"I Should Be Pregnant So Many Times By Now": Risk Perception, Numeracy, and Young Women's Contraceptive Use

Abstract

Conception is probabilistic: any instance of sexual intercourse without contraception may lead to a pregnancy, but whether any one instance of sex results in pregnancy is a matter of chance. The underlying probability of conceiving is not easily inferred from either outcome, and judgments about risk require fairly sophisticated quantitative reasoning skills (numeracy) that many Americans adults do not have. Still, women’s deliberations about risk may be tremendously important for their subsequent behavior. In this dissertation, I consider the impact of risk perception and health numeracy on young women’s contraceptive use. I use data from the Relationship Dynamics & Social Life study, a longitudinal survey of women aged 18/19 living in a Michigan county. RDSL data include weekly measurement of sex, contraception, pregnancy desire, and pregnancies, and quarterly measurement of women’s estimates of their pregnancy risk. The first empirical chapter finds evidence of a reciprocal relationship between women’s pregnancy risk estimates and their actual experiences with sex, contraception, and pregnancy. I find that women’s pregnancy risk estimates tend to decrease over time. Women who avoid pregnancy despite having sex without contraception tend to revise their estimates of pregnancy risk downward; in turn, lower estimates of pregnancy risk predict sex without contraception in later weeks. The second empirical chapter examines a competing set of risk perceptions: women’s concerns about side effects and other long-term health consequences of hormonal contraception. Sex without contraception is more likely and more frequent among women expressing greater concerns about side effects. Side effect concerns predict less use of the contraceptive pill/patch/ring, intrauterine devices (IUDs), and contraceptive implants, and more reliance on non-hormonal methods such as condoms and withdrawal. The final empirical chapter considers whether low numeracy is related to ineffective contraceptive use. In this analysis, I categorize women as low, medium, or high numeracy based on the logical consistency of their answers to survey items about the risk of pregnancy. Numeracy is not associated with women’s sexual behavior, but it does predict contraceptive use among sexually active women. Lower numeracy is associated with a higher likelihood and higher frequency of sex without contraception, more gaps in contraceptive use, and more switches to less effective contraceptive methods. Lower numeracy women also have lower odds of using the pill/patch/ring, IUD, and implant versus condoms as their primary method of contraception. Collectively, these analyses demonstrate that misunderstandings about the risk and probabilistic nature of pregnancy, concerns about contraceptive side effects, and poor numeracy are barriers to effective contraceptive use among young women who wish to avoid or delay pregnancy.