Comment on Johnson's "On the Prevalence of Rape in the United States"

Abstract

Professor Johnson's paper (Sings 6, no. 1 [Autumn 1980]: 136-146) poses some thorny questions when seen from the perspective of how social science can be used to help people deal effectively with acts of violence in our society. Demographic studies usually make for dull reading, though when used as an approach to studying interpersonal or social violence they can introduce a note of sanity by testing popsociological explanations.1 However, when they are used to sharpen public consciousness and to achieve the other goals that Johnson sets forth, the burden upon the analyst grows heavier in two respects. First, the data must be of high quality, able to sustain the uses to which they are put; and second, the assumptions governing the methods employed must be reasonably well met. If not, the conclusions drawn are likely to be misleading or convincing only to the already persuaded. I find this analysis to be deficient in certain key respects. Let me specify the principal flaws that continue to dilute the force of Johnson's argument.2 First, as to mission: The language used throughout (e.g., oppression of women, patriarchal society) and his conclusion (Sexual violence against women is part of the everyday fabric of American life) reflect a deeply committed personal stance. In itself, this is not very surprising, given the current rejection of value-free social analysis. Moreover, rape cannot-indeed, should not-be taken lightly, either in moral terms or in one's professional work. But at some point, ideology or moral revulsion can tempt the analyst to exhort rather than to examine. I fear this is the case here, in the author's choice of problem and in his handling of data and inferences. Second, the marshaling of imperfect data: Here one can see a struggle going on between a desired result and the author's intimate knowledge of the imperfections of the data at hand. Many times he cautions us about the limitations of self-reports or official records, estimating national rates from thirteen-city survey data, race differences in reporting attempted or completed rapes, etc. This leads him to attempt to construct life tables based on survey data and then to estimate women's presumed lifelong risks in terms of a range of values rather than a single estimate.