Hello
Wanted to find out if Its ok to compare beta coefficients of a predictor variable from two different regression models. Is there a way to standardise them before comparing or can i just go ahead and compare them. Please help
I don’t think that you can or should compare them. The meanings of the coefficients depend on the whole model, and they will have different meanings in different models. There might be exceptions, but I would generally not compare coefficient values between different models.
Systematic reviews and meta-analyses are, essentially, the practice of comparing beta coefficients.
Do you have a specific use case?
What i have are results from my model which has very similar variables including the output variables with another published study. It is however in a different population and different setting. I was very much tempted to compare what they had found by coefficient with my model findings as opposed to me just discussing coefficients within my model. I hope i have answered your question.
It is however in a different population and different setting
You are definitely asking the right questions. Study design and generalization to other populations are key, along with modelling approaches. I’d suggest reviewing an introduction to meta-analyses. I can’t think of the best reference for your question but will post if one comes to mind.
2 Likes