Tobacco.org

Response to March 1997 Critique of 'The Effect of Ordinances Requiring Smoke-Free Restaurants on Restaurant Sales'

Response to March 1997 Critique of

"The Effect of Ordinances Requiring Smoke-Free Restaurants on Restaurant Sales"

Stanton A. Glantz

Lisa R. A. Smith

June 3, 1997

In July 1994, we published a five page paper, "The Effect of Ordinances Requiring Smoke-Free Restaurants on Restaurant Sales" in the American Journal of Public Health (1994;84:1081-1085), the leading peer reviewed journal in public health. This paper analyzed sales tax data from the first 15 cities with laws requiring 100% smoke-free restaurants and demonstrated that, contrary to claims of the tobacco industry and its front groups, such laws did not affect restaurant revenues.

On April 24, 1997, the National Smokers' Alliance, an organization created by the public relations firm of Burson Marsteller for Philip Morris Tobacco Company, held a press conference in Washington, DC to release a four pound critique of this study, prepared by economic consultant Michael K. Evans, which claimed that it was seriously flawed and that a proper analysis of the data showed that these ordinances harmed the restaurant business. This critique was widely distribued to the press and on April 29, 1997, the National Smokers' Alliance issued a press release announcing that it had written to Richard Atkinson, President of the University of California, demanding that disciplinary action be taken against Stanton Glantz. They have also written to the editor of the American Journal of Public Health complaining about the peer review process and written to health officials around the country attacking our paper. They have also created "Californians for Scientific Integrity" and started recruiting plaintiffs for legal action.

Neither Michael Evans, nor the National Smokers' Alliance, nor Burston Marsteller, Californians for Scientific Integrity, nor Philip Morris have provided a copy of this critique to Dr. Glantz or Ms. Smith. We obtained a copy from one of the reporters that covered the story.

We have thoroughly reviewed Evans' critique. Evans did identify a few minor errors in the effective dates of the ordinances. We have corrected these dates and rerun the statistics. While the numbers change slightly, the conclusion that smoke free restaurant ordinances do not affect business remains valid.

He also claims to use the same analytical techniques (after correcting the data) as we did and reach different conclusions. A close reading of his critique, however, reveals that he has not based his conclusions on the full data set and that he has manipulated his statistical modeling in ways that we do not deem appropriate.

We stand behind our paper.

Misclassification of Ordinances as 100% Smoke Free When They Were Not

Evans states that all but one of the ordinances permitted smoking in bars or bars attached to restaurants and this means that the businesses are not 100% smoke free. It is important to emphasize that our paper was a study of the effects of 100% smoke free restaurant ordinances. At the time these ordinances were passed, bar areas in restaurants and freestanding bars were excluded because of vigorous lobbying by the tobacco industry and its front groups. We recognized that bars are not included in these ordinances and discussed this point in our original paper (see page 1085). As we noted there, this is not a significant limitation.

Evans also noted that several of the ordinances permit smoking outdoors (including outdoor dining areas or take out food eaten outside the restaurant) or in rooms reserved for private functions. We do not see this as a problem. At the time that these ordinances were being passed (and vigorously opposed by tobacco interests on the grounds that business for all restaurants would drop by 30%) the term "smoke free restaurant" allowed for these exceptions. Evans does not present any reliable, independent quantitative evidence to support his assertion that these exclusions account for our results or would seriously bias our findings.

Evans also notes that a few cases (Aspen, Snowmass, and Telluride) permit creation of separately ventilated smoking rooms. We have contacted the authorities in these cities and been informed that they are not aware of a significant number of restaurants who have gone to the expense of creating such special smoking rooms. Again, Evans fails to provide any reliable, independent quantitative evidence to support his assertion that this exemption has been exercised in a significant number of restaurants to affect the results.

All the cities included in our study meet the definition of "smoke free restaurants" that was in common usage at the time (and continues to be used in may places), i.e., no smoking permitted in the indoor dining areas.

Effective Dates of Ordinances

Evans identified several cities in which he states that our paper incorrectly identified the effective dates of the ordinances. He is correct. We had based our effective dates in these cases on when the ordinance passed or information obtained from the respective city clerks. On close reading of the ordinances, we agree that some of these ordinances delayed implementation of the 100% smoke free provisions and that we had not used the correct dates. (We used the effective date of the ordinance which sometimes was a few months earlier than the effective date of the 100% smoke free provision.) We have corrected our data set and re-run the statistical analysis (revised tables are attached); while the numbers changed slightly, the overall conclusion that these ordinances did not affect revenues did not change.

Objections to Comparison Cities

Evans notes that several of the comparison cities had some sort of smoking restrictions. We recognized this fact in our paper (page 1082) and noted that to qualify as a comparison city, there had to be less than 60% of seats in the nonsmoking area. Nothing in Evans' critique identifies any comparison cities that fail to meet our stated criterion. Table 1 in our paper, however, mislabels Oroville and Lakewood as having no restrictions, whereas it should have said "some" and Breckenridge and Steamboat Springs as having "some" restrictions when it should have said "none." These errors have no effect on the statistical analysis and do not affect our conclusions that smoke free restaurant ordinances do not affect restaurant sales.

Objections to Using Sales Tax Data

Evans objects to using sales tax data. As we noted in our paper (on page 1084), sales tax data are the best source of objective information on the effects of smoking restrictions on restaurant revenues. It is complete (i.e., it covers all restaurants) and it is collected by unbiased tax authorities who have no interest in whether smoke free restaurant ordinances affect restaurant sales, and it is a crime to report incorrect results.

Evans also noted that government officials who collect sales tax data sometimes make errors. While this is true (and, indeed, we identified a few such errors and helped correct them), there is no evidence that these errors are systematic in any way that would bias the study. This complete data is preferable to samples of selected restaurants such as Evans suggests, which can be manipulated. Government tax officials also have no interest in the outcome of studies of the effects of smoke free ordinances on restaurant revenues.

Evans' Reanalysis of the Data

To a reader not familiar with statistics, it may appear that Evans did the same analysis we did, yet came up with different conclusions, thereby drawing our results in to question. This is not correct; there are important differences in how Evans did his analysis. In particular:

• Evans did not use historical data for at least 5 years before the 100% smoke free ordinance went in to effect. This failure reduces his ability to identify underlying time trends and variability that have nothing to do with the ordinance.



• Evans incorrectly states that data prior to 1991 are not available; the aggregate statistics necessary to do the study are still available for eleven of the cities in our study (Bellflower, Beverly Hills, El Cerrito, Roseville, Sacramento, and San Luis Obispo, Aspen, Snowmass, and Telluride). Moreover, much of the data needed to do the analysis was included in our paper (in Figure 1).



• Evans eliminated 3 cities (Aspen, Beverly Hills, and Ross); we used all the cities.



• Evans dropped time trends or seasonable variables if they were not statistically significant; this step could lead to model misspecification and the accompanying biases. The fact that many of Evans' analyses exhibited significant serial correlations in the residuals strongly suggests that his model was misspecified.



• Evans states that the adverse effects of smoke free restaurant laws will intensify over time, but presents no formal analysis of the data that supports this point. None of the many different statistical models he uses allow for an increasing (or reducing) impact of such ordinances over time. On page 14 he says that this point will be "argued in the conclusions" but in the conclusions (on page 15) he states this result as if it had been demonstrated earlier.

Other Research

It is important to note that since our original paper was published, 5 additional analyses of the effects of smoke free ordinances have been produced by legitimate scientific sources. All of these analyses indicate that smoke free ordinances either do not affect restaurant revenues or are associated with an increase in revenues. They are:

1. Taylor Consulting Group. The San Luis Obispo Smoking Ordinance: A Study of the Economic Impacts of San Luis Obispo Restaurants and Bars. San Luis Obispo, CA: Taylor Consulting Group (under contract to the City of San Luis Obispo). January, 1993.

2. Maroney N, Sherwood D, Stubblebin W. The Impact of Tobacco Control Ordinances on Restaurant Revenues in California. Claremont, CA: The Claremont Institute for Economic Policy Studies; The Claremont Graduate School. January, 1994.

3. Hwang P, Toblas S, Kohout S, Harris M, Satterwhite D, Simpson D, Winn L, Foehner J, Pedro L. Assessment of the impact of a 100% smoke-free ordinance on restaurant sales -- West Lake Hill, Texas, 1992-1994. MMWR. 1995;44:370-372.

4. Pope G, Bartosch W. Effect of local smoke-free restaurant policies on restaurant revenue in Massachusetts. Waltham, MA: Center for Health Economics Research (conducted for Massachusetts Department of Public Health). April 22, 1997.

5. Sciacca J, Ratliff M. Prohibiting smoking in restaurants: Effects on restaurant sales. Am. J. Health Promo. 1997;(in press).

Conclusion

We stand behind our paper.

Search News