Why Minitab is robust enough ?
June 17, 2002 at 11:50 am #29667
Ovidiu ContrasParticipant@Ovidiu-Contras Include @Ovidiu-Contras in your post and this person will
be notified via email.
I’ve been told that I can use ANOVA and T tests to non-normal data . Now ,I read Mr.Shree Padnis article about handling non normal data , so my question was ” why?” .
The answer I got was :”Minitab is robust enough…”.
Please ,can someone develop further on this topic ? What are your recommendations ?
Thank you0June 18, 2002 at 1:07 am #76476
Its not so much MINITAB as the ANOVA/t-Test methods themselves. They both involve only the assumption that the means are normally distributed – not that the original data are normally distributed. If the sample size is “largish” (n>10 for mildly symmetric data, n>30 for fairly skewed data) then the central limit theorem can prove that the means will be nearly normally distributed. Thus the ANOVA and t-Test can be used even for nonnormal data.
Nonparametric tests should rarely be needed when testing means. If the sample sizes are so small that normality doesn’t hold, then it is probably small enough that the likelihood of a Type II error is quite large anyway.
The forum ‘General’ is closed to new topics and replies.