06-21-2016 04:51 PM
I have run the following two-way mixed ANOVA and now need to report the results in standard APA style notation:
proc mixed data=pk(where=(paramcd="LNAUCINFD"));
class allonum dose sex;
model aval=dose sex dose*sex/ ddfm=kr;
random intercept / subject=allonum;
estimate "10 mg vs 25 mg" dose -1 1 / cl alpha=0.1;
estimate "Male vs Female" sex -1 1 / cl alpha=0.1;
lsmeans dose / cl alpha=0.1;
lsmeans sex / cl alpha=0.1;
lsmeans dose*sex / cl alpha=0.1;
When reporting ANOVA results in standard APA style, you are required to report the degrees of freedom of the error term as part of the standard notation. However, since there is no error term in this model, what is the best way to report the results of my ANOVA? Any help would be greatly appreciated!
06-28-2016 09:58 AM
Unfortunately, APA style doesn't recognize mixed models. It would also like you to report effect sizes, which are not as easily calculated. However, unless your model is overspecified, the table Type 3 Tests of Fixed Effects should report denominator degrees of freedom, which would be associated with error df in a non-mixed model. With the Kenward-Rogers adjustment, this value may differ for each of the tests, depending on homogeneity of variance and equality of cell sizes.