Translator Disclaimer
1 May 2008 Response from Singer
Author Affiliations +

Many thanks to Peter Bednekoff for his comments on my essay “Dualism, Science, and Statistics.” I'd like to respond briefly to his second point about how data should be presented and to his third point about confidence intervals.

Regarding the second point, I applaud both the intent and the application of writing “prose with a living entity as the subject and its actions as a verb...and statistics as a form of technical validation.” In my example, I described a weak student who was not sophisticated enough to follow this advice. My intent is to provide a framework in which the weak student is coerced into confronting and understanding the science, even if his presentation is weak. I agree that we must work on improving the quality of the presentation while addressing the quality of the science.

Regarding the third point, confidence intervals are very useful for reminding students that mean values of samples may be very different from actual mean values of populations. They also have an advantage over p values in that they are easily represented in graphs. But for testing the predictions generated by research hypotheses, they still suffer from the drawback of having a sharply drawn line (at 95 percent). I would not want my students to think that a hypothesis is on or off the table on the basis of whether a sample mean value falls just within or just outside of a 95 percent confidence interval.

Fred Singer "Response from Singer," BioScience 58(5), (1 May 2008).
Published: 1 May 2008

Back to Top