and final reports). The Court ruled that disclosing the sponsor’s identity and total expenditures on qualification was sufficient to address the state’s interest in controlling the domination of the initiative process by affluent interest groups. However, the added benefit of revealing the names of and amounts paid to circulators was not demonstrated. Hence, the burden imposed on initiative sponsors and potential circulators, as above, was not justified.
In terms of interpreting the implications of Buckley for future reform of the initiative process, this case is less straightforward. In Meyer, the challenge was to a blanket prohibition on paid petition circulators. Few questioned whether this prohibition imposed a severe burden on initiative proponents’ First Amendment rights, and so the Court’s decision to apply the standard of exacting scrutiny was straightforward. At that point, the only remaining question was whether the state’s interest was sufficiently compelling to justify the burden. The Court ruled that it was not.
In Buckley, there were a number of restrictions under challenge, and the exact burdens they imposed were not obvious. The provisions did not directly prohibit paid circulators, they merely made the act of collecting signatures for a fee more onerous and hence less readily available. The Court therefore had to determine, in the absence of any reliable empirical evidence, whether such burdens were, in fact, severe. In addition, the potential burdens created by these restrictions affected both initiative sponsors and signature gatherers themselves. The Court therefore also had to determine the nature of First Amendment protections on both sets of individuals.
To summarize, the Court’s major ballot access decisions indicate that direct prohibitions on political speech are virtually impossible to justify, since the Court requires proponents to establish that (1) no less burdensome alternative exists, and (2) that the state’s interest is sufficiently compelling to justify the burden. This is the case even if targeted groups or individuals have other means for expressing their political views. Regulations that burden political speech by requiring identification of the speaker are also problematic, since they burden both the speaker (in this case, the circulator) and his/her principle (the initiative proponent). Disclosure requirements may or may not be viewed as burdensome, depending on what activities must be reported. In Buckley, the burden on circulators was deemed to be greater than the burden on sponsors.
Given the Court’s position, it is clear that for reformers who wish to reduce the role of interest group money and influence in attaining ballot access, their most viable strategy is to work to empower other interests, thereby reducing the relative advantage money may provide. Reforms such as increasing the length of the circulation period or instituting a two-tiered signature requirement make it easier for citizen groups to qualify measures with volunteer signature gatherers. These reforms do not directly restrict the ability of economic interest groups to expend their resources on political speech, since these groups can continue to qualify initiatives using paid signature gatherers. Rather, they work by empowering citizen groups, thereby reducing the advantage of economic groups relative to others. This example suggests that successful, constitutional attempts to change the distribution of power between economic and citizen groups must come through the empowerment of citizen groups rather than through the undermining of economic groups.
Common regulations on ballot measure content include word limits, language restrictions, prohibitions on the naming of individuals or corporations, prohibitions on regulations of the judiciary, and limiting content to a single subject. Most challenges to content restrictions have been made in state courts. In some of these cases, proponents of initiatives that are challenged as violating the content restrictions argue that the regulations themselves are invalid and fail to serve legitimate state interests.
The US Supreme Court in an advisory opinion reviewed one such case. The Florida State Attorney General ruled that a proposed amendment to the Florida State Constitution violated the state’s single subject law. The proposed amendment prohibited any government entity in the state from passing laws establishing rights or privileges for members of various social groups. The Attorney General ruled that the amendment covered multiple subjects, including civil rights, the power of state and local government bodies, the basic rights of all natural persons, and the right of employees to bargain collectively. It also dealt with protections for ten different classifications of people.
The Attorney General petitioned the Supreme Court for an advisory opinion on the validity of the initiative. The Court agreed with the Attorney General’s conclusion that the proposed amendment violated the state’s single subject restriction. The Court then went on to articulate its view of the purpose and legitimacy of the restriction. They noted that the single subject law is a legitimate constitutional constraint that protects citizens from being compelled to vote for provisions of a law or constitutional amendment they oppose in order to vote for other provisions they support.
This opinion illustrates some parameters of the Court’s position on content restrictions. To a limited extent, content restrictions such as single subject requirements share some of the same characteristics of campaign finance and ballot access regulations. They restrict or limit the ability of initiative proponents to place certain measures on the ballot. In a weak sense, this can be interpreted as a burden on initiative proponents’ rights to engage in a form of political speech, namely advocating the initiative of their choice. In the campaign finance and ballot access cases, the Court vigorously protected a proponent’s political speech rights, and required the state to provide extremely strong arguments to justify their abridgement. In the content restriction opinion, the rights of initiative proponents to advocate the initiative of their choice were viewed as much less compelling. Note that the type of political activity being restricted in the two sets of cases is quite different. In the campaign finance and ballot access cases, the political activity in question is pure political speech: expending one’s resources to promote one’s political views. In the content restriction case, the political activity in question involves setting the policy agenda. In the court’s view, these activities warrant weaker protections.
A second way the content restrictions differ from the campaign finance and ballot access regulations is that they pit the interests of one group of political actors – initiative proponents – against the interest of another group – voters. In advocating content restrictions, the state’s interest is in protecting voters from elite (i.e., initiative proponent) machinations. Content restrictions simplify the voters’ choice and allow them to make “better” decisions. For the purposes of understanding the implications of reform, however, it is important to note that “better” in this sense does not necessarily mean “more competent” in terms of information, as we discussed above. Rather, it means decisions that are closer to voters’ true issue-specific preferences. Hence, in the Advisory Opinion, the Court clearly gave the rights of voters more weight than the rights of initiative proponents.
To summarize, the Court’s position on content restrictions, as articulated in its Advisory Opinion to the Florida Attorney General, is that these restrictions on initiative proponents’ activities can be justified to the extent that they protect citizens from having to approve provisions of an initiative they oppose. In this way, the rights of voters outweigh the rights of initiative proponents.
A final area of Supreme Court positions I will address is judicial review. Cases in this area address rather different issues than cases in the previous three areas since they tend not to deal with First Amendment protections. However, they are important for our understanding of what reforms the Court is likely to permit, since a number of reform proposals advocate more extensive judicial, as well as administrative or legislative, review.
Judicial review can take place either before or after the election. Above, I discussed one of the Court’s important decisions regarding post-election judicial review in Pacific States Telephone and Telegraph v. State of Oregon. In this case, the Court ruled that judicial review (at least in the federal courts) must consider the content of initiative legislation and not the legitimacy of the initiative process itself. In other words, it ruled that the courts should treat initiatives no differently than regular legislation.
In Wyoming National Abortion Rights League v. Karpan (1994), the Court considered the legitimacy of pre-election judicial review. The issue in this case is whether state courts can remove a measure from the ballot before it is passed. The Supreme Court ruled that a measure, once qualified according to the state’s procedures, could be blocked from the ballot only if it is found unconstitutional in its entirety. If found constitutional in part, then its submission to the electorate cannot be restrained. The Court did not comment, however, on whether the finding of unconstitutionality can be included on the ballot or what the implications of pre-election findings of partial unconstitutionality were for post-election review (i.e., whether such measures would automatically be subject to post-election review). The Court’s findings in Wyoming impose rather severe limitations on prospects for active pre-election judicial review. It found that the right of voters to consider initiative legislation, once it is legally qualified, is quite strong. This finding also implies that prospects for pre-election administrative and legislative review may also be quite limited.
Summary and Conclusions
I conclude by discussing what I interpret from the empirical evidence as the most important problems with the initiative process and the types of reforms most appropriate to address those problems. I will also consider some normative concerns raised by this discussion.
The empirical evidence fails to support the criticism that there is too much money and/or too much interest group influence in the initiative process. Economic interest groups with vast financial resources are severely limited in their ability to use the initiative process. Most of their activities are defensive – aimed at defeating initiatives they oppose – and the set of initiatives that pass do not reflect their interests or support. To the extent that interest groups can use their monetary resources to gain an advantage in the initiative process, it is through their superior ability to qualify initiatives. The US Supreme Court has found most attempts to directly limit this ability to be unconstitutional. Therefore, any attempts to reduce the relative advantage of wealthy economic interest groups to gain ballot access must be through empowering citizen interests. Reform proposals such as increasing the length of the circulation period or instituting a two-tiered signature requirement achieve this goal. The cost of empowering citizen interests, however, is that the total number of measures to qualify for the ballot may very well increase, perhaps creating greater informational demands for voters.
This minor informational concern aside, there is little empirical evidence that voters are incompetent to decide policy via the initiative process. At least when campaigns are vigorous and information is available about a measure’s supporters and opponents, voters are able to figure out a measure’s likely impact and vote as they would if they were better informed. Reforms that simplify voters’ informational problems may have some benefits, such as increasing voter participation and feelings of efficacy, but they are not likely to change outcomes dramatically.
In contrast, there is compelling evidence that initiatives affect the functioning of representative government. The initiative process is often used to directly or indirectly limit the powers of legislative actors, and policies in areas such as term limitations, campaign finance, tax limitations, and spending limitations differ systematically between initiative and non-initiative states. Interest groups of various types also use the initiative process to pressure legislative actors, and while the empirical evidence is somewhat mixed, the general finding is that they are often successful.
Whether these effects are “good” or “bad” depends on the analyst’s view of the status quo and whether he or she feels legislatures currently have too much or too little power. My own view is that state legislatures are already quite constrained in their ability to respond to changing political, economic, and social conditions. Legislators operate in an environment of myriad mandates imposed by the federal government, voters, and the courts. Recent research suggests that these mandates may not be as binding as previous estimates suggest; nevertheless, they clearly limit the flexibility legislators have to pursue their policy preferences and the interests of those they represent. When coupled with the fact that many initiatives are difficult to amend in most states, and virtually impossible to amend in others, it follows that legislators can hardly be held accountable for legislation increasingly produced in the initiative process. This lack of accountability further weakens the link between voters and their elected representatives.
In previous work (Gerber 1995), I have argued that to increase legislative flexibility and accountability, states should increase legislative involvement in both the pre- and post-election phases of the initiative process. This includes pre-election legislative review with an opportunity for legislative amendments, mandatory pre-election legislative votes on all qualified initiatives, and post-election legislative amendments with a graduated waiting period (after three years for measures passed by the legislature, 5 years for measures passed by voters). All of these provisions involve empowering the legislature, rather than restricting the powers or activities of initiative proponents, and so are likely to withstand Court scrutiny. All are already used in part or in whole in other states. The main concern with this proposed reform is its political feasibility. Since implementation would require a constitutional amendment or revision, and therefore would be subject to popular approval, one must consider its likely acceptance by voters. In the current political environment, it is unlikely that voters would support this reform in sufficiently large numbers. Voters report a reluctance to empower their state legislators, and most political reform initiatives have aimed at reducing, rather than increasing, the legislature’s power. To the extent that observers agree that such reforms would benefit representative government, our task is to think of ways to make them acceptable to voters.
These considerations tie into my view of the effects of the initiative process on minorities. The empirical results presented in this paper show that the effect is conditional upon the preferences of the majority. Sometimes majorities have preferences that are detrimental to minorities, and sometimes they do not. The problem is that, in the short term at least, we cannot ensure that the former will not occur. In Madison’s spirit, it is therefore necessary to control the effects, rather than the causes, of these anti-minority tendencies. Policy minorities tend to have better access via the legislative process. The same sorts of reforms that increase legislative flexibility and accountability may therefore help to protect minorities. These include more legislative involvement in drafting legislation and easier post-election legislative amendments.
Finally, I will address some normative considerations. Throughout this chapter, I have tried to report the empirical evidence from an objective, “positive” perspective. I have also tried to assess the proposed reforms in light of the empirical evidence without bringing in any particular normative judgement. Of course, all observers have their own normative perspective that filters into and colors their perceptions of evidence and their conclusions about the proposal. Indeed, one can quite legitimately object to the initiative process even if it is not broken, simply because one disagrees with the types of interests that are able to use the process or the types of policies that result. If this is the case, it is important to be clear that these are one’s objections are based on normative considerations and not allegations that the process is broken.
 While the set of reform proposals I discuss here are far from exhaustive, they represent a wide sampling of the types of reforms that are being proposed.
 For an alternative view of the Progressive movement, see Deverell and Sitton (1994).
 I use the term “successful” to refer to initiative campaigns that ultimately result in the policy change desired by the campaigners. Depending on the campaigners’ goals, this can include campaigns that result in the passage of a new initiative, the blockage of an initiative, or a favorable response by other policy actors. I describe each of these forms of influence in Gerber (forthcoming) and below.
 While voter decisions in the direct initiative are unmediated by other government actors prior to the election, they are still subject to judicial review.
 Voters in direct legislation elections may use other cues not available in candidate elections. Research by Lupia (1994) shows that voters who know the interests of groups involved in direct legislation campaigns can use this information to infer how to vote. For example, Lupia found that voters who were able to identify which of the five insurance initiatives on the 1988 California ballot were endorsed by the insurance industry were more likely to cast votes that were in their economic self-interest.
 An alternative interpretation is that a modern populist movement can and must rely to some extent on organized interest groups. What defines a modern movement as “populist” is the types of organizations that are involved (citizen versus economic) and the types of interests that are promoted (broad-based versus narrow).
 To accurately assess the extent of bias, one must compare the current ballot to the hypothetical ballot that would occur in the absence of a financial advantage.
 “Economic interests” includes economic interest group, professional groups, and businesses. “Citizen interests” includes citizen interest groups, labor unions, and individuals. “Other contributors” includes political parties and candidates for elected office.
 Garrett (1999) notes that groups who rely on professional circulators may be able to exaggerate public support for their measures.
 Legislators may respond to interest groups, even if their policies are not majority preferred. Indeed, it is quite likely that many laws passed by state legislatures would fail to obtain majority voter support, especially when they deal with political, economic, or social regulations that affect narrow constituencies.
 There are at least two reasons legislators may respond to an interest group’s credible threat. First, if a law passes by initiative, legislators cannot claim any credit for the policy. Second, if a law passes by initiative, legislators are limited in their ability to act in that policy area in the future.
 Gerber (forthcoming) defines economic interest groups as those whose members are businesses or organizational representatives. Citizen interest groups are defined as those whose members are autonomous individuals. Note that the critical distinction lies in a group’s membership composition. Citizen interest groups may pursue policies with economic consequences, and economic interest groups may be involved in issues of only limited economic importance. The importance of membership composition is that it directly affects a group’s ability to mobilize and dedicate political resources.
 The one initiative on which voters did report high levels of awareness was Proposition 5, the Indian Gaming initiative. By late October, an estimated $71 million had been spent on the Prop 5 campaign (Sweeney 1998).
 Several observers have suggested that proponents may deliberately write their initiatives in complex language to deceive voters (e.g., Cronin 1989, p. 208).
 An example of the former is California Proposition 104, an auto insurance initiative in 1988 that contained over 29,200 words and a multitude of technical provisions (DuBois and Feeney 1992). An example of the latter is California Proposition 209, the anti-affirmative action “California Civil Rights Initiative.” Only 369 words in length, this measure had vast implications for public university admissions, hiring, and contracting policy (California Secretary of State 1996b).
 Gerber and Lupia employ a heteroscedastic logit model which weights the preferences of informed voters more heavily and the preferences of uninformed voters less heavily.
 Of course, many interests are excluded from drafting regular legislation as well. The point is that there are fewer formal opportunities for interest group input in the initiative process.
 For these hearings to be an effective means for improving initiative legislation, proponents must be allowed the opportunity to amend their propositions prior to the election. However, some states such as California that require public hearings for qualified initiatives do not allow such amendments since they change the legislation that originally qualified.
 Such pre-election review may also help proponents clarify language and otherwise improve the drafting of initiative legislation.
 For example, a number of states have passed tax limitation laws by initiative.
 See Smith 1998.
 Gerber estimates the state’s median voter’s preference as the percent of respondents in each state who reported favoring the policy the Pooled Senate Election Study surveys.
 In this section, the term “minorities” refers to any group that comprises less than a majority of the total population. These policy minorities include but are not limited to racial and ethnic minority groups.
 Some observers have argued that these moderate referendums were intended as “killer initiatives,” placed on the ballot with the intention of drawing support away from the extreme initiative but with little real support from the legislature. See Dubin et al. 1992.
 Furthermore, restricting competing initiatives distorts the policy agenda. It ensures that the choice presented to voters is between only two stark policy alternatives, rather than several from which voters can select the most preferred (see Ferejohn 1995).
 One reason for debating reforms that the Court has ruled unconstitutional is the possibility that such decisions may be modified in the future.
 ACLF challenged three other provisions in the lower courts: (1) requiring circulators to be at least 18 years old; (2) limiting the circulation period to six months; and (3) requiring circulators to attach an affidavit to petitions containing the circulator’s name and address. The Tenth Circuit court upheld these provisions and they were not challenged in the Supreme Court (Buckley v. ACLF).
 However, one could argue that setting the political agenda by determining the content of initiative legislation is not much different than setting the political agenda by determining which measures get on the ballot (i.e., through ballot access).
 However, one could argue that setting the political agenda by determining the content of initiative legislation is not much different than setting the political agenda by determining which measures get on the ballot (i.e., through ballot access).