On Fri, 26 Jul 2002, Wojtek Sokolowski wrote:
[Three strikes research snipped]
>
> This is a prime example of researchers drawing conclusions beyond the scope
> of their data. To demonstrate what motivates criminals, one needs to
> collect data on their thought process (e.g. interviews). All they
> seemingly have, however, is the "before" and "after" homicide rates (I
> wonder if they compared that increase to changes in regions that did not
> implement the said laws?) which can be explained by a myriad of alternative
> factors (e.g. changes in demographic composition or the economy).
>
> wojtek
>
>
I think it's important to keep in mind this is exactly the type of quasi-experimental research that has provided clear evidence that smoking causes health problems. True, if you do not control the independent variable (here, three strikes legislation), drawing causal conclusions is challenging. However, reasonable scientists often provide striking evidence of causality with carefully designed and analyzed quasi-experiments. We really don't know whether they controlled for the important confounds here until we read the research report.
Miles