Manuel Eisner, in his article Long-Term Historical Trends in Violent Crime (esp. pp. 115—8), proposes an interesting hypothesis on the basis of several studies: in the late medieval and early modern period, upper-class members committed as many violent acts as the lower class, and the poor really became the “violent class” (in terms of statistics, not of social representation) from the 16th century on (of course, as he does not fail to remark, it is also clear that our data are skewed and tends to record more extensive evidence on “high-profile” cases than on lowly disputes; we may even wonder whether the appearance of “lower class violence” does not stem from to the creation of a social problem rather to a real shift in behaviour). If the fragmentary picture of the studies he quotes could be confirmed, however, it would tend to show that (violent) crime is not mechanically “poverty's handmaiden”; in societies where violence had a strong social value, it was rather everyone's handmaiden. The current situation reflects a world where violence is strongly repressed by society—and naturally, less integrated parts of societies tend to be less affected by its norms, including the repression of violence.

(of course, this analysis does not hold true for other types of criminal activities, such as “property crimes”)