Abstract

ABSTRACT Analysis of smokers' posts and behaviors on Twitter reveals factors impacting abstinence and relapse during cessation attempts. Combining automatic and crowdsourced techniques, we detect users trying to quit smoking and analyze tweet and network data from a sample of 653 individuals over a two-year window of quitting. Guided by theory and practice, we derive behavioral, social, and emotional measures to compare users who abstain and relapse. We also examine the cessation process, demonstrating that Twitter can help chronicle how some people go about quitting. Among other results, we show that those who fail in their smoking cessation are far heavier posters and use relatively less positive language, while those who succeed are more social in both network ties and in directed communication. We conclude with insights on how intelligent intervention systems can harness these signals to provide tailored behavior change support.