Information and Safe Sex: Are Better Informed Youth More Likely to Use Contraceptives and Condoms?

Monday, June 11, 2018: 10:40 AM
Salon IV - Garden Level (Emory Conference Center Hotel)

Presenter: Dean Lillard

Co-Author: Yehia Mekawi

Discussant: Christopher Carpenter


We investigate whether information changes an individual's sexual behavior. We focus on information about sexually transmitted infections and diseases (STIs). We study three types of sexual behavior. We model whether information delays the age at which a person first has sex and the probability he or she uses contraceptives and condoms. To characterize information we use magazine articles about STIs. We will create separate measures of information about HIV and other STIs. We collect articles published in popular consumer magazines from 1970 to 2014. With individual data, we predict magazine reading habits. Specifically, we estimate models of whether and how intensively individuals read specific magazines. We merge the coefficients to 1997 National Longitudinal Surveys of Youth (NLSY97) respondents' data. The NLSY97 includes data on individual sexual practices and the demographic information we use to predict each individual's magazine reading. After merging data on articles appearing each year in each magazine, we construct measures of the information each person potentially saw. We also use instrumental variables methods to account for three types of endogenous behavior: 1) people choose which magazines they read, 2) editors decide whether or not to publish an article, and 3) editors decide how long each published article will be. We predict each of these behaviors with an instrument that is plausibly uncorrelated with each sexual behavior we model. Our instruments include the count of articles appearing in the scientific press (i.e. academic journals) on the same topic as the consumer magazine articles, the cost of advertising in the consumer magazines, and the price of a subscription to each magazine. We construct a “naive” and instrumented version of each individual's information exposure in each year. We use those measures to construct not only the information stock and flow for each individual, we also construct comparable measures for each individual's mother, father, and siblings. In every case, we construct two different measures of information. One counts the number of articles a person saw each year. The other counts the number of pages in the articles a person saw. We use those measures to fit the models of sexual behavior on the combined sample of men and women and separately by sex. We find that men and women change their sexual behavior as their stock and flow of information changes. Women's use of condoms responds more to information about STIs than men's. Both men and women initiate sex later when they see more information about STIs.