在MERGE和SET之间选择以在SAS中合并数据

Question

I have a general question on methodology. How do I know whether match-merging (MERGE) or interleaving (SET) is better for combining datasets? If I have two related datasets, that seem to contain many of the same variables (but not all), but I don't know whether or not the information in said variables is the same, which is better?

Is there some sort of general rule of deciding which is better?

Thanks for your advice.

Answer 1

There really isn't a good answer to this question; there are fundamental differences between what "merging" and "interleaving" do. Take a few minutes and read the example in the SAS Concepts manual, particularly here .

Answer 2

I think that's a question that is very much specific to your data and what you are trying to achieve. You shouldn't combine the datasets at all until you know enough about the data to know whether or not you can combine them (set) or want to match-merge them. There cannot be a general rule because it simply depends on your data - if I had two datasets

data have_1;
input x y;
datalines;
1 2
2 3
3 4
;;;;
run;

data have_2;
input x y z;
datalines;
1 2 3
2 3 4
3 4 5 
;;;;
run;

You could guess that have_1 and have_2 are the same observations, just with an additional variable z; but they easily could be different observations as well. If I told you that 'x' was the unique identifier, then you would suspect these are the same records; but if I told you that 'x' and 'y' were qualitative features, then they could easily be different observations that happen to be similar qualitatively.

The point here: know your data before doing anything with it. If you don't know your data you shouldn't be working with it in the first place.