Type-I error in SAS/SPSS 1 Differences of Type I error rates using SAS and SPSS for repeated measures designs Nicolas Haverkamp 1,a , André Beauducel 1,b 1 University of Bonn, Institute of Psychology, Kaiser-Karl-Ring 9, 53111 Bonn, Germany a nicolas.haverkamp@uni-bonn.de, b beauducel@uni-bonn.de April 24 th 2018 Abstract In this study, we examined the Type I error rates of Multilevel Linear Models (MLM) and repeated measures Analysis of Variance (rANOVA) as well as their respective correction methods for nine and twelve measurement occasions using the SAS and SPSS software packages. As MLM has been repeatedly proposed for the analysis of small samples, we performed a simula- tion study with the following specifications: To explore the effect of both the number of measure- ment occasions and the sample size on Type I error rates, measurement occasions of m = 9 and 12 were investigated as well as four sample sizes of n = 15, 20, 25 and 30. The effects of non- sphericity in the population on the mean Type I error rates for different analysis methods were also examined: 5,000 random samples were drawn from two populations that contained neither a within- subject effect nor a between-group effect. The random samples were analyzed with different methodological approaches to include the most common and relevant options to correct rANOVA and MLM-results: The correction by Huynh and Feldt (1976) for rANOVA (rANOVA-HF) as well as Kenward and Roger's (1997, 2009) correction for unstructured MLM (MLM-UN-KR), which could help to correct progressive bias of uncorrected MLM with an unstructured covariance matrix (MLM-UN). The resulting mean Type I error rates showed a progressive bias for the uncorrected rANOVA and heterogeneous results for MLM-UN when sphericity was violated as well as a progressive bias for MLM-UN for small sample sizes in general which was stronger in SPSS than in SAS. Moreover, an appropriate bias correction for Type I error via rANOVA-HF and an insufficient bias correction by MLM-UN-KR for n < 30 were found. These findings plead for a use of rANOVA or MLM-CS if the sphericity assumption is not violated and for a correction of a sphericity violation rather via rANOVA-HF than MLM-UN-KR. If an analysis requires MLM, however, SPSS yields more accurate Type I error rates for MLM-CS and SAS yields more accurate Type I error rates for MLM-UN. Keywords: Multilevel linear models, software differences, repeated measures ANOVA, simulation study, Kenward-Roger correction, Type I error rate