Fact-free gun policy?

• Author: Cook, Philip J.
• Position: Response to article by Dan M. Kahan and Donald Braman in this issue, p. 1291

INTRODUCTION

Dan M. Kahan and Donald Braman's interesting article is addressed to those "academics and others who want to help resolve the gun controversy." (1) That's not us. The pragmatic goal of our labors in the consequentialist realm of empirical research has been to determine what works and at what cost. (2) Good answers to these questions strike us as important precursors to any satisfactory "resolution" of the debate. Indeed, our goal is not to end or resolve the debate, but to develop sound information that at least some voters and authorities may find useful. We aspire to be a sort of Consumer Reports for gun policy, not multicultural marketers.

But do facts actually affect gun policy in the real world? Kahan and Braman argue, persuasively, that "cultural worldviews" influence how individuals perceive gun control measures. (3) We are convinced. From there, however, Kahan and Braman draw a somewhat puzzling--and almost certainly erroneous--conclusion that culture matters, and therefore, evidence on consequences does not matter. Why can't both culture and consequences matter? The fallacy is the same as in the old question: Do you walk to school or carry your lunch?

As a matter of fact, factual information has helped steer popular opinion in debates where voters held strong and conflicting cultural attitudes, including in the area of guns. And, equally important, empirical research may affect public policy directly, independent of its influence on public opinion, by informing the decisions of courts, bureaucrats, and other actors in the policymaking process.

1. CULTURE, THE PARTIAL EXPLANATION

   Needless to say, there are differences in opinion about gun control, and those differences tend to follow certain patterns. Views on gun control measures have been correlated with objective characteristics such as gender, race, region, religion, and military experience, as well as with such subjective matters as trust in government and judgment concerning the protection offered by the police. (4) Kahan and Braman demonstrate that the two scales they construct from answers to survey questions, which, according to them, measure two dimensions of the respondent's "cultural worldview," are also associated with views on gun control. They find that people who are opposed to gun control tend to be less supportive of government social and regulatory programs (their individualism-solidarism scale) and tend to be less progressive with respect to race, sexual orientation, and capital punishment (their hierarchy-egalitarianism scale). But while circumstances and culture help "explain" (in a statistical sense) attitudes toward gun control, that is not the whole story.

   For one thing, people's views of gun control depend on the specific policy measure in question. Arguably the best data available on attitudes toward gun policy come from nationally representative surveys collected in 1996, 1997, and 1998 by the University of Chicago's National Opinion Research Center (NORC). The NORC surveys show overwhelming support for moderate gun control measures: 88% of respondents believed that all handguns should be "child proofed," 81% believed that handgun sales should be limited to one per person per month, 82% believed that handguns should be registered, 77% supported background check requirements for private gun sales, and 89% believed that people who have been convicted of domestic-violence misdemeanors should be prohibited from purchasing guns. (5) But broad support for more stringent measures is lacking; the same NORC surveys found that only approximately 37% of the respondents favored a general ban on handgun possession outside of their use in law enforcement. (6)

   If support for gun control were driven entirely by scorn for guns and the desire to "equalize wealth, status, and power," (7) why is it that over half of those who support moderate controls oppose a ban on the private ownership of handguns? It seems that much of the public is not judging these matters solely from their cultural gut but instead have a rather nuanced view of gun policy. More striking still is the finding that a majority of gun owners support each of the moderate gun control measures discussed above. (8) Thus, it appears that even "individualistic, hierarchical" gun owners may support the "egalitarian, solidaristic" goal of reducing gun violence through government regulation.

   Further evidence that there is more to the story comes from Kahan and Braman's own analysis. (We set aside for the moment the irony of their using multivariate regression analysis as a tool of persuasion to argue that multivariate regression analyses are not persuasive!) Kahan and Braman's measures of cultural attitudes among respondents to the General Social Surveys (GSS) plus their measures of the sociodemographic, religious, and regional backgrounds of respondents, together only explain 8% of the variation in people's attitudes toward gun control. (9) Is there some reason to believe that facts are entirely irrelevant in explaining the remaining 92% of the variation in the public's attitudes about gun policy?

2. DO STATISTICS EVER PERSUADE?

   Kahan and Braman argue that people will "credit or dismiss empirical evidence ... depending on whether it coheres or conflicts with their cultural values." (10) While this response undoubtedly occurs, facts surely are not irrelevant. But the Kahan-Braman analysis is not suited to determine the role of factual information. We illustrate this point by replicating their analysis for two other outcome variables: whether the respondent smokes or not, and whether the respondent used a condom during her last sexual encounter.

   Our Table 1 reports the results of our effort to replicate Kahan and Braman's regression analysis presented in their Table 1 and then reports the results of the same regression specification with the two other dependent variables. (11) As seen in the first column, our numbers confirm the Kahan-Braman results: the likelihood that a GSS respondent during the 1988 to 2000 waves reports favoring "a law which would require a person to obtain a police permit before he or she could buy a gun" is positively and statistically significantly related to the two "culture" variables, even after conditioning on race, gender, religion, region, educational attainment, socioeconomic status, and political...