For one campaign, I have X as independent variable, Y as dependent variable, i.e., X is income, Y is credit score. I can use X to predict Y with proper model. Now I want to know how Y would perform if I added one more independent variable Z into the model, i.e., age. The challenge is I don't have any data on Z, that is to say, I don't know the relationship between Y and Z for this campaign. But I do have data that illustrates the relationship between Y and Z for other campaigns. Is there any model that can be used to answer this question? Any thought is helpful.
A possible solution is to pick a set of Z data available from other campaigns and use it for this one. If you believe that there is a relationship between Y and Z and there are theoretically sound reasons for it, then the particular sample of Z you use is irrelevant, assuming the sample is large enough.
Now, if this bugs you and you don't want to pick a set of Z data arbitrarily, just use an average of all sets available.
Hope this helps, Arman