Methods

Ballot measure language and arguments, campaign themes and elections outcome were collected from state voter pamphlets, secretaries of state, academic8 34 37 41 44 48 51 53 and advocacy4 reports and newspaper archives (http://infoweb.newsbank.com). To characterise tobacco industry strategies, we searched tobacco industry documents in the UCSF Legacy Tobacco Documents Library (http://legacy.library.ucsf.edu) dated between 1986 and 2008. Initial search terms were “ballot measure”, “initiative”, “tobacco tax”, “excise tax” and specific initiative titles and numbers for a particular year (for example, “Proposition 10” or “Prop 10” for California in 1998). Surrounding Bates numbers were searched for useful documents as well as specific names of individuals or groups that were identified (such as “Ballot Issues Committee” or “initiative strategy team”) using standard snowball search techniques.54 We used public and private polls (obtained from proponents of the ballot measures, pollster websites or the tobacco industry documents) to identify early levels of support for the tax proposals and how this support compared with the final election outcomes. We divided the campaigns into four categories based on whether or not there was a substantial opposition campaign and whether voters passed or defeated the ballot measures, then looked for consistent themes.

Results

There was no clear trend suggesting a certain tax increase amount could predict the outcome of a campaign, including whether the total tax would have exceeded that of surrounding states.

Changing tobacco industry strategies on tobacco tax ballot measures

Between 1988 and 1998 the tobacco industry mounted extensive opposition campaigns to all nine tobacco tax ballot measures, but only defeated four (44%) of them21 24 25 29 34 55; from 1998 to 2008 it only challenged five out of 13 of the tobacco tax measures and defeated four out of five (80%) (table 1). A March 1997 industry memo, “Tobacco Industry Ballot Issues Goals and Procedures” stated “there is no reasonable expectation of the lessening frequency of these battles nor can there be any assumption that our chances for victory will be increased due to attitudinal changes in the electorate”.56 A 1997 draft script for a Philip Morris Issues Management presentation listed the traditional reasons for raising cigarette excise taxes—economic (the need for revenue), social (user fees, luxury taxes), political (smokers make up a minority of votes) and personal (special interests)—but noted that “over the last two years, these traditional reasons for excise tax increases are becoming supplemented by new—more difficult to fight—justifications”,57 including cigarette tax increases as a deterrent to consumption (particularly youth), a revenue substitute for other types of more unpopular taxes and “Taking on the Evil Tobacco Industry”.57

By 1997 the Tobacco Institute (the tobacco industry’s political and lobbying arm until 1998, when it was dissolved as a result of litigation by state attorneys general) had organised its Tobacco Industry Ballot Issues Committee, which provided “oversight of all prevention, preparation, and execution of tactics relating to ballot issues, as well as the place to discuss new strategies and technologies related to battles in this arena”.56 The committee suggested that the industry work to “prevent” these initiatives by, among other things, lobbying legislatures to “reform initiative and referendum laws to make qualification of ballot issues more difficult”, “encourage[ing] third party ballot issues which threaten our opposition and impede their progress”, mounting “legal challenges which complicate opposition progress” and “preventing the opposition from using inappropriate funding sources for their political activities”.56 They also increased their knowledge of the initiative process, conducted public relations campaigns, conducted benchmark research and built partnerships with other organisations before they would be needed in a campaign.

After its 1996 loss in Oregon, the Tobacco Institute commissioned extensive market research to improve the voter segmentation model used to assess the chance of defeating tobacco tax initiatives.58 (Voter segmentation models are based on similar techniques from marketing research to divide potential customers into different “segments” based on their attitudes, beliefs, and needs, which is then used to develop specific marketing messages to appeal to people within each segment.) This 1997 national survey58 explored a wide range of attitudes towards smoking, government regulation, tobacco control policies (not just taxes) and the tobacco industry, as well as personal attitudes and characteristics (including parenthood). Agreement with two statements improved the model’s ability to predict election outcome: “Even though I don’t smoke, I am pretty much opposed to all tax increases, so I oppose cigarette taxes as well”, and “Although I don’t like smoking, some of the proposals to reduce smoking just seem to be going too far”.58 The analysts concluded, “Of these two, the ‘I hate all taxes’ item did the better job of differentiating between the ‘harder core’ and ‘not so hard core’ antis,” which led to the algorithm to identify voter segments and the share of each segment that the industry needed to win to prevail in a campaign (table 2).58

Another strategy shift came after a 1998 loss in California. An industry poll found that 48% of the voters had made up their minds three weeks before the election, showing the value of early campaigning.59 The Tobacco Institute’s Initiative Strategy Team also realised in 1999 that “early involvement by the industry in states with threats has had an impact on the decisions by tobacco initiative proponents [emphasis in original]”.60 Similarly, a July 2001 focus group report to Philip Morris concluded that early opposition would be a crucial factor in the outcome of Washington state’s upcoming Measure 773:

The good news: voters can be persuaded to our side of this argument. The bad news: their strong, first inclination is to vote against us… It will be strategically imperative to frame this debate at the earliest possible junction: “folks, this is NOT about health care…its [sic] about more money—BIG money—for the HMOs.” If we can credibly take their best arguments away early, if we can put them on the defensive early…we’re in the ballgame. If not, it will be uphill all the way.61 [emphasis in original]

The report also noted that having many messages—for example, anti-tax, HMO get-rich schemes, distrust of government—would have a cumulative effect and distract the voter from the initiative’s general concept, which voters tended to support.61

A third strategy decision came after Measure 773 passed with 66.1% of the vote. The Associated Press reported that the industry did not oppose the initiative because they “didn’t think it was winnable” and because of the “animosity toward tobacco companies”.62 A Seattle Times story quoting a representative of a national anti-tobacco organisation saying, “I know several states right now that would jump on this [tobacco tax] if they saw that they weren’t going to get any opposition”63 caught the attention of a Philip Morris lawyer heavily involved in policy matters, who distributed it two days later with the comment: “The article below suggests that the industry’s lack of financial support for fighting the Washington tax may encourage other states to move forward with excise tax increases.”64 The industry would pass on the next five tobacco tax ballot measures in 2002 (Arizona, Oregon and Missouri) and 2004 (Colorado and Montana). The industry fought and lost in Oklahoma in 2004, but won the next three it opposed in 2006 (California and Missouri) and 2007 (Oregon).

The industry’s increasingly sophisticated polling and voting models, and increasing awareness of factors like early involvement and the state’s political environment and budget issues, may underlie the shift in the industry’s opposition strategy in the 2000s. Whereas the industry often mounted multimillion dollar campaigns, it became much more selective in its opposition to tobacco tax measures beginning in 2001. In addition to helping public health advocates understand and predict when tobacco companies will oppose initiative campaigns, this voter segmentation could help tobacco control advocates assess their chances of winning.

Key elements of individual campaigns

Nicholl21 provides brief summaries of the salient qualitative features of eight of the early campaigns. We prepared case studies of the remaining 14 campaigns. Table 1 summarises the key elements of these campaigns, including the tax proposals, key campaign messages, campaign budgets, initial levels of support and outcomes. The appendix summarises important qualitative aspects of the campaigns, including the political alliances that came together to work to pass the tax, endorsement or opposition of key political leaders, other issues on the same ballot (sometimes related to tobacco, such as a strong proposal for smoke-free environments in Arizona, sometimes unrelated, such as funding for stem cell research in Missouri, both in 2006) and the broader political environment. These details, while hard to quantify, can exert substantial influences on the outcomes.

Base level of public support

Beginning with Washington’s Measure 773 in 2001, the tobacco industry did not contest every tax initiative; all seven that were unopposed between 2001 and 2008 passed (table 1). Strong and early public support for tobacco taxes may have dissuaded the industry from wasting money on opposition campaigns.

Support for tobacco tax initiatives almost always drops as the campaign progresses (table 1). Compared with the first polling conducted a median of 8.0 months (interquartile range (IQR) 4.0–13.0) before the election, the percentage of voters voting Yes fell by a median of 12.5% (5.0%–17.3% drop). The drop in support was larger, 15.0% (10.0%–20.0%) when there was active opposition to the initiative than when there was not (4.0%, 1.0% gain in support to 6.0% drop). Winning campaigns started with a median of 67.5% (64.5%–70.0%) supporting the proposed tax, compared to 59.0% (40.0%–66.5%) in campaigns that went on to lose. Detailed data from early polls that distinguished between “strong support” and “support” were available for 13 elections; the nine for which initial “strong support” was at least 49% all won, while the four that were 44% or lower lost.

Discussion

Faced with hostile state legislatures, advocates for tobacco tax increases, like others seeking to enact tax policies that were blocked in the legislature,10 have sought to shift the venue9 and use ballot measure processes to enact tobacco tax increases. Previous studies,65 66 67 tobacco industry research and our work suggest that tobacco tax initiatives have more successful outcomes when there are substantial revenue allocations to tobacco control efforts. The persuasiveness of an argument criticising a measure for not giving enough tobacco taxes to tobacco control was identified by the tobacco industry’s research. Following the tobacco industry’s loss in the 1988 California Proposition 99 campaign, RJ Reynolds commissioned a post-election survey that concluded, “A theme that still has potential usefulness was ‘only a small amount of the tax increase will go to education about the hazards of smoking.’ Among the general public 55% agreed, 19% disagreed, and 27% had no opinion. However among proponents 44% agreed, 23% disagreed and 34% had no opinion. Fully one-third were available to be persuaded on this issue”.68 Ten years later, in a post-election survey conducted for the tobacco industry after its narrow loss in the 1998 California Proposition 10 election, 55% of those surveyed agreed, 28% disagreed and only 8% had no opinion on the statement: “It is not fair to single out smokers to pay for programs that are everyone’s responsibility and have nothing to do with smoking”.59 Proposition 99 (some money for tobacco control) won with a 58% Yes vote, and Proposition 10 (no money for tobacco control) squeaked by with a 50.4% Yes vote.

The tobacco industry also identified themes of “anti-tax/HMO scheme” and “what they aren’t telling you” as effective arguments against ballot measures.61 Our findings are consistent with the industry’s results and suggest that advocates need to be careful in choosing their stakeholders and partners. The involvement of hospital associations and HMOs in several measures proved to be fatal moves that were exploited by the industry. The tobacco industry also learned to combine the argument that the tax would primarily benefit hospitals and HMOs with lack of funding for benefits for smokers. When either of the two arguments was used alone, the tobacco industry lost three out of four elections, but when they were combined, they won three out of four elections.

One reason that health groups partner with hospital and physician groups has been the hope that these wealthy interests would provide substantial resources for the campaign (for example, California’s Proposition 86 in 2006; see appendix for details). In hindsight, Jack Nicholl, a political consultant to the voluntary health agencies on the campaign, opined in 200748 69 that the decision to work with the hospitals was a mistake and that they could have won their original proposed $1.50 tax increase by modestly increasing their campaign funding to $5 million and avoiding the hospitals’ baggage and for a substantially higher tax to get more money for their programmes.

A common characteristic of losing campaigns was the failure to effectively mobilise broad grassroots support. Support for a tobacco tax typically drops as the election approaches; proponents need to anticipate this fall in support and ensure that they have built strong public bottom-up support before the election campaign begins. Having “strong” support from 49% of the public early in (or before) the campaign was strongly predictive of success. The criteria the industry developed58 (table 2) to decide which campaigns to enter could also be used by health groups to decide where they have a strong chance of winning.

It is also increasingly important to begin campaigns early, before most voters form opinions.70 Most states now have extended voting periods beginning up to 4 weeks before election day.71 This fact has an impact on the timing of media campaigns for ballot measures. In the tobacco industry post-election survey following its narrow loss in the November 1998 California Proposition 10 election,59 21% of the voters had made up their minds before September, 12% during September, 34% 1–4 weeks before the November election and only 26% during the last week. Likewise, a July 2001 poll conducted for Philip Morris on the Washington state tobacco tax in anticipation of the November election identified the significance of early voting by mail: “We clearly start out and end up with more “No” votes among people who prefer to vote at their precinct polling place than we do among people who prefer to vote by mail; high mail voting hurts”.61 Because the tobacco industry has essentially unlimited financial resources, it will be active from the outset in shaping public perceptions of key issues in the campaign. Public health advocates, however, are devoting substantial resources to these tax campaigns (table 1), so it should be possible to have some media presence early in the campaign to present their framing of the issues, as well as continuing to advertise through election day to reach people who make their voting decisions at the last minute.

The fact that the tobacco industry generally substantially outspends tobacco tax advocates does not appear to explain the outcome of the elections. This situation may reflect the facts that pro-tax advocates devoted substantial resources to their campaigns and that advertising in support of initiatives seems to be more effective (measured in terms of percentage change in the vote per advertising dollar72).

In the three elections where both smoking restrictions and tax increases were on the same ballot, the tobacco industry gave priority to opposing the smoking restrictions, perhaps because the reduced social acceptability of smoking codified in smoke-free policies reduces cigarette consumption more than a tax increase reduces cigarette consumption.73 74

Limitations

The largest limitation of this research is that the campaigns we analysed took place over 20 years, during which time many elements of the larger social and political environment were changing and that, except for Missouri in 2002 and 2006, no two campaigns were the same.

Conclusion

Ballot measures are an important venue for making tobacco tax policy by tobacco control advocates. Health groups have succeeded in passing cigarette taxes devoted to a variety of purposes and the tobacco companies have become more selective in the tax increases they oppose. The degree to which tobacco control advocates can build public support for cigarette tax increases well in advance in a way that anticipates tobacco industry arguments, may play a part in discouraging the industry from investing in an opposition campaign. In addition, tobacco tax advocates may be able to use the voter segmentation algorithm that the industry developed to assess their ability to prevail in campaigns (table 2). While ballot measures have commonly allocated substantial funds from cigarette taxes to provide medical services, the tobacco companies seem to be becoming more successful in making this use of funds an issue.

Direct democracy is spreading around the world, making initiatives available to tobacco control advocates in many countries. As countries that have ratified the WHO Framework Convention on Tobacco Control75 move to implement Article 6’s tax increase provisions, the lessons learnt in tobacco tax ballot initiatives in the United States should prove instructive on the tactics to expect from the tobacco industry and the pitfalls to be avoided by the proponents of a tobacco tax increase initiative. Just as the tobacco industry has done with efforts to eliminate secondhand smoke exposure,76 77 78 79 80 it could be expected to utilise tactics developed in the US in other countries.

Our results suggest proposing tax increases with a large fraction of the money devoted to tobacco control activities or, for larger taxes, allocating the “first dollars” to tobacco control programmes may increase the likelihood of success. The Centers for Disease Control and Prevention’s Best Practices for Comprehensive Tobacco Control Programs—200781 provides an independent authoritative source for the US and provides an approach that could be adapted to other countries. US states that failed to pass tobacco tax increases in 2006 and 2007 could have fully funded state tobacco control programmes with smaller cigarette tax increases than those proposed. For example, a cigarette tax increase in 2006 of only 58 cents in California would have brought the $77 million it spent on its state tobacco control programme up to the CDC’s recommended level of $442 million48 81 82 83 (in 2007). Likewise, in Oregon a cigarette tax increase of only 31 cents would have increased the $7 million Oregon spent to the CDC recommended level of $43 million.81 82 83 84 Including an inflation adjustment (perhaps tied to the medical component of the consumer price index) would maintain the effect on consumption and purchasing power of the tax to finance the tobacco control programme over time as prices rise.85 This solution provides ongoing funding for a tobacco control programme that saves lives,86 87 prevents billions in healthcare costs85 and avoids the tobacco industry’s argument that the tax increase is “unfair” because it fails to provide funding for programmes to help smokers and other tobacco users quit and prevent new people from starting.