返回Google圖書搜尋

Elusive Facts About Guns Violence

其他書名	Where Good Surveys Go Bad
出版	SSRN, 2014
URL	http://books.google.com.hk/books?id=3F3gzwEACAAJ&hl=&source=gbs_api

註釋Sometimes a survey is well designed, but the resulting estimates are demonstrably wrong, and by a wide margin. For that reason, we believe that if getting a reasonably accurate estimate is important (and if it is not, why bother?), then the analyst should ask and attempt to answer the following prosaic question: “Given everything we know, both from the survey in question and other sources, is this estimate in the right ballpark?” We might call this a “plausibility test.” It may seem like common sense, but a quick scan of reports of survey results will demonstrate that a discussion of procedure is far more common than a discussion of plausibility. In what follows we consider three examples from the study of gun ownership and use (or misuse). The first example is gun ownership - the household prevalence of guns, and the number of guns in private hands. The second is the number of individuals who are shot and wounded in assault circumstances. And the third is the number of instances in which a private individual uses a gun to defend against crime. In each case the apparent bias in estimates based on population surveys is remarkably large.