Rapid responses are electronic letters to the editor. They enable our users to debate issues raised in articles published on thebmj.com. Although a selection of rapid responses will be included online and in print as readers' letters, their first appearance online means that they are published articles. If you need the url (web address) of an individual response, perhaps for citation purposes, simply click on the response headline and copy the url from the browser window. Letters are indexed in PubMed.
Price is right in concluding that statistical understanding cannot
rely on sexy presentation alone. Sure, this may be feasible for the
description of data. However, the issue is whether inferential statistics
can be so-treated, since this aspect of the discipline provides the
decisions about the worth or not of experimental interventions.
Inferential statistics retains an aura of opaqueness for many
students. In part, this rests on the predominant model based on normal
distributions and - more specifically - sampling distributions. This has a
number of consequences which are not always easily appreciated.
One concerns the fact that computed probababilities may approach -
but never attain - a value of zero. Such an issue raises many questions
regarding how far the model can be violated. The suggestions often seem
rough and ready - and cannot aid the students' basic understanding.
Another concern is that the outcome of significance testing reflects
the size of the sample: the larger the sample, the more likely a
significant effect is to be demonstrated. It is not often clear to many
students that a feable effect can be significant if we just use a large
Such important issues are never likely to be sexy, no matter how
potent they may be! Nor are they likely to vanish: alternatives to
significance testing never seem to make much headway.