topic Re: ancova in Statistical Procedures

ancova

clcl — Sun, 06 Jan 2013 10:00:49 GMT

does anyone know if when i check for an interaction in the ancova model and i get a significant interaction, can i use the results i recived ?

if not what do i do ???

thanks

Re: ancova

PGStats — Sun, 06 Jan 2013 18:28:11 GMT

The presence of an interaction means that the effect of your covariate is not the same in every group (class variable). A comparison of the dependent variable will thus depend on the value of your covariate. What you can do is look at the ancova plot and compare your groups at a specific value of your covariate, as in this example (modified from SAS doc) :

data DrugTest;
   input Drug $ PreTreatment PostTreatment @@;
   datalines;
A 11 6   A 8 0   A 5 2   A 14 8   A 19 11
A 6 4   A 10 13   A 6 1   A 11 8   A 3 0
D 6 0   D 6 2   D 7 3   D 8 1   D 18 18
D 8 4   D 19 14   D 8 9   D 5 1   D 15 9
F 16 13   F 13 10   F 11 18   F 9 5   F 21 23
F 16 12   F 12 5   F 12 16   F 7 1   F 12 20
;

proc glm data=DrugTest plots(only)=(ancovaplot(clm));
   class Drug;
   model PostTreatment = Drug|PreTreatment;
   lsmeans Drug / at PreTreatment=15 pdiff;
run;

Re: ancova

clcl — Mon, 07 Jan 2013 06:20:12 GMT

In your attached example the interaction not significant.

I would like to know if it is a problem to use the ancova results when i get a significant interaction (is it an assumption that the slopes must be the same in order to using the ancova results?)?

thanks

Re: ancova

PGStats — Mon, 07 Jan 2013 13:06:33 GMT

You cannot fit an ANCOVA model (model PostTreatment = Drug PreTreatment;) when the interaction is significant. My example shows you what can be done when the interaction IS significant. - PG