Knight's 2003 analysis of the effect of the WAIS–III instructions on the Matrix Reasoning subtest was based on multiple t tests, which is a violation of conventional statistical procedures. Using this procedure significant differences were found between the group who know the subtest was untimed versus the group which did not know if the subtest was timed or untimed. Reanalysis of the data used three statistical alternatives: (a) Bonferroni correction for all possible t tests, (b) one-way analysis of variance, and (c) selected t tests with the Bonferroni correction. All three analyses yielded nonsignificant differences between means, thereby changing the conclusions of Knight's study.
Get full access to this article
View all access options for this article.
References
1.
CharterR. A. (1982) Practical formulas of strength of association measures. Educational and Psychological Measurement, 42, 969–974.
2.
GordonL. V. (1978) One-way analysis of variance using means and standard deviations. Educational and Psychological Measurement, 33, 815–816.
3.
HaysW. L. (1973) Statistics for the social sciences. New York: Holt, Rinehart & Winston.
4.
KnightT. A. (2003) WAIS–III Matrix Reasoning: Instruction effects on task perception and performance. Psychological Reports, 93, 66–68.
5.
SidakZ. (1967) Rectangular confidence regions for the means of multivariate normal distributions. Journal of the American Statistical Association, 62, 626–633.
