�鶹������ý

The distorting influence of money and concentrated wealth on American democracy is a pervasive problem, and the ballot initiative process is no exception. But the role money plays in direct democracy differs in significant ways from its role in candidate elections and legislative policymaking, and those differences matter for how reformers, funders, and policymakers should evaluate the initiative process.

Initiatives require substantial resources. Qualifying a measure for the ballot can be expensive, especially in large states or under restrictive qualification rules. Once qualified, campaigns for and against a measure often demand considerable spending on voter education, persuasion, and turnout.

This report explores both the costs and the value of statewide ballot initiatives in light of two key concerns that advocates of the initiative process routinely confront and that lawmakers have invoked in recent years to justify to it in their states. The first is that initiatives have become too expensive for grassroots efforts to access, allowing wealthy interests to dominate the process. The second is that initiatives invite even more money into a political system already saturated with it, potentially reinforcing problems of inequality and corruption rather than correcting them.

There is truth in both critiques. But cost alone is a poor metric for evaluating democratic institutions. The more relevant questions are what they cost relative to the available alternatives, and what they deliver in return and for whom. To answer these questions, we analyzed a decade of campaign finance data alongside the contemporary literature on money in direct democracy and interviewed ballot initiative practitioners.��

When examined in comparative context, against legislative lobbying, candidate campaigns, and the real-world constraints of contemporary policymaking, state ballot initiatives emerge as a high-cost, high-return investment for policy advocates as well as a critical democratic infrastructure in need of an upgrade. Initiative spending is unusually visible, bounded, policy-specific, and potentially educational. Costs are concentrated in a handful of states and campaigns. Money does not reliably buy outcomes. Perhaps most importantly, initiatives are often used only after persistent legislative inaction or obstruction. That is, the ballot initiative is sometimes the only avenue to achieve policy change when other representative channels are blocked.��

The evidence points toward a practical approach to redressing cost concerns in the initiative process. Under current Supreme Court jurisprudence, the kinds of contribution and spending limits that might most directly curb the influence of money in ballot campaigns are largely off the table. That leaves a narrower but still meaningful set of reforms to lower barriers to entry for genuine citizen-led campaigns and limit the extent to which financial resources impact participation and outcomes. These include relaxing qualification rules that drive up costs, strengthening voter information infrastructure, protecting statutory initiatives from immediate legislative override, calibrating disclosure requirements to campaign scale, and modernizing high-cost systems such as California’s so they are accessible beyond professionalized campaigns.��

Done well, this approach will improve the functioning of the process for grassroots campaigns and voters. It will also bring the initiative process closer to its original purpose of ensuring that citizens retain a viable mechanism to bypass wealthy interests and unresponsive institutions when necessary.

Ballot initiatives are costly. Qualifying a measure for the ballot, building a statewide campaign, and reaching millions of voters requires what can feel like obscene amounts of money. But how expensive are initiatives, really? We address this question by exploring the costs of signature gathering for qualifying initiatives and of campaigns for and against initiatives in even years from 2014 to 2024.¹ We recognize that focusing only on initiatives that successfully qualified for the ballot—and not on the many efforts that did not—presents methodological challenges. Initiatives that reach the ballot are likely to be better funded and organized from the start, and in more permissive institutional environments, than those that do not. As a result, our analysis likely underestimates the costs of attempting an initiative and may obscure how institutional barriers undercut campaigns before they become observable.

The most consequential expense in the initiative process occurs before voters even engage with a proposal. High qualification costs influence agenda control by determining which issues even reach the ballot. When ballot access depends heavily on fundraising capacity, well-resourced actors can gain disproportionate influence over which policy questions are posed to voters, while proposals lacking wealthy or institutional backing may never advance to public debate.��

Sponsors of initiatives typically use some combination of volunteers or paid circulators to collect voter signatures to qualify for the ballot. The signature-gathering process is a consistent and substantial expense, and one of the first barriers that sponsors of ballot measures encounter. As Figure 1 shows, in recent cycles, the average cost of gathering signatures represents about a fifth to a quarter of total spending in support of initiatives—a significant down payment that sponsors must make in the ballot initiative process.

These costs reflect specific process design choices. Signature thresholds, geographic distribution requirements, petitioner restrictions, and circulation timelines each affect the difficulty of qualification. In recent years, legislative changes in several states have raised qualification barriers. Arkansas, for example, increased the number of required counties for gathering signatures from 15 to 50, contributing to one of the highest cost-per-signature figures observed. Rules of this kind function as cost multipliers that disproportionately hurt grassroots campaigns.��

In addition to institutional design choices, the cost per signature can be influenced by the measure’s topic and popularity. For some measures, the heavy lifting is securing enough signatures to qualify, but once approved, the initiatives sell themselves to voters, and sponsors don’t spend much more to support them. For example, in 2024, about 70 percent of the money spent in support of Maine’s 2024 initiative to limit contributions to Super PACs went to gathering signatures to qualify the measure for the ballot, and only about a third went to other supporting costs. The measure passed with close to 75 percent of the vote.��

In other states, as shown in Figure 2, signature-gathering labor costs can be as low as zero because sponsors used volunteers to collect signatures, as they did in Idaho and South Dakota in 2024. However, no petition process is totally free: Even all-volunteer campaigns face administrative and operational expenses—including printing, materials, and voter data—that can be significant depending on the state and scale of the effort. The figure also shows that for the 2024 cycle in Maine and Oregon, most of the money spent in support of initiatives went toward signature gathering. Maine had only one initiative: the initiative limiting super PAC contributions mentioned earlier. Oregon had two. One was a failed initiative to enact a corporate tax revenue rebate for residents, for which supporters spent $702,676 to qualify the measure for the ballot out of a total of $873,369 spent in support of the measure (a total of $17,481,126 was spent to defeat it). The other was a successful initiative for the unionization of cannabis workers, for which supporting groups spent close to 60 percent of support resources in gathering signatures.

Figure 3, which includes data from 2012 to 2024, shows that signature-gathering costs vary dramatically by state. Signature-gathering costs have also risen sharply from a median of $314,789 in 2012 to $5,822,657 in 2024, and the range of spending on signature gathering has widened, driven by increased spending in a few states.��

On average, sponsors of initiatives in California and Florida spent more in absolute terms on signature gathering than sponsors in other states in most years. These averages hide variation in the number of measures in these states and in the amount spent to get them on the ballot. Given its large population and the number of initiatives that make it to the ballot, sponsors in California consistently outspend sponsors elsewhere in signature collection almost every year. In 2012, groups in California spent a total of $28,244,069 collecting signatures for 13 initiatives (an average of $2.17 million per measure); in 2024, they spent a total of $42,318,319 on five initiatives (an average of $8.46 million). In contrast, groups in Florida spent a total of $41,236,917 on signatures for just two measures in 2024 (an average of $20.62 million).

In other states, the average amount of money spent on signature gathering is orders of magnitude lower than in California and Florida. In Colorado, which had the most ballot measures in 2024, sponsoring groups spent an average of $1.49 million on signature gathering. Groups in Massachusetts spent an average of $1.29 million. In 2024, the lowest average amounts spent on signature-gathering labor were by groups in North Dakota ($70,051) and Alaska ($368,215), though the direct costs could be as low as zero when groups relied on volunteers, as in Idaho and South Dakota. Again, these low-budget, volunteer-run petition drives are rare.��

Another useful metric to compare the costs of collecting signatures—and one that gives a better sense of the distinct financial challenges of signature gathering across states—is Ballotpedia’s . This metric divides the total amount spent on signature collection by the signatures required to qualify a measure for the ballot. For example, sponsors of Arkansas’s Issue 2 in 2024 (for casino license regulations) needed to collect 90,704 signatures to qualify and spent $3,750,123, for a cost per required signature of $41.34. Meanwhile, in California, qualifying a measure on the ballot required 623,212 signatures in 2022. For Proposition 32 (on minimum wages and on the 2024 ballot), sponsors spent $8,837,160, for a cost per required signature of $14.18. The average cost per required signature inflates the cost per signature borne by campaigns as it divides by the minimum number of signatures required and not the total number of signatures a campaign actually obtained, which is usually substantially higher.

Figure 4 shows that the average cost per required signature per state has increased from 2012 through 2024. Unlike total and average spending, the states with higher average costs per signature are not California or Florida. In 2024, the highest average cost per required signature was in Arkansas ($41.34), followed by Montana ($27.42). In 2022, Arkansas also had the highest average cost per required signature ($25.28), followed by Missouri ($21.33). In 2020, Montana had the highest cost per required signature ($24.02), followed by South Dakota ($16.22). In Arkansas, the legislature has recently added restrictions and additional requirements to the signature-gathering process: In 2021, SB 614 added restrictions on the payment and recruitment of circulators; and in 2023, HB 1419 increased the number of counties from which petitioners need to gather signatures from 15 to 50. Laws like these add burdens to the initiative process, raising costs for sponsors.

Petitioning costs are often steep, but they are not inherent to direct democracy itself. Rather, these costs reflect the current practice of electoral politics in the United States. Signature-gathering costs are largely a product of unreasonable campaign finance laws, population size, the measure’s popularity, and process design choices: signature thresholds, geographic distribution requirements, and circulation timelines.��

Of these factors, we found that petitioning for constitutional amendments (rather than statutes or referendums), geographic requirements, and the state’s population significantly affect signature-gathering costs, as shown in Figure 5. Our analysis indicates that, controlling for year-specific effects, the cost per required signature for initiated constitutional amendments was, on average, $3.76 higher than for initiated statutes or veto referendums. The cost per required signature was also higher among measures subject to geographic requirements. Each percentage-point increase in the share of voters whose signatures are required is associated with an average increase of $0.66 in the cost per required signature. We also analyzed differences in the share of total support expenditures spent on signature gathering and found that, on average, it increases with population. To be sure, these findings reflect the costs associated with successful ballot qualification. Because this data excludes initiatives that failed to qualify or were deterred by high entry barriers, these figures may be subject to selection bias.

Overall campaign spending is even more concentrated than signature costs. A small number of initiatives in a few states—especially California and, more recently, Florida—account for a disproportionate share of ballot initiative expenditures nationwide. As shown in Figure 6, in general election years from 2014 through 2024, 16 of the 20 ballot measures with the most total expenditures were in California, two were in Florida, one was in Nevada, and one was in Michigan. California’s 2022 reported the most total spending in this time period by far ($403 million). It was followed by California’s 2020 ($223 million), and Florida’s 2024 ($185 million).

These campaigns fuel public perceptions of exorbitant spending on initiatives, but they are not representative of a typical campaign. Of the 318 individual ballot initiatives in our dataset, 29 measures (9 percent) had associated expenditures greater than $50 million, and 21 of these were in California and two in Florida.² Florida, however, is also a case of how process choices and regulations impose financial barriers to initiatives and prevent initiatives from even seeing the light of day: The state only allows initiated constitutional amendments, requires at least 60 percent of the vote to pass measures, and recently implemented restrictions on petitioning so severe that zero initiatives qualified for the ballot in 2026. This is an example where, because of restrictions on the initiative process, we do not observe how much money would be spent on possible ballot initiatives, but if a ballot initiative does qualify, it would likely require huge amounts of money.

Figure 7 shows the median amount of dollars spent on initiatives per year in each state. We report medians since massive spending on a few initiatives severely skews averages. The figure shows that in every even-numbered year from 2014 through 2024, California or Florida has the highest median spending on ballot initiatives, often by a large margin. Over this time period, median spending on initiatives across all states was $6.63 million, and this figure remained relatively stable between 2014 and 2024. In 2024, median overall spending on initiative campaigns was $10 million. Excluding California and Florida, it was $8 million.

Breaking down total expenditures into support and opposition provides a clearer view of how and when money is deployed within initiative campaigns. Figure 8 plots expenditures made for and against specific initiatives between 2014 and 2024. Across cycles, there is a consistent pattern: Far more money is spent supporting ballot initiatives than opposing them. In 2024, for example, groups spent a total of $900 million supporting initiatives, compared to $362 million in opposition.��

Part of the imbalance is structural. Proponents must bear the full lifecycle cost of an initiative campaign, whereas opposition campaigns are usually reactive. If opposition campaigns emerge at all, they do so later, once the measure has qualified and its likelihood of passage is clearer. This allows them to target spending more selectively and efficiently. But greater support spending can also reflect the types of policies that make it to the ballot. Many initiatives concern issues with broad support, such as minimum wage increases and expanding access to health care. These measures still often require substantial investment to qualify and communicate with voters, but they don’t always generate equally well-funded opposition. As a result, support spending can dominate total expenditures even when outcomes are not closely contested.

In the period from 2014 through 2024, the most expensive support campaigns were California’s 2020 Proposition 22 ($204 million, with sizeable donations from companies like Uber, DoorDash, and Lyft); California’s 2022 Proposition 27 ($164 million in support of legalizing sports betting); and Florida’s Amendment 3 to legalize marijuana in 2024 ($154 million). Meanwhile, the largest opposition campaigns were California’s 2024 Proposition 27 ($239 million) and, in a distant second place, California’s Proposition 61 in 2016 ($111 million, in opposition to the Drug Price Standards Initiative).��

Again, most campaigns fell well below topline spending levels. As shown in Table 1, of the 288 unique initiatives we analyzed (excluding initiatives for which we identified one committee contributing to multiple campaigns to avoid double-counting), 72 percent spent less than $10 million, 55 percent spent less than $5 million, and 23 percent spent less than $1 million. The comparable figures for opposition spending were 83 percent, 74 percent, and 57 percent, respectively.

These figures emphasize multiple recurring themes: Spending is geographically concentrated in a few high-profile campaigns, particularly in California and, recently, in Florida; and support spending usually exceeds opposition spending. For advocacy organizations, this data provides realistic benchmarks for estimating campaign costs and assessing the financial risk of pursuing an initiative strategy. For stakeholders in states without an initiative process, these numbers can help ground debates over whether to adopt a process in typical spending patterns rather than extreme outliers.

Every pathway to policy change in modern American politics costs money. Legislative advocacy calls for sustained lobbying, polling, communications, coalition management, and often multiple election cycles of candidate support with no guaranteed end point. Statewide candidate races in today’s nationalized and polarized political climate routinely cost tens of millions of dollars. Inaction has a price too: years of staff time and sunk advocacy costs that produce nothing.��

How do ballot measures compare to other policy change pathways? First, each pathway is structured differently, as shown in Table 2, so costs and returns are not directly comparable. A ballot measure affects only one policy at a time in a state, for example, while investing in a U.S. senator can lead to changes across multiple policy domains nationwide. A citizen-led ballot measure targets a specific, predefined policy outcome, whereas even a fully aligned senator or governor must negotiate with other actors, often resulting in legislation substantially different from what they started with. The . Ballot measure campaigns are usually subject to fewer restrictions because they are largely considered as . An individual donor or one company can bankroll an entire campaign for a ballot measure. In contrast, donations to candidate campaigns are subject to contribution limits and other restrictions to avoid corruption. This also probably means there is less dark money circulating around ballot measure campaigns, whereas the restrictions associated with candidate campaigns encourage workaround and subterfuge.

Despite these differences, comparison can help contextualize initiative campaign costs and help organizers within initiative states decide how to spend limited resources.

In Figure 9, we compare the average amount spent on ballot initiatives at the state level with the total amount of reported expenditures on U.S. Senate and gubernatorial races for states that had both ballot initiatives and either a U.S. Senate or gubernatorial race in the same year in recent cycles. To be sure, these are not straightforward comparisons: As explained before, money going to ballot initiatives works differently than money going to candidate campaigns. Still, these comparisons help contextualize the amounts spent on these different avenues for policy change.

Points below the dashed line correspond to elections where more was spent on candidate races than on the average ballot measures, and points above the line correspond to elections when spending on the average ballot measure was greater than on candidate races. Because there are often multiple measures on the ballot but only one race for governor or senator, we show the average amount spent on ballot measures in each state. As the figures show, more money is often spent on Senate or gubernatorial candidates than on the average ballot initiative. This is especially true for Senate contests, likely because their fundraising is more nationalized and the stakes are perceived as higher than in a state race. As usual, California and Florida are exceptions. In 2024, two ballot initiatives in Florida reported an average of $159 million in expenditures, compared to an estimated $95 million spent on the Senate race. In California, an average of $139 million was spent on six ballot measures, while an estimated $6 million was spent on the Senate election.��

As another point of comparison, John Matsusaka finds that, in California, far than on state legislative races. From 2000 to 2020, about $4.2 billion was spent on ballot measures compared to $1.5 billion spent on legislative races in the state. This difference likely reflects the potential policy impact of ballot initiatives but also the decline in electoral competition for legislative seats in California.

Figure 10 compares ballot measure spending with state-level lobbying expenditures in even-numbered years from 2014 to 2024, using data for the states that report it (Alaska, California, Colorado, Florida, Maine, Massachusetts, Michigan, Montana, Nebraska, Oregon, and Washington). Again, this is not a direct or straightforward comparison: Lobbying money is often spread across different policy issues, while initiative spending is concentrated on a few distinct policy proposals. Lobbying may take years of relationship-building and repeated legislative pushes to produce uncertain results. Ballot initiatives can secure a clear, yes-or-no outcome in a couple of years.

Of course, lobbying and citizen initiatives can be complementary, with lobbying helping to find allies in the legislature for specific ballot measures. For the most part, ballot measures and lobbying expenditures cluster around the equivalence line. The biggest exception is, as usual, California. In 2016, 2020, and 2022, California reported more spending on ballot measures than on state-level lobbying, but in 2014 and 2024, more was reported on lobbying than on ballot measure campaigns.

Critiques of ballot initiatives tend to rest on an implicit comparison between direct democracy and an idealized vision of representative democracy. In this idealized system, legislators are broadly responsive, policymaking is done in-house by lawmakers, and public preferences are translated into law through routine legislative functions. In practice, that is not the system policy advocates are operating within. On many high-salience or structural issues, legislatures are constrained by a cocktail of partisan incentives, capacity limitations, institutional self-interest, and procedural rules that make action extremely unlikely regardless of public support.��

The relevant comparison, then, is often not “initiative versus lobbying the legislature” but “initiative versus permanent inaction.” Sometimes, it’s lobbying and initiative. In the 26 states where citizens cannot place new laws or amendments on the ballot, less money is almost certainly spent on policy influence—but that cost savings is cold comfort for residents who are chronically underserved or ignored by their representatives. The initiative process is expensive. The alternative, for many advocates, is not a cheaper legislative path but no path at all.

Medicaid expansion is a case in point. While most Medicaid expansions under the Affordable Care Act occurred through ordinary legislation, the initiative process was decisive in several conservative states where legislatures had repeatedly refused to act, including Idaho, Missouri, Oklahoma, Nebraska, South Dakota, and Utah. Most tellingly, , eight lack any statewide initiative options (the other two, Wyoming and Florida, technically permit at least one form of statewide initiative, but their processes are so restrictive as to be functionally inaccessible).

In proposing that Wisconsin adopt an initiative process, he cited eight issues that polling shows enjoy overwhelming bipartisan support but continue to stall in the legislature, including Medicaid expansion, mandatory background checks for gun purchases, and redistricting reform. (This tracks with previous research findings by this report’s authors that Wisconsin has among the lowest policy congruence on social and economic issues of all non-initiative states.)

Against that backdrop, the cost comparison is not between expensive ballot initiatives and a cheap, frictionless legislative process, but between ballot initiatives and the real-world alternatives available when legislatures will not move. Initiative costs look different: They are often higher up front, but they are finite. They usually culminate in a clear, time-bound decision. Most importantly, they offer a chance of success when other avenues have repeatedly failed.��

This point is especially pronounced for structural reforms that by design threaten the interests of incumbent officeholders and institutions. The legislative pathway holds little promise for achieving significant electoral and good governance reforms. One strategist who specializes in ranked-choice voting reforms said that you usually assume a minimum of four cycles to get a reform bill through the legislature, which he estimates could run a minimum of $400,000. While that is less than the cost of a ballot initiative, he also observed that, despite years of lobbying statehouses, no legislature had yet passed a bill to adopt ranked-choice voting statewide, but two states had enacted the voting system via ballot initiative.

Similarly, after repeated legislative and partisan resistance to meaningful anti-gerrymandering reform, grassroots organizers in Michigan used the initiative process to create an independent redistricting commission. This case, as well as Maine’s multiple citizen-led measures to pass and defend ranked-choice voting, is emblematic of direct democracy’s power, as President Theodore Roosevelt put it, “.” 

To be clear, ballot initiatives are seldom the cheapest or best option for advocates. Practitioners attest that lobbying (and financing candidate campaigns) is a more cost-effective strategy, and one to be tried first or alongside initiative campaigns. The advocates behind many of the high-profile initiatives discussed in this report often pursued legislation repeatedly before turning to the ballot. Sometimes, though, the initiative is simply the only politically viable path to force an issue onto the legislative agenda or effect policy change—no matter how much money you’re prepared to throw at the effort.

A central fear about ballot initiatives is that well-funded interests can exploit the process to buy policy outcomes. The empirical record presents a more complicated picture.��

Across the initiative campaigns analyzed here, money is necessary for campaigns to compete, but they do not reliably determine outcomes. While there are several cases of monied groups spending excessively to influence the outcome of an initiative, there are also high-profile cases where money was not enough to overcome public opinion, particularly on issues that people care about. Finally, our analysis builds on previous studies that found well-funded actors are more influential when playing defense.

Figure 11 plots the difference in the amount spent in support and in opposition of a ballot initiative from 2013 through 2024 and the winning or losing margin of each initiative. If money were determinative, the figure would show all the points clustered in the upper right quadrant (indicating successful measures when more money was spent in support of them) or in the lower left quadrant (indicating failed measures when more money was spent against them). The data show a different pattern. While the majority of initiatives (69 percent) with more money in support than against passed—showing that money in support helps secure wins—almost a third of initiatives that received more money in support than against failed. Spending advantage matters, but it is neither necessary nor sufficient for victory.

Meanwhile, nearly 71 percent of initiatives that received more funding in opposition than support failed, while only 22 percent passed. This finding reinforces research by Elizabeth Gerber and John Matsusaka showing that money—especially from wealthy economic interests—is . In essence, campaign advertising is most effective when it raises doubt—reinforcing voters’ natural risk aversion and status quo bias—rather than building support. This tendency is more acute in .

Some of the initiatives that failed despite having more money supporting them than opposing them did so because of procedural barriers. For example, Florida’s two ballot initiatives in 2024—one to enshrine abortion rights and another to legalize marijuana—were defeated despite receiving millions more in support than in opposition and a majority of votes (with 57 percent and 56 percent, respectively), because Florida has a 60 percent threshold for initiatives to pass. Ohio’s 2024 ballot initiative, which would have established a nonpartisan citizen redistricting commission, received $35 million more in support than in opposition, but it failed, in part because of confusion introduced by the ballot summary language written and approved by Ohio’s secretary of state.

A lot, of course, has to do with whether what the initiatives propose is actually popular among voters: There is only so much money can do when the topic of the initiative is something voters really want or really dislike. Figure 12 shows the winning and losing margins of initiatives from 2013 to 2024 by topic and the amount spent in support of them. The figure indicates that there are some topics that are popular and tend to pass when put to a vote while others are unpopular and often fail.��

For example, abortion-related initiatives have been mostly successful. Of the 15 initiatives on the topic, nine were approved by voters, and only one of these was to restrict abortion access ( in 2024). Of the six that were not approved, three sought to restrict abortion, and three sought to enshrine abortion as a right (including Florida’s 2024 ballot initiative that received a majority of votes but didn’t reach the 60 percent threshold). Similarly, of the 20 initiatives related to labor rights or wages, 18 were approved by voters. The majority of these initiatives (17) sought increases to the minimum wage, adoption or expansion of paid sick leave, or stronger collective bargaining rights. Similarly, the majority of initiatives (65 percent) seeking to legalize or decriminalize drugs like marijuana (and in some cases psychedelics and psilocybin) were approved by voters.��

The lone successful initiative that favored corporations was California’s Proposition 22 in 2020 (proposing that app-based drivers be classified as contractors), for which supporters outspent opponents by roughly ten to one. Post-election polling and subsequent analysis suggested that many voters remained confused about the measure’s practical effects, a reminder to be extra cautious when overwhelming spending advantage collides with confusing or technical ballot language.

Some topics tend to be unpopular with voters, regardless of how much money is poured into the initiative. This is particularly the case for initiatives dealing with gambling, hunting, alcohol, and tobacco, as Figure 12 shows. There were 32 initiatives on these topics, of which only 12 were approved by voters. Many of the initiatives that voters did not approve had to do with increases in taxes on cigarettes or with the legalization or expansion of betting, gambling, or casinos. Of these, the most notable examples are California’s 2022 initiatives to legalize sports betting, Propositions 26 and 27. Proposition 26 won 33 percent of the vote despite supporters far outspending the opposition ($127 million in support versus $45 million in opposition). Proposition 27 won only 17 percent of the vote despite having $164 million in support, but it faced an opposition campaign that spent $239 million.

These numbers also demonstrate that the kinds of expensive campaigns that spark concerns about money in the initiative process are more likely when corporate interests in lucrative industries try to expand their markets or block taxes and regulations. Grassroots advocacy groups interested in protecting and expanding health care access (excluding reproductive health measures), economic justice, and civil rights are unlikely to need to raise and spend as much money as sports betting companies trying to legalize gambling, for example.

One way to think about this dynamic is that when a proposal is not popular to begin with, heavy spending alone cannot compensate for a lack of public support. On the other hand, when proposals are popular with voters, heavy spending is not necessary to win their support. This supports the idea that direct democracy serves as a path to enact policies that are broadly popular with the public but not elected officials, and it suggests that the ballot initiative process is not fully captured by deep-pocketed interests.

Modern ballot initiative campaigns are often portrayed as dominated by wealthy donors, corporate interests, and out-of-state money. That critique captures part of the story, but it misses important context and evidence. Following the money in ballot initiative campaigns reveals a more complex financing structure that combines broad-based individual participation with extremely concentrated funding at the top and how elite donors operate differently at different stages of the process. Finally, by the admittedly poor standards of American campaign finance, there is also an unusually high degree of transparency and legibility. Understanding how these structures and patterns interact is important for accurately diagnosing the capture problem and evaluating what reform can and cannot accomplish.

To investigate claims about elite capture of ballot initiatives, we analyzed to 667 ballot measures across 25 states from 2006 to 2022, which includes amounts donated, contributor type, and location, among other details. We find that funding of ballot measures is typically dominated by organized groups rather than individual donors. Non-individual contributors—such as corporations, labor unions, trade associations, PACs, and advocacy organizations—provided roughly 70 to 90 percent of total campaign funding in most cycles, as shown in Figure 13. And as Figure 14 illustrates, individuals made the highest number of contributors but they supplied a much smaller share of total dollars.

These patterns align with Schroder’s finding—published in the 2025 article, “”—that out-of-state donor participation is overwhelmingly small-dollar and widely distributed, but overall amounts come from a small share of large donors. Total contributions from out of state more than doubled between 2006 and 2022, rising from roughly $200 million to nearly $500 million per cycle. Among out-of-state donors, more than 86 percent of contributions are $100 or less. Yet donations above $10,000—less than 1 percent of total contributions—account for roughly 86 percent of total dollars. Schroder finds that contributions disproportionately come from ZIP codes with low-to-middle median incomes. Individual donors also tend to give more than once, often across multiple states, and to social or moral issues, particularly abortion and LGBTQ+ protections, substance use policy, and election reform. A sizable share of the out-of-state contributions comes from New York and Texas, states without initiative processes (see Table 3). These findings suggest that participation in out-of-state ballot initiatives is nationalized and, in some cases, that voters are using other states’ direct democracy processes as an outlet for political expression unavailable to them at home.

The surge in out-of-state donations cuts two ways. Initiative campaigns enable a distributed, issue-based engagement that crosses state lines and can boost progress on national issues by showing support in unlikely places. On the other hand, these numbers raise the question of whether residents of one state should be able to influence the policies of another. High-salience measures now routinely attract national donor networks, which sometimes finance a substantial share of campaign activity in states where these donors cannot vote. Perhaps cross-state diffusion effects and the idea of states as testing grounds for future national policy give voters in one state a legitimate stake in the outcome of an initiative campaign in another.

At the same time, it would be a mistake to overcorrect and treat initiative campaigns as out-of-state-driven affairs. In most cases, out-of-state donors are not displacing local participation. In-state donors still make up a majority of donors, and the campaigns themselves remain anchored in state-level political contexts. Figure 15 shows that the majority of contributions to state initiatives originate in-state. Over nine election cycles examined, contributions from in-state individuals trumped those from out-of-state individuals in every cycle but one, 2022. Nevertheless, state legislatures have recently moved to restrict outside participation in initiative campaigns by requiring signature gatherers to be in-state residents or imposing extra compliance burdens on nonresident circulators.

Despite a broad base of individual small donors, the funding that makes initiative campaigns viable in the first place remains heavily concentrated at the top with large individual donors and organized advocacy groups. This is especially true in states where qualification costs are high. (Economically driven interest groups such as corporations and unions do spend heavily on campaigns, but because they are more likely to fund opposition campaigns, their influence often begins after the qualification stage.) 

Within the category of ultra-rich individual donors, there are several visible cases where the critique of elite influence lands cleanly. Among individual contributors to initiatives, Michael Bloomberg holds the record for most money given, having contributed nearly $46 million to California’s Proposition 31 (a successful veto referendum to uphold a ban on flavored tobacco products) in 2022. Bloomberg’s contribution represented 92 percent of the total amount spent. Charles Munger’s $10 million contribution represented 98 percent of the total contributed to California’s Proposition 54 in 2016, which succeeded in requiring the public display of legislative bills prior to a vote. Henry Nicholas was responsible for 96 percent of the total amount contributed to Ohio’s Issue 1 in 2017 to secure rights for victims of crimes. Of course, many individuals channel their donations through political organizations, like Brian Heywood, who has , mostly through a political action committee. Regardless of whether these causes reflect , when rich individuals exploit the initiative process to set the agenda, it weakens the democratic value and credibility of initiatives as the “people’s tool.” 

Crucially, the presence of large donors does not mean they are consistently getting what they want. Research by John Matsusaka provides a critical corrective. In , Matsusaka analyzed over 2,600 state initiatives and found that only 4 percent of successful initiatives benefited business interests, compared to 16 percent of legislatively referred measures. Matsusaka’s of whether ballot initiatives are good for business and organized labor found that, of all business-related state-level initiatives from 1904 to 2021, anti-business initiatives were significantly more common (56 percent) than pro-business initiatives (37 percent), and anti-business initiatives were more than twice as likely to pass. When comparing the subset of measures that were constitutional amendments—which can be placed on the ballot by the legislature or citizens—Matsusaka found that legislatively referred amendments were pro-business by roughly a three-to-one ratio, while citizen-initiated amendments were about evenly split between pro- and anti-business.��

Zooming in on contributions to California ballot measures between 2000 and 2020, initiatives are systematically less favorable to business interests than laws produced through legislatures. In addition, business groups directed about , compared to less than 1 percent to opposing legislatively referred propositions; unions show a similar shift, with a majority of spending devoted to defense. In dollar terms, business groups have spent nearly twice as much opposing initiatives as supporting them.

These patterns indicate that the role of money in initiatives differs from the standard model of a system captured by economic interests. Those organized players remain major funders, but much of their spending is reactive, aimed at blocking proposals rather than advancing them. By contrast, two-thirds of contributions by wealthy individuals in Matsusaka’s dataset were made in support of measures. The result is a system in which funding is concentrated, but results are not consistently aligned with the preferences of big corporations and financial elites. Matsusaka concludes that “nothing akin to capture emerges when citizens make the laws; if anything, the situation is the reverse—citizens use the process to capture benefits from businesses through taxes or regulations that transfer rents from companies to taxpayers, consumers, or the general public.” Matsusaka theorizes that organized interest groups are less effective at capturing the ballot initiative process than legislative politics because channels like contributing to “reelection campaigns, lobbying, and revolving doors [are not] available when an issue is decided by a public vote. The only lever available in ballot proposition elections is mass-media campaign spending, which seems to be a tool with limited effectiveness.”

Taken together, the data suggests that direct democracy does not eliminate elite influence, as the Progressives who first championed initiatives in the early 1900s had hoped, but it does redistribute it away from the interest-group capture of representative democracy toward episodic agenda-setting by well-resourced and ideologically-driven actors. Initiative campaigns do rely on concentrated funding, and wealthy donors can determine which issues reach the ballot. But once measures qualify, organized economic interests are more often defending against proposals than securing favorable outcomes, and voters frequently act independently of funding advantages.��

This distinction matters for reform. If the primary leverage point is at the start-up stage, then efforts to broaden access and reduce entry costs for grassroots campaigns may do more to counterbalance concentrated funding than attempts to regulate campaign spending alone. And given the broader legal environment, where limits on political contributions and spending remain constrained, restricting direct democracy on the grounds that wealthy actors participate risks closing off one of the few arenas where voters can directly override entrenched interests.

“If we don’t let ourselves have nice things because people we don’t like might also benefit from [the process], then we’ll never have nice things,” one expert told us. Wealthy actors will continue to spend in politics across all institutional settings. The relevant question is whether to maintain mechanisms, however imperfect, that also allow the public to translate preferences into policy.

Ballot initiative spending differs from other forms of political spending in structure and legibility. In traditional legislative policymaking, influence is fragmented across years of candidate campaigns, lobbying, and informal relationships. Campaign donations, lobbying activity, and coalition-building may shape policy over time, but the causal chain is often opaque. Even when spending is disclosed, the motivations behind it—and its ultimate policy impact—are not always clear. As a result, the total cost of influencing legislation is difficult to measure, and attribution is murky.

Initiative spending, by comparison, is time-bound and legible. Donors fund signature gathering, advertising, litigation, and voter outreach over a defined period, and these expenditures are generally disclosed in ways that clearly link donors to advocacy around specific policy campaigns.

This relative clarity gives initiative spending a practical advantage for funders and organizers, who can plan around predictable timelines and budgets. It also provides a degree of transparency that is less common in other domains of political spending. Still, concerns about the cost and capture of direct democracy have prompted states such as California, Washington, and Alaska to adopt stronger initiative campaign finance disclosure rules. These include prominent “top-donor” disclaimers on advertisements, rapid reporting requirements, true-source disclosures designed to prevent donors from hiding behind nonprofits, and requirements that petitioners identify the measure’s primary funders during signature gathering.

Disclosure is the only available tool for states to regulate campaign finance in ballot campaigns. As mentioned earlier, the Supreme Court has ruled against contribution and spending limits on the grounds that ballot campaigns are political speech. Enhanced disclosure rules make it easier for journalists, opposing campaigns, and regular voters to identify sources of funding for initiative campaigns. But to what extent does disclosure impact donor behavior and overall campaign spending?

There is no strong evidence that disclosure reduces overall spending or deters most major donors from participating. Initiative spending has risen sharply in the years following the implementation of more stringent transparency measures. Some reputation-sensitive donors have probably adapted by shifting contribution channels or strategies. Yet there is credible evidence that disclosure deters some donors at the margin, particularly in certain policy and geographic contexts.��

For example, in larger states, where most ballot campaign funding is heavily concentrated, initiative campaigns are already highly visible, often nationalized, and professionalized. The policy stakes tend to be higher. Corporations, industry groups, and wealthy individuals from both in state and out routinely participate with the expectation that their role will be public. In this setting, disclosure affects how money is interpreted more than whether it enters the campaign.

In smaller states, enhanced disclosure may have more bite. Where political and economic networks are smaller and more localized, in-state donors are more likely to be directly connected to the communities affected by the initiative. Identification may carry reputational and relational costs that outweigh the potential policy gain. Outside spending is also more likely to be scrutinized and stigmatized, making big national or out-of-state donors hesitant to engage. In some cases, that visibility might even undermine the campaign. One grassroots organizer suggested their state’s disclosure laws are at least somewhat effective in deterring big out-of-state donors. In their experience, disclosure concerns have been a deal-breaker for multiple would-be donors. One donor even made repealing the disclosure law a condition of future support for their organization’s initiative efforts.��

Of course, if the measure is ideologically or commercially important to even a few deep-pocketed interests, money is going to flow regardless of state size or disclosure regime. After California strengthened top-donor formatting requirements, online ad coverage, and original-donor disclosure rules, ballot measure spending has continued to break spending records thanks to a handful of expensive campaigns. Washington has one of the nation’s strongest ballot measure disclosure systems, including top-donor identification on ads, and still produces some of the country’s most expensive and donor-driven initiative cycles. Alaska added true-source disclosure rules in 2020 and major money continued to flow—the 2024 campaign to defend top-four voting reportedly raised more than $15 million, much of it from out of state.

The literature on the effects of campaign finance disclosure laws on donor behavior is sparse. However, there is some indirect and revealed-preference evidence suggesting marginal deterrent effects on donations. Donors rarely admit to being afraid of disclosure, but consistent patterns of litigation following the passage of new transparency laws—particularly top-donor ad disclaimers and true-source disclosures that complicate hidden pass-through giving—combined with evasive structuring of contributions, suggest that there are plenty of anonymity-seeking or -demanding donors. Alaska is the clearest recent example: Opponents litigated true-source and disclaimer rules through multiple rounds in federal and state courts and ultimately lost. Maine offers another case: During the same-sex marriage referendum fights, national advocacy groups fought donor-disclosure requirements in court, implying that public identification was viewed as costly or .��

By comparison, multiple recent studies have sought to understand whether disclosure helps voters make more informed decisions about ballot measures. Because ballot initiative contests lack party labels and incumbent or candidate performance cues, voters have to rely on other heuristics to interpret complex proposals. In this context, campaigns often spend heavily on ads and messaging, . This is where disclosure can serve as a much-needed informational shortcut, and, if presented optimally to voters, could disincentivize expensive misinformation campaigns.

Research on Washington State’s Initiative 1634—a 2018 measure backed by the soft drink industry to preempt local beverage taxes—found that revealing top donors shifted voter attitudes when paired with explanations that connected donors directly to the policy’s purpose. Similarly, suggests that donor information can shift voter opinions at levels comparable to party cues or policy information, but its effects are limited to more informed voters, who are better able to interpret donor interest. A different study found that who supports and opposes a ballot measure, and that this endorsement cue, rather than information about the measure’s policy substance, .��

That insight further explains why top-donor disclaimers on ads are more valuable to voters than searchable databases. Database disclosure is passive: The information exists, but voters must go look for it, and most never will. Top-donor ad disclaimers convert that into active disclosure by placing funder identities directly on television, radio, mail, and digital ads. Donor identities effectively function as endorsements; endorsements guide voters’ decisions in the absence of party labels and other heuristics that we know influence candidate selections. Washington’s requirement that political committees list their top five contributors on ads and California’s DISCLOSE framework are significant because they make donor identity part of the campaign message itself. That increases reputational risk, especially for corporations and trade groups with public-facing brands. Adding top-donor tables to official voter guides could serve a similar function, particularly in all-mail states—including Washington, Oregon, and Colorado, which are also some of the most active initiative states—where voters complete ballots at home with the official guide nearby. In states where voters go to the polls and the guide sits unopened on the kitchen counter, its informational value would be considerably lower.

Unfortunately, disclosure systems may be more effective at undermining grassroots-led initiative campaigns than highly professionalized ones. The four most active initiative states west of the Rockies, plus Idaho and Alaska, which recently adopted some of the country’s most aggressive transparency rules, illustrate how much compliance burdens vary in practice. (See the Appendix for more details.)

States such as California prioritize expansive disclosure through low committee thresholds and early itemization requirements: Ballot-measure committees generally qualify at $2,000—pocket change for state ballot campaigns that run an average of $8 million just for signature gathering—while contributions of $100 or more must be itemized, alongside rapid pre-election reporting and prominent top-donor ad disclaimers. That creates a rich public record, but it also means that modest grassroots campaigns can be drawn into full compliance regimes at a very early stage.

Idaho sits at the lighter-touch end of the spectrum, imposing fewer burdens on campaigns generally but also providing voters with less timely and less intuitive information about who is financing ballot campaigns. Again, because Idaho is a small pond, weaker disclosure rules might be offset by stronger incentives for journalists and other campaign stakeholders to identify and expose major sponsors of ballot campaigns.

Disclosure costs are not borne evenly across campaign types. A committee backed by a single corporation or trade association that writes a seven-figure check may generate relatively little reporting complexity. A citizen-led campaign raising the same amount through thousands of $25 or $50 donations can face far greater burdens in data collection, treasurer capacity, software, legal compliance, and other risks. Low thresholds can therefore operate regressively, advantaging concentrated wealth over distributed participation. The most efficient systems likely combine light-touch treatment for genuinely small or volunteer-driven efforts with strong transparency for large donors and voter-facing advertising. Top-donor disclaimers on ads, true-source tracing for intermediary entities, and rapid reporting of major late money likely generate more value to voters than forcing grassroots, small-donor-funded campaigns into dense reporting systems that few ordinary voters will ever consult.

A final implication is that disclosure, while valuable, does not fully resolve the deeper capture problem. Intense transparency can expose who is trying to buy influence, but it does not necessarily prevent dominant economic actors from overwhelming the field. In Montana, organizers advanced a 2026 initiative, the “Montana Plan,” that would prohibit corporations and other artificial entities from spending money on candidates or ballot measures, explicitly attempting to challenge or route around Citizens United.

Rising costs and the dominance of money in the initiative process are a function of identifiable political and legal trends and institutional choices, not an inherent flaw of direct democracy. Since 2010, growing partisan polarization, the proliferation of state-level trifectas, and increased gerrymandering have all . More recently, in the wake of several successful progressive initiatives in conservative states, many state legislatures have imposed new qualification and petition rules that further raise the cost of qualifying for the ballot while increasing the rate at which they .

Many of these purported “reforms” are process killers masquerading as process improvements. Higher signature thresholds, expanded geographic distribution requirements, shortened circulation windows, burdensome and expensive new petition regulations, and often increase costs while doing little to improve integrity, homegrown authenticity, or policy quality. In practice, such measures disproportionately hurt the volunteer-driven and grassroots campaigns that the initiative was originally created to empower. Any reform agenda should start by avoiding such counterproductive changes. Beyond that baseline, our work identifies several opportunities for reform.

Most states with initiative processes adopted them long before the Supreme Court ruled against capping contributions and expenditures for ballot campaigns. As long as those rulings stand, money will remain a dominant feature of direct democracy. Rather than bury our heads in the sand, it is better to adopt new processes to ensure that money cannot easily substitute for public support.��

The evidence presented in this report suggests, for example, that the initiative process’s principal vulnerability to concentrated wealth lies more in ballot access than in Election Day persuasion. Reform efforts focused exclusively on campaign spending thus risk missing the more consequential ways in which money sets the agenda and structures participation.��

For these reasons, the appropriate response to concerns about money in direct democracy is to level the playing field for grassroots, volunteer-driven efforts rather than abandoning the tool.��

One of the most direct ways to reduce costs and boost agenda-setting power for authentic people-driven initiative campaigns is to revisit qualification rules that function as cost multipliers. Signature thresholds and geographic distribution requirements in particular materially affect the cost of qualification. Modest adjustments to these rules can broaden participation without eliminating legitimate safeguards.��

Additional reforms may further facilitate greater grassroots participation relative to elites and provide helpful informational cues to voters about the motivations of initiative sponsors. As and other direct democracy scholars have argued, public financing or in-kind support for grassroots qualification efforts—such as subsidized signature verification or matching funds for small-dollar initiative campaigns—warrant consideration, though such methods remain underexplored in the United States. Technological reforms could likewise lower costs over time. Secure digital signature collection, administered by a nonpartisan authority, could significantly reduce qualification burdens. Expanded translation and language-access support may further improve participation.

Indirect initiative pathways can also reduce costs as legislatures can adopt proposed measures before they reach the ballot, allowing some campaigns to avoid the far greater expense of full election contests. Across the nine states with an indirect initiative process, 11.5 percent of qualified indirect initiatives since 2014 were acted on by the legislature instead of going to the ballot. (It should be noted, however, that some of these measures were accepted by the legislature only to be altered by the legislature or overturned later via veto referendum.)

Many indirect initiative states also allow some combination of direct initiated statutes and indirect or direct initiated constitutional amendments. To incentivize grassroots campaigns in these states to make greater use of the potentially less expensive indirect statutory route, states that don’t currently protect indirect initiatives from immediate legislative alteration, or from being , should add such protections. In the smaller subset of states with both direct and indirect statutory initiative pathways, reformers should also experiment with setting lower signature requirements for qualifying indirect initiatives for the ballot.

States should also avoid unnecessarily pushing campaigns toward constitutional amendment pathways when statutory routes would otherwise suffice. As discussed earlier in this report, constitutional amendment campaigns are substantially more expensive to qualify than statutory initiatives. Yet where legislatures may immediately amend or repeal initiated statutes, organizers often rationally choose the more expensive constitutional route to secure policy durability.

A more balanced system would preserve the relative accessibility of statutory initiatives while providing greater assurance that voter-enacted laws will receive meaningful protection. More states should require initiated statutes to remain in force for a defined cooling-off period (two years, for example) before legislative amendment, or permit modification only through bipartisan or supermajority legislative approval. Such measures reduce incentives to constitutionalize ordinary policy contests while preserving legislative flexibility over the longer term.

States can also strengthen the informational environment in which initiatives are considered. Sustainably funded Citizens’ Initiative Reviews and voter guides that include independent fiscal and constitutional reviews can reduce voters’ reliance on paid campaign messaging. In a of California residents, 77 percent of respondents in California reported favoring an independent citizens’ initiative commission that would hold public hearings on initiatives and then make ballot recommendations in the official voter guide. More ambitious proposals would move deliberation upstream by empowered to review or advance proposals before they reach the ballot. While such approaches remain largely theoretical and would require substantial experimentation, they illustrate one possible path toward reducing the role of money in agenda-setting without eliminating citizen-led policymaking.

Disclosure rules remain the only available method of campaign finance regulation in initiative campaigns. While stricter disclosure regimes have failed to shrink the overall scale of initiative campaigns, it is hard to argue against rules that protect the integrity of the process. Certain disclosure requirements make hidden channels riskier, make donor identities more salient to voters, and probably deter some reputation-sensitive contributors at the margin.��

However, disclosure systems should be calibrated carefully. Overly complex or labor-intensive campaign finance reporting requirements can impose disproportionate burdens on leaner, volunteer-run campaigns already struggling to compete in a highly professionalized environment. Disclosure systems that preserve robust transparency for large donors and professionalized campaigns while simplifying compliance for smaller efforts can strike a better balance.��

Top-donor disclaimers and true-source reporting appear to offer the greatest promise for improving voter decision-making, particularly when donor identity is visible at key decision points. If voters can readily infer the interests and motivations behind a campaign, disclosure may help reduce reliance on costly and misleading advertising. This could lower overall spending in the persuasion phase—the most expensive phase of the campaign, as shown above—and discourage campaigns that depend on voter confusion or misalignment with underlying preferences. However, these effects are not well understood. More research is needed to determine which formats, delivery mechanisms, and messaging strategies enable disclosure to serve as an effective signal of donor intent, especially to counter high-cost campaign tactics that can lead voters to support outcomes inconsistent with their own preferences.

More structural reforms may be necessary in especially high-cost environments. Again, California offers a useful stress test because it combines high usage, high cost, and heavy professionalization in a single system.��

Notably, even in California—a cautionary tale for most critics of modern direct democracy—voters remain broadly supportive of the initiative process despite expressing significant concerns about its operation. , 67 percent of California voters said it’s a good thing citizens can make laws via initiative, and 63 percent reported being at least somewhat satisfied with how the process is working. At the same time, the vast majority of respondents said the process is influenced by special interests and requires at least “minor changes.” 

The state’s crowded ballots demonstrate that initiatives remain formally accessible, but in practice that access is largely limited to well-funded campaigns. One alternative approach for California and other high-cost qualification states would be a differentiated, two-track qualification system separating professionalized campaigns from genuinely citizen-led efforts. California’s current rules require hundreds of thousands of signatures within a relatively short 180-day period, effectively forcing campaigns to rely on paid signature-gathering firms. Under a two-track model, campaigns that use paid circulators or exceed certain fundraising thresholds could face higher qualification requirements or disclosure obligations, while volunteer-driven campaigns could qualify under more accessible rules, such as lower thresholds, longer timelines, and reduced compliance burdens.

Reforms to ballot access and initiative durability, however, do not fully address what initiatives are being used to decide. California also illustrates the governance challenges that arise when initiatives are used to resolve highly technical fiscal and budgetary questions. in 1978 locked in tax and budget rules that were shifting more fiscal decisions onto the ballot and contributing .

For this reason, subject-matter restrictions deserve greater consideration from pro–direct democracy reformers. While broad exclusions risk neutering the process (as in Illinois), restrictions on technical tax and budget rules could simultaneously mitigate some of the problems facing direct democracy and our representative institutions. Reserving certain fiscal matters for legislative negotiation and decision could reduce the kinds of confusing technical ballot questions that make voters most susceptible to manipulative advertising, lower administrative costs, and strengthen legislative accountability by ensuring elected officials bear responsibility for ongoing fiscal tradeoffs and votes.

The challenge for reformers, whether they’re seeking to proactively strengthen existing initiative processes or adopt the initiative in their state for the first time, is to ensure that processes are designed for contemporary political realities: accessible enough to permit meaningful citizen participation, structured enough to maintain long-term legitimacy, and calibrated to prevent wealthy interests and individuals from determining whose ideas are heard or what measures pass or fail.

For more detail on disclosure rules, including reporting triggers, itemization minimums, accommodations, and requirements, see Table A1.