Attitudes towards Risk and the Framing of Bioterrorist Prevention Programs

Principal investigators:

Cindy D. Kam

Vanderbilt University


Elizabeth N. Simas

University of California, Davis


Sample size: 1013

Field period: 02/11/2005-02/16/2005


Political debate routinely focuses on the potential costs and benefits of policies and the uncertainty surrounding political consequences. Given that these debates often involve competing frames, we apply framing theory and the classic Tversky and Kahneman “Asian disease problem” (1981) to contemporary policy decisions. We present subjects with two pairs of objectively equivalent choices about a hypothetical small pox breakout, manipulating whether they receive a survival or mortality frame. Our design is unique in that it allows not only for the traditional between-subject comparison, but also for within-subject analyses, as all subjects were exposed to both frames. In addition, we introduce a 7-question battery for assessing risk orientation and test its ability predict the individual’s choice. We find that both frame and risk orientation are significant predictors of preference for the probabilistic outcome and that risk orientation adjudicates between those who opt for a sure-thing on both trials and those opt for the probabilistic outcome on both trials, suggesting that heterogeneity in risk orientation provides one means of explaining why the classic preference reversal is not observed in all individuals.


1: Risk acceptance will increase the probability of opting for a gamble over a sure-thing.

2: The effect of risk acceptance will vary with frames. The risk accepting will be especially likely to respond to the region of losses with a willingness to take a gamble. The risk averse will be especially likely to respond to the region of losses with an unwillingness to take the gamble. In other words, the effect of risk acceptance will be especially pronounced in the mortality frame. Here, we argue that the mortality frame (by situating individuals in a region of losses) should do more to activate attitudes towards risk than the survival frame, by conjuring the “worst-case-scenario” among the risk averse and by activating a “nothing-to-lose” scenario among the risk tolerant.

3: An individual’s preference reversal across successive frames will depend upon risk acceptance. The risk-accepting (averse) will be especially likely to opt for the gamble (sure-thing) regardless of framing condition.

Experimental Manipulations

All subjects then received a description of a smallpox policy proposed by the Centers for Disease Control; one-half of the subjects were randomly assigned to receive policy choices that were described using a survival frame; the other half received a mortality frame. This between-subject design enables us to compare different subjects’ responses to the survival and mortality frames. The experiment was then repeated, but this time the assignment of frames was reversed. This within-subject design enables us to capture preference reversals among subjects, as a single subject is exposed to both frames.


Following exposure to each frame, subjects were asked to indicate which policy they would choose and how certain they were of their choice.

Summary of Results

Consistent with the work of Tversky and Kahneman (1981), we find a significant effect of frame, as the probability of selecting the risky outcome increased with exposure to the mortality vs. survival frame. Probit analyses also confirm H1: the probability of opting for the risky outcome increased among the risk acceptant. The interaction suggested by H2, however, is not supported. The effect of Risk Acceptance does not appear to depend upon which frame is presented to respondents, suggesting rather that risk orientation predicts preference for a probabilistic outcome over a sure-thing – regardless of how that outcome is framed. This idea is further confirmed by ordered probit analyses of the within-subject component of our design. Here we find that in the line with H3, risk orientation does a better job discriminating between those who opt for the sure-thing in both trials and those who opt for the probabilistic outcome on both trials than the initial frame. Those are who are more risk acceptant are more likely to choose both risky policies regardless of whether they received the mortality or survival frame first.