The Other Side of Statistical Significance: A Review of Type II Errors in the Australian Medical Literature

Abstract

: The other side of statistical significance: a review of Type II errors in the Australian medical literature. J. C. Hall, Aust. N.Z. J. Med., 1982, 12, pp. 7–9. A review of 1416 scientific papers which appeared in the Australian medical literature between 1976 and 1978 revealed 191 instances (13%) of authors claiming “negative” results. In 72 (38%) of these papers there was enough information present to estimate the probability of a false-negative (Type II) error. Forty-three (60%) of these admissible studies had a greater than 50 per cent chance of missing a true 10 per cent difference between the groups under investigation. Critical readers should be suspicious of unqualified statements about the inability of a variable to discriminate between two groups, especially when small sizes have been used.