Population parameters, sample statistics, and individual observations: A lesson in the law of small numbers.

Comments on the original article "Predoctoral internship training in Canada--I: Internship settings and supervisory issues," by J. L. Howes, T. M. Vallis, A. Wilson, M. Ross, and H. Louisy (see record 1996-06731-004). It has come to our attention that the data presented for Manitoba are misleading. It is stated (p. 175) that the average stipend in Manitoba decreased from 1992-93 to 1996-97. However, this is a statistical artifact due to the addition of a second program which paid less than the existing program. Rather than a reduction, the program participating in the 1992-93 survey actually increased its funding by $1,746 (from $28,671 to $30,417). A second program opened between '92-3 and '96-7 (with a stipend of $19,000). As well, on page 174 of our paper Manitoba's stipend is listed as $27,671 when it in fact was $28,671. We regret any confusion that reporting provincial averages may have caused. To clarify the funding issue, we reanalyzed our data on a program basis, not provincial basis (i.e., programs are the unit of measurement). We included only those 30 funded programs operating in 1992-93 in this re-analysis, the results of which are presented here. (PsycINFO Database Record (c) 2010 APA, all rights reserved)