"More sex is safer sex" is the title of Freakonomics author Steven Landsburg's latest book. The title idea is taken from Landsburg's 10 year old Slate column, based on a paper by Michael Kremer, but the book was published last year. The argument seems to be that the most promiscuous spread disease, so if people with few sex partners each added a sex partner, those with many sex partners would have to have fewer partners, and there would overall be less disease transmission. In other words, abstinence messages appeal to those with few lifetime sexual partners, increases the variance in the number of sex partners, and spreads more disease.
Gil Kalai covers the idea briefly in his review of the book in the AMS bulletin, and phrases it in terms of graph theory: if people are nodes in a graph, and each sexual encounter is an edge, then "if we fix the number of edges (sexual encounters), then the epidemic will spread more rapidly when the variance in the degrees of the vertices is larger."
That assumption that the number of sexual encounters in a population is fixed would be interesting to test. Population-level properties are usually not investigated when studying interventions: people tend to look at individual-level outcomes. Since the goal of abstinence education is to decrease the total number of sexual encounters in a group, or in wonky jargon "decrease the size of the sexual pie", it would be interesting to see if that were true. To extend the idea even farther, we could go from micro-level behavior to the macro, and look at the elasticity of demand for sexual encounters in a population, using macro-level data and standard (flawed) macro demand equations.
Another way of saying "More sex is safer sex" is "If sex is made deviant, only deviants will have sex." Logically, I can't help but think of, "If guns are outlawed, only criminals will have guns," but, oddly, the overlap between the gun control and sex control lobbies is very small.