The tests have a number of subtests (eg Malin's Intelligence Scale for Indian Children, which has 11 subtests) and various parameters ( eg Wisconsin Card Sorting Tests , has a number of parameters on which the groups were compared) on which both the groups have been compared.
for using bonferroni correction , how will we calculate the number of tests. will it be all the subtests irrespective of the number of parameters?
plz help in understanding this.
Bonferroni is a/n where n is the number of pairwise comparisons. Sidak correction is just a s easy today 1-(1-a)^(1/n), and might be considered slightly better. But why not use an ANOVA, and then some a post-hoc procedure?
The Bonferoni test values are significant only if the dispersions within the groups are homogeneous.
Homogeneity of dispersions can be calculated with the Levene test programmed in the SPSS program in the Explore: Plots box (Spread vs Level with Levene Test).
Another way to compute the homogeneity of the variant is to tick the Homogeneity of variance option in the One Way ANOVA box: Options.
If the Levene test does not confirm the homogeneity of the variant then the significant values will be those of the Tamhane test.
- Badges
- Science topic
- Similar topics
- Medicine
- Epi
- Age Groups
- Children