[WS:] But there is also a possibility of a statistical artifact due to rather small number of cases in the "sample". To illustrate, if your sample consists of 1,000 individuals then the inclusion or exclusion of 10 cases will not make that much of a difference - it accounts for 1% of you sample. If your sample has only 20 individuals, then 10 cases accounts for 50%. Now finding 10 cases of defendants who were innocent is not that difficult, but due to the sampling error and small sample size of the capital cases, the results may be very skewed. Daniel Kahneman discussed this kind of perception error in which people try to find causal explanation to statistical artifacts at length in his latest book "Thinking, Fast and Slow" (chapter 10). I highly recommend it.
-- Wojtek
"An anarchist is a neoliberal without money."