Hi Brandon, if I understand your experimental design, it looks like a nested design (sections nested on treatments), then you must run a nested design anova with four replicates, instead one-way anova.

Best regards,

Federico Nave

Dear Brandon,

I understand that you have all 5 sections form the animals subjected to all 3 types of treatment? If yes, try two-way repeated-measures ANOVA. There are two factors: within subject (sections) and between subject (treatment). If you are looking for the effect of treatment to all the sections - check the effect of treatment only, but you may also look for the difference in sections under different treatments - then look for the interaction in ANOVA. Although one-way ANOVA sounds also ok if you look for the effect of treatment only, the proposed analytical design gives you more possibilities.

Good luck,

Michal

Thanks all!

Actually, the animal groups were individually subjected to treatment (control group, treatment 1 group, treatment 2 group). The brain region I'm interested in spans multiple sections and I'm counting the number of neurons activated by treatment in each section that contains that region. Thanks again for all your help, it's pointed me in the right direction.

Brandon

If for each individual the same number of sections and the same kinds of sections (brain regions) - at least approximately - are taken, then you can see the set of the sections per animal as one larger section. This is what you do be "averaging" the values of the sectione per animal.

Whether "ANOVA" is ok depends on the numbers of counts you actually have (on the "large" section, so the total per individual). Actually, as you have counts, you ought use a Poisson model or a negative binomial model. However, for large counts these models can be well approximated by the normal model (which is the basis of an ANOVA).