Highly publicized sex surveys suffer severe self-selection and response bias

Evidence

But as David Myers (1992, p. 12) has pointed out, these polls are often poorly done and almost certainly give a distorted account of what is really going on. Shere Hite’s (1989) popular Women and Love was based on responses from 4,500 women, which impresses one initially. But the sample was drawn exclusively from members of women’s organizations, and less than 5% of the surveys mailed out were completed and returned. So Hite’s report that 70% of the respondents who had been married at least five years were having affairs does not necessarily apply to your Aunt Bertha and Uncle Harry. Ask them, if you do not believe me. — Bob Altemeyer, The Authoritarian Specter, 2203

Categorization

• XREF: Connects to Milgrom’s work on institutions and reputation — both about how sampling and information costs distort what we can know; also connects to the chapter on research methodology in the reader’s knowledge base.
• SHARE: The Hite report detail — 4,500 respondents drawn only from women’s organizations with under 5% response rate — is a nerdy statistics-vs-media warning worth sending to a stats- or research-methods-minded friend.