Sir.—In a recent issue of AJDC,1 Brown commented on the inappropriateness of calculating the mean and SEM of ordinal data. To quote, "A precondition for the use of SEM is interval or ratio data."
This reader is disconcerted by how frequently the editors of AJDC allow this type of statistical blunder to appear on the pages of their journal. In the very issue containing Brown's article, the mean±SEM of Apgar scores are calculated and used to compare groups of babies.2 This same error appears in subsequent AJDC articles in reference to Apgar3,4 and risk5 scores. Perhaps the editors of AJDC should pay closer attention to Brown's article.
POWERS WF. Discrepancies in Statistical Evaluation. Am J Dis Child. 1985;139(10):962. doi:10.1001/archpedi.1985.02140120008005