How Not To Use Statistics

Guest Post: The Nature of Privilege | Dispatches from the Culture Wars:

"Those half dozen instances are the tip of the iceberg in terms of the avalanche of attacks on trans* people throughout the world, particularly against transwomen of color. A nationwide survey of bias-motivated violence against LGBT people from 1985 to 1998 by the National Coalition of Anti-Violence Programs found that incidents targeting transgender people accounted for 20% of all murders and about 40% of all police-initiated violence. According to the same project, in 2010 44 percent of LGBTQH (lesbian, gay, bisexual, transgender, queer and HIV-affected) murder victims were trans women, and in 2009 trans women were 50 percent of murder victims. Yet trans people as a whole are only about 1 percent of the LGBTQH population. Trans women also more often experienced multiple forms of violence and more severe violence, as well as more police bias and violence."

'via Blog this'

Reading that, you would think there is an epidemic of murder involving transgender folk. Fortunately this isn't actually the case. In 2003 and 2004 there were a total of 38 murders in the LGBT hate-crime category. I can't tell you how many there were in 2010 or 2009 because the pdfs are mislinked and I can't read the damn things. I am guessing that it is a roughly similar number.

I always get suspicious when someone talks in percentages rather than absolute numbers and I get particularly annoyed when someone who I would probably agree with on most things does a bullshit move like this.