Correction Appendix for “Sexually Transmitted Infections, Sexual Behavior and the HIV/AIDS Epidemic”, Quarterly Journal of Economics May 2005

Subsequent to the publication of this paper I have become aware of two errors in Table I, which calculates the transmission rates for the United States and Africa. The numbers for the United States are correct, but those for Africa are not. First, in the Quinn et al (2000) paper the transmission rates should be 21.9% from women to men and 21.3% from men to women; the original paper used 17.5% and 26.7% respectively. Second, the Grosskurth et al (1995) paper uses transmission rates from the wrong population; the transmission rates used in the original paper are 29.3% and 14.5% (M-to-F, and F-to-M, respectively); the correct transmission rates are 23.1% and 12.7%. In addition, while addressing these errors, it has come to my attention that the Gray et al (2001) paper and the Quinn et al (2000) paper use the same data. More specifically, the Gray et al (2001) paper uses a sub-sample of the data in the Quinn et al (2000) paper. This was not explicit in the original papers, but correspondence with an author on both papers has confirmed it. Given this, it is not appropriate to use both papers and I should have originally used on the Quinn et al (2000) paper, which covers the larger sample. This appendix presents corrected tables and figures adjusting for these three sources of error. Throughout (in footnotes) I will also give results adjusting only for the transmission rate error, and not for the study overlap. The first change is, obviously, in Table I. The new version of Table I shows the studies used to calculate transmission rates in Sub-Saharan Africa, with the corrections for errors made in the previous version. The new estimated transmission rates (based on weighted averages) are: 24.5% M-to-F and 12.4% F-to-M.1 This change will also affect Figures II, V, VI and VII and Table V. The revised version of Figure II is shown below. The estimated prevalence rate in Africa is 12.2%, relative to 12.7% in the earlier version of the paper. This is lower, but the change is relatively If we do not leave out the Gray et al (2001) study, these numbers are 23.7% and 13.5%.