19

In the 2010 Summary Report of the CDC's National Intimate Partner and Sexual Violence Survey, there are two figures that are surprisingly similar:

In Table 2.1, they claim the 12-month weighted rate of a US woman being raped is 1.1%.

In Table 2.2, they claim the 12-month weighted rate of a US man being "made to penetrate" is also 1.1%.

It would appear that these risks are comparable.

Despite these figures being similar, the lifetime weighted rates are very different (18.3% versus 4.8%, respectively).

Are these figures correct, and is it reasonable to directly compare them?

  • 1
    Welcome to Skeptics! The quoted sentence doesn't appear to be in that document. (Also, the title is biased, inviting only answers in one direction.) – Oddthinking Jun 21 '15 at 13:34
  • @Oddthinking Sorry, I've tried to improve the question slightly, if it's inappropriate for the site, I'll happily remove it. – deworde Jun 21 '15 at 13:37
8

Page 12 says,

Lifetime and 12 Month Prevalence Estimates of Violence

Lifetime prevalence refers to the proportion of people in a given population who have ever experienced a particular form of violence. Lifetime prevalence estimates are important because they provide information about the burden of violence within a population.

12 month prevalence provides information about the proportion of people in a given population who have experienced a particular form of violence in the 12 months prior to taking the survey. Twelve-month prevalence estimates provide a snapshot of the recent burden of violence in a population. When collected over multiple years, 12 month estimates can be used to assess trends in the burden of violence over time (suggesting whether violence may be increasing or decreasing).

I suspect the figures which you quote are as reported.

The "weighting" is to try to account for people who don't answer the survey -- page 13 says,

Statistical inference for prevalence and population estimates were made based on weighted analyses, where complex sample design features such as stratified sampling, weighting for unequal sample selection probabilities, and non-response adjustments were taken into account. The estimates presented in this report are based on complete interviews.

So given that the sets of numbers don't match, I assert that the reason for those numbers is (as for all their numbers) that they're simply reporting what they were told by the people being surveyed.

FWIW my hypotheses for the discrepancy are that it could be either:

  • Misreporting by the people being surveyed (the men's looks less likely to me than the women's, with the men reporting a lifetime rate of only 4 times higher than the yearly rate)

  • A recent significant increase in the rate of violence against men (reflected in the recent 12 month figures but not evident yet in the lifetime figures), which is the kind of phenomenon which the "12 month prevalence" questions were intended to assess (as stated at the end of the second paragraph of the "Lifetime and 12 Month Prevalence Estimates of Violence" quote above).

  • 1
    Is there any sort of analysis of reliability of the answers? – user5341 Jun 26 '15 at 1:37
  • Good question and well asked. There's a 3-page "Chapter 8", including (see especially) the paragraph which begins "As an example of" on page 84. I don't know of (haven't looked for) peer analysis outside this document. That (Chapter 8) analysis is perhaps not quite as lengthy as the analysis I may remember having seen (many many months ago) associated with the corresponding Australian survey. – ChrisW Jun 26 '15 at 6:18
  • 1
    Almost all the male data in the survey has a 4:1 ratio for lifetime:12-month. I think it reflects experiences that aren't particularly memorable for men. For example, the survey is worded so that if a man receives oral sex while he is too drunk to consent, he was "made to penetrate". – DavePhD Jun 30 '15 at 13:11

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .