Dear List Members:
I have received a number of helpful suggestions regarding my 'Mixed ANOVA using Excel' dilemma (one between subjects factor and one repeated measure with unequal Ns). The consensus seems to be that Excel is not capable of performing such an analysis and that, if attempted, it would likely return inaccurate results. Some responders suggested that Excel is not the best choice for any analysis (including one who 'tested' Excel using obviously collinear variables and found that it could not detect the collinearity). As this gentleman pointed out, in the long run it is better to invest in a good statistics package than to discover after your analyses that you and your program have failed to detect defects in your model. Several suggestions for alternate statistics packages included Stata, SAS, SPSS, GENSTAT, SPLUS, Analyse-It, Statistica, and Modstat.
As for my dilemma, I did the analysis by hand using unweighted means [Keppel, G. (1991). Design and analysis: A researcher's handbook (3rd ed.) Englewood Cliffs, NJ: Prentice Hall] and the G-G correction (luckily it was only a preliminary analysis based on N=16). The whole time, I kept hearing my grad school professors' promises that after their courses I would never have to do another analysis by hand. Actually, it was kind of fun--I suppose it is useful to reacquaint ourselves with our (hand-held) calculators and (hard-copy) data sheets from time to time.
Thanks again to all who replied.
Sincerely,
Sara Pearson, Ph.D.
[log in to unmask]
Orthopaedic Research of Virginia
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|