Caught in the crossfire: a defense of the cultural theory of gun-risk perceptions.

• Author: Kahan, Dan M.
• Position: Response to articles in this issue, p. 1329, 1341, 1349, 1371, 1387

Anyone who despairs of the prospect for consensus in the American gun debate should be heartened by the commentaries on our article. (1) Cook and Ludwig finally see eye-to-eye with Lott and Mustard.

Of course, the consensus among the commentators that we must be wrong about something doesn't hearten us. We are truly gratified by their thoughtful commentaries, though, and we will now try to return the favor--by returning fire.

1. A CULTURAL THEORY OF SMOKING- AND UNSAFE-SEX-RISK PERCEPTIONS

   Cook and Ludwig think we are too quick to dismiss the power of empirical information to influence individuals' positions on gun control. "[W]e know," they write, "that people's attitudes and behaviors about smoking and unprotected sex have changed dramatically over time" as a result of "epidemiological research about the health risks associated with each of these activities." (2) If "statistical evidence can affect people's attitudes and behaviors even in areas where cultural attitudes run deep," (3) why assume that some future expert consensus on the consequences of gun control can't do the same?

   It turns out, though, that we don't know that changes in smoking and sexual practices reflect the dissemination of information on health risks. On the contrary, public health professionals have concluded that managing the social meanings that individuals attach to smoking and unsafe sex is the only effective means of reducing such behavior, and that risk information abstracted from such cultural resonances is inert. We will summarize the evidence for these conclusions. But first we want to show how one reasonably could have predicted it on the basis of Cook and Ludwig's own remarkable finding that our cultural orientation scales explain a significant amount of the variance in smoking and condom use. At that point, we'll be in a good position to clarify why our data suggests that empirical evidence on the consequences of guns can't settle the gun debate.

   How should we expect empirical data to influence individual perceptions about the risks of smoking and unsafe sex? Consider the answers suggested by three different models of risk-information processing. The rational-weigher model assumes that individuals will respond to empirical data in a manner that maximizes their welfare. Accordingly, as information accumulates that smoking, unsafe sex, or some other activity is in fact riskier than individuals had previously believed, we can expect them to reduce participation to the optimum level. (4)

   The irrational-weigher model, in contrast, asserts that individuals are incapable of responding to risk information in a way that maximizes their well-being. Cognitive biases will cause them systematically to underestimate certain risks, while volitional defects will prevent them from conforming their behavior to their judgment of what is in their interests. Accordingly, we should expect individuals not to be nearly so responsive to information about the risks of smoking or other dangerous activities as they should be. (5)

   Finally, the cultural-evaluator model holds that individuals' risk perceptions are constructed by their cultural worldviews. Culture supplies the meanings--of courage or irresponsibility; of self-reliance or self-centeredness; of trust in or indifference to others--that determine the benefit that individuals derive from any particular form of risky behavior. Equally important, cultural orientations affect individuals' perceptions of the cost of such behavior by inclining them either to invest credence in or to dismiss evidence of risk. Data purporting to show the risks of smoking, unsafe sex, and the like will thus influence behavior, on this account, only when that data is congenial to individuals' visions of a properly ordered society. (6)

   These views generate not only different predictions about how individuals will respond to information about risk but also different strategies for how to manage risk taking. The right strategy under the rational-weigher model is simply to furnish individuals with accurate information about smoking or unsafe sex, at which point they will make their own best judgment about the costs and benefits of engaging in it. The irrational-weigher model, because it denies that individuals can accurately form and act on cost-benefit judgments, favors regulatory management of smoking, unsafe sex, or other forms of risk taking consistent with expert cost-benefit determinations. (7) The cultural-evaluator view says that only information or policies that influence the meanings that individuals attach to particular types of risk taking can change their attitudes or behavior. (8)

   One can imagine various ways to test the relative power of these three models. But one way is simply to measure, as Ludwig and Cook have done, the relationship between such risk taking and individuals' cultural orientations. For it turns out that the models generate competing hypotheses on what that relationship will be.

   The rational- and irrational-weigher models both imply that there shouldn't be a significant correlation between cultural orientations and either smoking or unsafe sex. As Cook and Ludwig point out, evidence of the health risks of these activities is massive and widely available. (9) Because there is no reason to doubt that individuals of all cultural orientations have been exposed to it, there is no reason to think that individuals of only certain orientations and not others would revise their welfare-maximizing strategies in the manner predicted by the rational-weigher model. Likewise, there is no reason to believe that cognitive biases and volitional defects are spread out unevenly across individuals of diverse cultural orientations, and thus no reason under the irrational-weigher model to believe that individuals of one orientation or another will react differently to risk information.

   There obviously is a reason under the cultural-evaluator model, however, to expect individuals of different cultural orientations to react differently to smoking- and unsafe-sex-risk data. Individuals whose orientations invest smoking and unsafe-sex with (relatively) positive social meanings will either derive more compensating gain from such activities or be less likely to credit the information about the risks associated with them than will individuals whose cultural orientations invest smoking and unsafe sex with relatively negative social meanings. Alone among the three theories, then, the cultural-evaluator model predicts a significant correlation between worldview and willingness to smoke or engage in unsafe sex. (10)

   Lo and behold, that is exactly what Cook and Ludwig find in their regression model. So, if that were the only evidence one had about what influences such behavior, one would already have reason to believe that health-risk information, by itself, can't be expected to reduce smoking or increase safe sex.

   In fact, however, social scientists and public health professionals have compiled many additional sources of evidence that confirm this interpretation of the Cook-Ludwig data. For example, the most recent and authoritative national survey on smoking among youth flatly contradicts claims that "young people reduce their smoking because of their heavy exposure to warnings about cigarette risks." (11) The most comprehensive meta-data study on the effectiveness of smoking-cession messages echoes these findings, concluding that information about the short-term and long-term health effects of smoking is just "not effective" in shaping behavior. (12) And studies of smoking reduction show that the campaigns containing scientific information--ones Cook and Ludwig would presumably want to emulate for guns--"are not particularly effective at preventing young people from smoking or motivating adults to stop. (13)

   If individuals aren't responsive to information about the consequences of smoking on their health, what are they responsive to? The answer, it appears, are the social meanings that they attach to smoking, meanings that have more to do with conceptions of the moral qualifies of tobacco companies and smoking than with knowledge about smoking's health effects. Some of the best known and most effective of these antismoking ads contain no information about smoking risks; instead they show tobacco officials testifying that smoking is not addictive and has no proven health consequences. (14) The information conveyed in such messages is clearly not empirical data about the harm cigarettes do, but information about the moral status of big tobacco that effectively taints cigarettes and smoking by association.

   Striking evidence on the relative efficacy of social meaning versus health-risk-oriented smoking campaigns comes from California, where the tobacco industry convinced the state to pull moralizing anti-industry ads and substitute traditional health-information ads. This shift in advertising strategy--from culture-based appeals to health-risk ones--created a natural experiment on the relative power of each to influence behavior. The result? Smoking reductions slowed and, eventually, smoking rates increased). (15)

   The literature on smoking also highlights the distinction we make between the manner in which risks can be either cognitively or morally derivative of social norms. (16) It turns out that people are not impervious to health-risk information. People's perceptions of factual health risks are moderately related to their exposure to clear and consistent information (moderated, we think, by the cognitive filters that comport with our norm-pervaded moral evaluations). (17) But to the distress of public health advocates everywhere, increased knowledge about smoking's effects is minimally related--if at all--to smoking behavior. (18) That is, unless the information conveyed by antismoking campaigns lends a negative moral valence to smoking, researchers have found lit tie or no effect on behavior. (19)

   Those...