a1 University of Chicago
Scott Ashworth, Joshua Clinton, Adam Meirowitz, and Kristopher Ramsay (2008) allege that I have committed the sin of sampling on the dependent variable by considering only the universe of suicide terrorist attacks rather than the universe of all imaginable instances when potential or actual terrorists might have committed suicide attacks, and so cannot measure the effects of any independent variables. They go on to describe a method that they say I should have used, which is not of interest because the accusation that is supposed to motivate this discussion is inaccurate.
The main claim—that my work on suicide terrorism samples on the dependent variable—is simply wrong. Indeed, the authors paid no attention to the large portions of my recent book that explain what we know about factors that make resort to suicide terrorist campaigns more or less likely, and how we know it. Hence, this letter is mainly devoted to updating Ashworth, Clinton, Meirowitz, and Ramsay on my work. I also make a few comments about the general question of whether concerns about “sample bias” should carry significant weight when dealing with the complete universe of a phenomenon, as is the case in my work on suicide terrorism.