Given that the majority of accident reports I have read have some degree of mistaken causal reasoning in Section 3, and that the statistics are directly derived from uncritical acceptance of the assertions in this section, differences are as likely to reflect what is currently "in" to say and what "out", as they are to reflect objective differences in flying airplanes.
Those mistakes can be trivial or they can be profound, and I know of no way of controlling for this confounding factor. That said, however, I'll certainly read the article at some point for the insight any such well-performed longitudinal study must provide.
PBL