What specific type of GLM to use?

#1
I'm trying to figure out what type of distribution (Poisson or binomial) to use to measure differences in proportions using logistic glm. Basically I want to know whether a proportion of active neurons (compared to inactive) in one group is significantly different from the proportion of active neurons in another group.

Experiment info:
- Measured total number of neurons in a region
- Measured active number of those neurons
- Want to compare the proportion of active neurons in one group (control) to active neurons in another group (shock)

Bonus: what would the name of the model be using this approach with a repeated measures design? So comparing between 3 groups within 3 regions.
 

hlsmith

Omega Contributor
#2
Would it be appropriate to say your outcome is a percentage between 0 and 1? If so can you post histograms of the outcome values.


You may be able to use linear regression if the value is centrally located with minimal dispersion, otherwise beta regression may be an option.