0
$\begingroup$

I'm sorry if this question is vague: I don't yet understand stats terminology very well.

Let's say I am interested in a measurement X. Imagine it is neuron diameter. I want to know if brain regions A, and B differ in average neuron diameter. I have 100s of measurements of neuron diameters sampled from regions A and B from different individuals of the same species: animals U, V, and W.

What is an appropriate way to check if the differences between regions A and B are statistically significant? Can I just lump all the individuals together for each region, so I compare 2 distributions? Or is there some multi-level approach that I should do?

$\endgroup$
0
$\begingroup$

I do not have enough points to comment and clarify, so I will make some assumptions.

Scenario 1

Each animal is measured twice, region A and region B. The animals come from three different species, U, V, and W. This can be a repeated-measures ANOVA (region) with three levels of one factor, species. You check if the difference in regions in statistically significant. You can also check if the interaction between species and region is statistically significant, maybe the dynamic changes depending on the species you are looking at.

Other scenarios

I would rule out any multi-level approaches to your analysis, as your cluster sizes are just too few. You have two regions, nope, and three species, nope. Unless each animal is measured several times, and then we can cluster at the animal level, but it would have to be at least three times for the multilevel approach to make sense. Any two points will form a line regardless of the trend in our data. Three points then we can start reject (establish) a linear trend in our data.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.