Tapsearch Com EditorTrends in interracial rape, 1996-2005
January 27th 2007 17:48
Over at Blogger News Network, a commenter challenged a critique I did of a Nicholas Stix article. Stix had asserted the black-on-white rape rate was 200 times the white-on-black one; I argued that the 200 number wasn't meaningful, but that blacks indeed rape whites more than the reverse happens. The commenter said even that was wrong. Check out the comment and my response here.
Dealing with such a touchy subject, I should say what interested me in the first place: the Duke lacrosse scandal. Liberals, including school faculty, painted the supposed rape as a common example of whites exercising power over blacks. The thesis of the original post was that, while white racism works its way into many aspects of American life, rape isn't statistically one of them.
Anyway, the exchange inspired me to look up the National Crime Victimization Survey survey data of crime victims to see how things have changed over the years. The government collects information about perpetrators' races (this is where Stix got his information). Numbers are available from 1996-2005, and I arranged them chronologically in this Google spreadsheet.
The first thing I would like to make clear is that the vast majority of rapes are intraracial -- the data do not justify panic by whites that black rapists are out to get them, even if, as I showed in the previous post, that scenario is more likely than its reverse by the NCVS data.
Also, many of the data points were astericked, indicating they were based on small sample sizes. Due to the large fluctuations, I had to run statistical correlations instead of simply looking to see if the numbers rose or fell. I present this not as hard fact but as the best information available.
White-on-black rape is so rare there really isn't any way to measure its ups and downs. 5 of the 10 years, the NCVS put the number at 0 for its respondents.
However, since 1996, white rape victims have been increasingly likely to report a black assailant. The first three years, 10, 10 and 11 percent of white rape victims reported black assailants; the last three, 21, 11 and 43 (as I said, the bouncing is worrisome).
Presuming the data reveal an actual trend, this could be due to (A) blacks raping whites more, (B) whites raping whites less or (C) some change in NCVS-answering trends.
(The NCVS is given to a sample of households, not only to crime-reporters, so it's an admittedly incomplete way to answer the "people report rapes by blacks to the police more" argument. The survey, to some extent, can show differences between actual and reported crime.)
Nerdy stats stuff:
Looking at the raw data in the spreadsheet, we see a negative, statistically significant (bold) correlation between the year and the likelihood a white victim reported a white assailant. You see a statistically significant positive one for black assailants.
One thing I noticed was that 2005 was a drastic year -- only 44.5 percent of white victims reported same-race assailants, and 33.6 reported black rapists. Those are the lowest and highest numbers respectively by quite a margin, so I re-ran the correlation without 2005. The correlations remain, but the increasing likelihood of a black assailant drops out of statistical significance.
Another thing I noticed was that the "Other" and "Unsure" categories varied significantly over the years. So, in the five columns on the right, I controlled for that by removing those categories. In other words, the right-side data only considers rapes within the white and black communities, not rapes involving other or unknown races.
The results were basically the same -- significant correlations when 2005 is included. However, once you exclude 2005, both correlations fall just shy of significance at the 5 percent level.
Dealing with such a touchy subject, I should say what interested me in the first place: the Duke lacrosse scandal. Liberals, including school faculty, painted the supposed rape as a common example of whites exercising power over blacks. The thesis of the original post was that, while white racism works its way into many aspects of American life, rape isn't statistically one of them.
Anyway, the exchange inspired me to look up the National Crime Victimization Survey survey data of crime victims to see how things have changed over the years. The government collects information about perpetrators' races (this is where Stix got his information). Numbers are available from 1996-2005, and I arranged them chronologically in this Google spreadsheet.
The first thing I would like to make clear is that the vast majority of rapes are intraracial -- the data do not justify panic by whites that black rapists are out to get them, even if, as I showed in the previous post, that scenario is more likely than its reverse by the NCVS data.
Also, many of the data points were astericked, indicating they were based on small sample sizes. Due to the large fluctuations, I had to run statistical correlations instead of simply looking to see if the numbers rose or fell. I present this not as hard fact but as the best information available.
White-on-black rape is so rare there really isn't any way to measure its ups and downs. 5 of the 10 years, the NCVS put the number at 0 for its respondents.
However, since 1996, white rape victims have been increasingly likely to report a black assailant. The first three years, 10, 10 and 11 percent of white rape victims reported black assailants; the last three, 21, 11 and 43 (as I said, the bouncing is worrisome).
Presuming the data reveal an actual trend, this could be due to (A) blacks raping whites more, (B) whites raping whites less or (C) some change in NCVS-answering trends.
(The NCVS is given to a sample of households, not only to crime-reporters, so it's an admittedly incomplete way to answer the "people report rapes by blacks to the police more" argument. The survey, to some extent, can show differences between actual and reported crime.)
Nerdy stats stuff:
Looking at the raw data in the spreadsheet, we see a negative, statistically significant (bold) correlation between the year and the likelihood a white victim reported a white assailant. You see a statistically significant positive one for black assailants.
One thing I noticed was that 2005 was a drastic year -- only 44.5 percent of white victims reported same-race assailants, and 33.6 reported black rapists. Those are the lowest and highest numbers respectively by quite a margin, so I re-ran the correlation without 2005. The correlations remain, but the increasing likelihood of a black assailant drops out of statistical significance.
Another thing I noticed was that the "Other" and "Unsure" categories varied significantly over the years. So, in the five columns on the right, I controlled for that by removing those categories. In other words, the right-side data only considers rapes within the white and black communities, not rapes involving other or unknown races.
The results were basically the same -- significant correlations when 2005 is included. However, once you exclude 2005, both correlations fall just shy of significance at the 5 percent level.
| 48 |
| Vote |
Add To: 
subscribe to this blog
