Logistic model in overall population or in sub-population?


I am not sure if this is the right place to ask my question. I really appreciate your help because I am really a beginner in bio statistics.

I want to develop a logistic model in a population. The outcome is yes/no and I have 12 predictors. I have no problem in the process of developing the model but the point is that I can divide my population into 2 different sub-population based on one of the predictors. these 2 sub-populations share all the same characteristics (predictors). I mean these 2 populations have been tested for the same variables but they have different exposures (exposed or non-exposed). my question is that how I can decide whether or not to develop one model in overall population or different models in the subsets??? Do I really need to develop separate models??? How can I descriptively infer that from the distribution of the predictors in overall population or in the sub-population?

Thank you a lot!
Hello and thanks for the answer.

I may not use the correct terms. I mean the population can be divided in two different populations base on one of the predictors( exposed to A or B) and then you can develop a model for each of them. which one to choose? developing a model in overall population or in sub-groups and then compare them?


Omega Contributor
The only difference is one exposure variable, then I would just run a single model and control for that variable as well. Does that subpopulation have its own independent risk for the outcome beyond the exposure status? If so it could get masked and running a single model may not be the best approach. Though running two you would have to interpret the exposure with reservations.