In it, Deborah Blum (author of the Poisoner's Handbook) asserts that homicide by poison is much more likely to be committed by a man (60.5%) than by a woman (39.5%), and that therefore the idea that poison was a woman's weapon was false. Her numbers come from this report...snapshot here:
Weapon Male Women
Gun homicide 92.1% 7.9%
Arson 78.8 21.2
Poison 60.5 39.5
So women are less likely than men to poison, but they are better represented in that group than the two others in the report. But I got curious....what does this mean in terms of absolute rates? From the way the data's presented, it does appear that if a woman murders, there's a good chance she used poison....but is this true?
I took a look at the FBI crime database to see what the absolute numbers were. The numbers above are for the years 1980-2008, and this report is for 2006-2010, but my guess is the order of magnitude holds.
For 2006-2010, there were 47,856 gun homicides, 505 fire/arson, and 49 poisonings. So despite the lower percentage, women are still almost 200 times more likely to kill using a firearm than poison*. So basically, there is no base rate fallacy going on here. If you hear someone was poisoned, it's more likely a male did it (at least for the years listed), and if you hear a women killed someone, poison was not her most likely method.
Of course poison may have fallen out of fashion a bit, so this trope could have been true in Sherlock Holmes' day, and all bets are off in the fictional world of Game of Thrones. In case you're curious, the FBI does not appear to keep data on crimes committed by eunuchs, so I can't verify any of that.
*This may be skewed, as it is far more likely that some poisonings got missed by coroners than gunshots...but I doubt the missed cases would make up much of the difference.