What is the difference between covariate with fixed effect and random effect?

Question

[Emy]

Hi everyone,

I am really confused about some concepts. I am reading papers on the subjects of mixed linear model for GWAS and variance components.

What is the difference between covariate with fixed effect and random effect? A covariate is separate in a model? or is it a part of those effects? what is its role in a model?

Thanks

Answer 1

[atorreso]

Check out this answer at stats.stackexchange for basic definitions:

https://stats.stackexchange.com/a/4702

A covariate is part of the model, and can be either a fixed effect or a random effect. In a GWAS context, your SNP would be a fixed effect, while random effects can include variables such as "Family" if your samples come from different families but some samples come from the same family, genetic structure if you are trying to correct for population stratification, etc.