I have a set of variables and I want to test their bivariate ot multivariate normal distribution, but I didn't know how! I want a method in excel or a statistical software such as Minitab or SPSS or SAS.
If you must use MS Excel, you should be able to use "COVAR" in Office 2007 and onwards to find covariance. As it happens, SPSS has better/sophisticated features to help with these.
I have no problem with calculating covaraince, I just want to know what should I calculate to carrying out this test?
thank you
It depends on the test you use. I suppose you are running a GLM univariate test in spss. 1. Select the predictive value and residuals under SAVE option. 2. run a scatter plot under the Graph test with the residuals score and the predictive value.
Abdolghani Abdollahimohammad
thank you very much, but I need multivariate normal test (not univariate!).
What if you have 2 or more variables and you want to carry out MANOVA not ANOVA??
Use GLM multivariate and follow the previous post procedure. You can also run the following syntax, only write your variable names instead of DEPENDENT VAR and INDEPENDENT VAR:
GLM DEPENDENT VAR BY INDEPENDENT VAR
/METHOD=SSTYPE(3)
/INTERCEPT=INCLUDE
/SAVE=PRED RESID
/CRITERIA=ALPHA(.05)
/DESIGN= INDEPENDENT VAR.
GRAPH
/SCATTERPLOT(BIVAR)=PRE_1 WITH RES_1
/MISSING=LISTWISE.
Paul R. Yarnold
Thank you very much.....
it was very interesting.....
- Badges
- Science topic