Networked collective intelligence improves dissemination of scientific information regarding smoking risks

Abstract

Despite substantial investments in public health campaigns, misunderstanding of health-related scientific information is pervasive. This is especially true in the case of tobacco use, where smokers have been found to systematically misperceive scientific information about the negative health effects of smoking, in some cases leading smokers to increase their pro-smoking bias. Here, we extend recent work on 'networked collective intelligence' by testing the hypothesis that allowing smokers and nonsmokers to collaboratively evaluate anti-smoking advertisements in online social networks can improve their ability to accurately assess the negative health effects of tobacco use. Using Amazon's Mechanical Turk, we conducted an online experiment where smokers and nonsmokers (N = 1600) were exposed to anti-smoking advertisements and asked to estimate the negative health effects of tobacco use, either on their own or in the presence of peer influence in a social network. Contrary to popular predictions, we find that both smokers and nonsmokers were surprisingly inaccurate at interpreting anti-smoking messages, and their errors persisted if they continued to interpret these messages on their own. However, smokers and nonsmokers significantly improved in their ability to accurately interpret anti-smoking messages by sharing their opinions in structured online social networks. Specifically, subjects in social networks reduced the error of their risk estimates by over 10 times more than subjects who revised solely based on individual reflection (p < 0.001, 10 experimental trials in total). These results suggest that social media networks may be used to activate social learning that improves the public's ability to accurately interpret vital public health information.

Conflict of interest statement

The authors have declared that no competing interests exist.

Figures

Fig 1. A schematic representation of the experimental design.

800 unique smokers and 800 unique nonsmokers were randomly assigned to one of three conditions: (1) a control condition where they interpreted the anti-smoking messages on their own; (2) an anonymous social network consisting of equal numbers of smokers and nonsmokers, where subjects exchanged views with peers for whom they were given no identifying information; and (3) a social network consisting of equal numbers of smokers and nonsmokers, where subjects exchanged views with peers while being aware of each other’s smoking status as either ‘smoker’ or ‘nonsmoker’. Every condition, in every trial, consisted of 40 unique subjects. Net. w. Sm. Habits Revealed, network with smoking habits of peers revealed.

Fig 2. Anti-smoking warning label used as a stimulus in the experiment.

This warning label was produced by the U.S. Department of Health & Human Services (© U.S. HHS) in 2011. The question that we used to elicit subjects’ judgements concerning the health risks of tobacco use was taken from the World Health Organization’s report on the global tobacco epidemic, 2015.

Fig 3. Changes in estimation error across experimental conditions.

Bars display the total change in estimation error from Round 1 to Round 3, averaged across all 10 experimental trials, where each trial provides one observation. All conditions are independent. The error bars show 95% confidence intervals. S, smoker; NS, nonsmoker; An., anonymous; Id., with the smoking status of contacts identified.

Fig 4. Changes in the average estimation error of individuals across experimental conditions, split by the smoking status of subjects in each condition.

(A) The performance of smokers, averaged across all 10 trials, for all conditions. (B) The performance of nonsmokers, averaged across all 10 trials, for all conditions. Network conditions contained 20 smokers and 20 nonsmokers, whereas each control condition contained 40 smokers and 40 nonsmokers. The average estimation accuracy of smokers and nonsmokers in networks was measured separately by computing the average estimation accuracy by each subgroup, within each network, thus producing two group-level observations for each network and 20 in total for each network condition. All conditions are independent. The error bars show 95% confidence intervals. w. Smoking Id., with the smoking status of contacts identified.