Working Paper: NBER ID: w24974
Authors: Jeffrey E. Harris; Mariana Gerstenbluth; Patricia Triunfo
Abstract: We asked 97 cigarette smokers to make a series of 12 binary choices between experimental cigarette packages with varying warnings and background colors. Each smoker had to decide which of the two packages contained cigarettes less risky for his health. We tested whether the smokers, confronted with warnings that were repugnant and threatening to many of them, could still make choices that adhered to the standard axioms of rational choice. We supplemented our observations on smokers’ choices with data on their eye movements. We find that participants universally made choices consistent with a complete, transitive preference ordering. We find little evidence of inconsistent choices violating the weak axiom of revealed preference. In a majority of smokers, we find strong evidence of the use of a lexicographic decision rule to assess the riskiness of a cigarette package. These smokers first ranked the two packages solely on the basis of their warnings. Only when the two packages had the same warning did they rank the packages on the basis of their color. The data on eye tracking strongly confirmed the lexicographic nature of the underlying decision rule. Our study represents an entirely different angle of inquiry into the question of rational addiction.
Keywords: Cigarette Packaging; Rational Choice; Eye Tracking; Discrete Choice Experiment; Lexicographic Preferences
JEL Codes: D12; D83; D87; D91; I12; M31
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| cigarette packaging warnings (D18) | decision-making processes of smokers (D91) |
| eye-tracking data (Y10) | decision-making processes of smokers (D91) |
| warnings on cigarette packages (D18) | choices of smokers (D91) |
| lexicographic decision rule (D81) | decision-making processes of smokers (D91) |
| warnings (Y20) | utility derived from different warning images (Y90) |
| cadaver warning (Y60) | perceived risk compared to boot warning (D80) |
| light brown background (Y91) | perceived risk compared to gray (D81) |