Information and sin goods: Experimental evidence on cigarettes

Abstract

We test the importance of information source on consumer choice in the context of sin goods, specifically electronic and tobacco cigarettes, among adult smokers. We proxy choice with intentions to vape and quit smoking in the next 30 days. We experimentally vary the information source: government, private companies, physicians, and no source. Our findings suggest that information source matters in the context of cigarettes choice for adult smokers. Private companies appear to be an important information source for cigarettes among adult smokers.

Keywords: choice; electronic cigarettes; experiments; information; signaling; sin goods; smoking.

Figure 1:

FDA as the information source

Figure 2:

The fictitious e-cigarette company (the Ave) as the information source

Figure 3:

Physicians as the source

Figure 4:

No information source

Figure 5:

E-cigarette at the time of survey

Figure 6:

Effect of information sources on e-cigarette use Notes: The Ave is the fictitious e-cigarette company created by the authors for the purposes of this study. All models estimated with an LPM. Outcome variables coded one if the respondent reports being extremely likely or somewhat likely to use an e-cigarette/quit tobacco cigarettes in the next 30 days, and zero otherwise. Controls include personal characteristics listed in Table 1. Reference category is no source. Circles reflect the coefficient estimate and vertical lines reflect 95% confidence intervals based on heteroskedasticy robust standard errors.

Figure 7:

Effect of information sources on tobacco cigarette use Notes: The Ave is the fictitious e-cigarette company created by the authors for the purposes of this study. All models estimated with an LPM. Outcome variables coded one if the respondent reports being extremely likely or somewhat likely to use an e-cigarette/quit tobacco cigarettes in the next 30 days, and zero otherwise. Controls include personal characteristics listed in Table 1. Reference category is no source. Circles reflect the coefficient estimate and vertical lines reflect 95% confidence intervals based on heteroskedasticy robust standard errors.