reports a BBC caption. Uh, no: According to the Justice Department's National Crime Victimization Survey (2005 data), table 66, handguns are used in 5.4% of U.S. assaults and 26.3% of robberies.

Thanks to Charles Curley for the pointer. I've e-mailed BBC with the correction; please let me know if you see them making it.

UPDATE: Even if one uses the Uniform Crime Reports data, which is generally thought to be less reliable because it focuses only on crimes reported to the police, only 42.2% of all robberies involved firearms generally (likely nearly all handguns, but still not "most" robberies), and only 21.9% of all aggravated assaults involved firearms generally, even though "aggravated assault" is a kind of assault that is especially likely to use a deadly weapon — the statistics for all assaults would surely be far lower. (I cite 2006 data here, but the 2003 data is comparable.)

I also should have pointed out that the error isn't just in the caption, but also in the article itself, which says "Handguns are used in two-thirds of robberies and assaults and in half of murders in the US, according to statistics from the Federal Bureau of Investigations." The murder statistics are about right, but, as I note above, the robbery and assault numbers are flatly wrong.

FURTHER UPDATE: The BBC has revised the page to eliminate these errors; the original is still available in the google cache.

So says the L.A. Times. But the National Crime Victimization Survey for 2005 (PDF page 81) reports that handguns are used in fewer than 8% of all crimes of violence. This doesn't include homicide, but homicides are fewer than 0.5% of all crimes of violence, so even if one includes homicides the answer would be about 8%.

Here's my guess about what happened: The petition does say that "Handguns are the weapon most likely to be used in a street crime," a statistic that the NCVS supports. According to NCVS, handguns account for a plurality (roughly 1/3) of weapons used in the about 25% of violent crimes that the victims believed were committed with weapons (as opposed to hands and fists), which is more than any other class of weapons. The petition's statement is thus accurate, though not terribly helpful and potentially misleading. But this then got transposed into the off-by-a-factor-of-6 L.A. Times statement that "[D.C. officials] argue handguns are involved in most violent crime."

Now I should stress that handguns are indeed involved in many crimes, and the majority of homicides. I oppose handgun bans because I think they do more harm than good, but I certainly don't want to pooh-pooh the harm done by criminals with handguns. But this sort of story helps show how skeptical one should be of media reporting about statistics (and likely about other things as well).

Thanks to commenter Tom Hynes for pointing this out.

The BBC story I discussed yesterday has been revised.

The "Handguns are used in most US assaults and robberies" caption has been changed to "The debate over handguns is politically charged in the US," and the "Handguns are used in two-thirds of robberies and assaults and in half of murders in the US, according to statistics from the Federal Bureau of Investigations" line has been replaced with "Firearms, including handguns, are used in two-thirds of murders and about 42.2% of robberies committed in the US, according to statistics from the Federal Bureau of Investigations." Both of the new statements are correct, though as I noted earlier, the FBI robbery statistics are probably less reliable than the Department of Justice survey-based statistics, which estimate handgun use at 26.3%.

In any case, I'm pleased that the BBC revised the story, whether based on my e-mail to them or on others'. (Note that the "Last Updated: Tuesday, 20 November 2007, 22:21 GMT" doesn't reflect this change; compare the cached version, which contains the errors but has the same timestamp.) If anyone can point me to any BBC page that specifically notes the correction, I'll be happy to note that as well.