Impacts of Four Title V, Section 510 Abstinence Education Programs: Final Report

2,057 youth in the final follow-up survey (1,209 program, 848 control) across four programme sites

Youth enrolled in grades 3-8 at four sites: Powhatan VA (semi-rural, mostly white working/middle-class two-parent families), Miami FL, Milwaukee WI and Clarksdale MS (predominantly African-American or Hispanic, poor, single-parent households). Mean age at final follow-up about 16.5.

A Congressionally mandated, independently conducted randomised trial of school-based abstinence education — one of the largest randomised tests of a life-skills curriculum ever run, and the reason it belongs in this topic rather than in a health file. Youth were randomly assigned to be offered the programme or not; the four programmes all exceeded 50 contact hours, two of them met daily for a full school year, and one ran up to four years. Follow-up was four to six years after enrolment, which is far longer than almost anything else in the life-skills literature. Weaknesses: outcomes are self-reported (though behaviour reporting was collected under conditions designed to protect candour), the four programmes are not a random sample of Title V programmes, and control-group youth in two sites had access to rich alternative health and sex-education services — a partially active control, which makes this a test of MARGINAL effect over what the schools already provided.

Nothing moved. Program and control youth were equally likely to have remained abstinent (49% vs 49%), and abstinence in the last 12 months was 56% vs 55% (not significant). Mean reported age at first intercourse was identical at 14.9 years in both groups, and the distribution of number of sexual partners was nearly identical. Rates of unprotected sex were identical: 23% vs 23% had sex and always used a condom, 17% vs 17% sometimes, 4% vs 4% never — so the critics' concern that abstinence education would raise unprotected sex was also not borne out. On knowledge, the programmes improved identification of STDs but had NO overall impact on knowledge of the risks of unprotected sex or the consequences of STDs.

None material. Random assignment at the individual level within sites. The trial's problem is that a genuinely randomised, well-powered, 50-plus-hour, four-to-six-year test of a values-and-behaviour curriculum found nothing on any behavioural outcome.

A one-off federal evaluation with no direct replication, which is why replication is recorded as not-applicable rather than unreplicated — nobody has re-run these four programmes. The wider literature is consistent with it: abstinence-only programmes do not delay sexual initiation, and the finding was influential enough to reshape US federal funding policy.