CIS — In February 2016 the Washington Post reported the findings of a study that said “there is no real link between immigration and terror.” A look at the study (behind a paywall, but you can find the abstract here and a brief précis and interview here) reveals that it is based on a very sophisticated technique called “spatial temporal regression” that is subject to significant error when misapplied by researchers who commit the sins of over-aggregation and aggregating unlike things. A classic proof of misapplication was established by Benoit Mandelbrot, in which the coastline of Britain is measured using a long yardstick, a medium yardstick, and a short yardstick, yielding different results: “it is inherently nonsensical to discuss certain spatial concepts despite an inherent presumption of the validity of the concept.”
Such a misapplication would seem to be the case here: Can one really measure migration and terror attacks equally for all countries in a 30-year period? For instance, is the risk of terror in the United States or France inherently the same as it is in Finland? And is it reasonable to measure the risk of terror by sheer volume of migrant inflows, rather than by assessing their countries of origin and cultural and religious mores? Doing so puts the terror risk from migrating Pacific Islanders on the same level as migrating Somalis. To a simple naif like me, and probably most of America, this doesn’t sound at all logical. But the study advances the desired narrative of progressives, and so the Post dutifully and somewhat gleefully publishes the article while managing to take some sideswipes at others who aren’t so perceptive as to recognize that of course there is no link between immigration and terror. What it doesn’t do is discuss the possibility that this study falls into the same foolish realm as measuring British coasts on a map by means of yardstick.