[SPSS - GLM-ANOVA] GLM-ANOVA an appropriate choice of statistics?

#1
Dear Community!

I have two questions regarding the GLM-ANOVA and post hoc tests!

1 ) Is the GLM-ANOVA an appropriate choice of statistical analysis in my study?
2 ) Post hoc test: Fischer´s LSD vs. Bonferroni

Study design:
I have 4 groups of patients with different diagnosis in which I have collected 4 samples in a serie, 1 sample before surgery and 3 after. The groups of patients varies from n=5 to n=10.

I want to analyze the measured levels of circulating collagen type IV in the same group of patient regarding to the difference over time. I have used SPSS- univariate GLM-ANOVA to do this. Can I use the univariate GLM-ANOVA in this way? (I have tested the groups for normality. and they are normaly-distributed)

I provide data for one group to help clarify my question!

Post hoc I have used Fischers LSD test, which according to several sources is an outdated post hoc test and shouldn´t be used. When I use this I get significant differences, when I use the Bonferroni test I don´t. The Bonferroni test is more strict but I think it tends to overcorrect when the number of groups increases.

Thanks in advance,
Gruvaeus
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
Seems fine. Possible correction, you want your residuals to be normally distributed not the variables themselves (which can or cannot be ~ N).

Yes, Bonferroni is more cautious.