Two effects, but not enough samples - what statistics can I do?

#1
Dear all,

I made measurements of oxygen consumption (VO2) during steady state running with 3 different shoes. The runs were performed in 5 series of 3 runs, where each of the 3 successive runs was performed with different shoes. From the results of VO2 it is seen that the place of the runs within a series has an important impact. Can I do any statistics if I have at least one sample of each combination (each shoe was used at least once in the first run of a series, in the second and the third run of a series)?
So I have two effects (shoe and place of the run), but for each combination of effects there is 1 to 3 samples.

Any help is appreciated. Thank you very much.

Jurij
 

katxt

Well-Known Member
#2
If you had done six runs, you could have made sure that each shoe was used twice in each run place - a nice balanced experiment.
However, with your data, if I interpret your description right, most stats programs will do a two way anova on unbalanced data. If that fails you can do a General Linear Model which will sort all this out as well as it can be done.
 
#3
The primary goal was to execute 6 series of three runs (to have an balanced experiment), but unfortunately some results got lost. Ok, thank you for your answer. I'll try two way ANOVA. Regards,

JH