Add this after rsubmit   %include '!SASROOT/wrdslib.sas' ;  ... View more

It depends partly on the procedure you're using. Some allow for CLASS variables, which then create the dummy variables.  Otherwise you need to create them. Here's a post and a reference to Ricks blog that goes over ways to create the dummy variables. https://communities.sas.com/t5/SAS-Statistical-Procedures/How-to-create-dummy-variables-Categorical-Variables/m-p/258702/highlight/true#M13685 ... View more

Hi @HannahPhD,   If I had to investigate this situation, I would create a temporary dataset from each of the three original tables by restricting them to the key variables.   Example: data keyall; set allanalyst(keep=permno year quarter); run; Using the resulting three comparably small datasets you can easily check for duplicate key combinations perform frequency counts sort by key variables (if necessary) eliminate duplicates (if any) create a "membership profile" for each key combination (e.g. "101" means "contained in the first and third dataset, not in the second") perform "test merges" The results of these operations will give you a clear picture of the situation and hence guide you how to proceed with your "big" datasets. You will then know for sure in advance how many observations to expect from your intended joins/merges.   You should also make sure that there are no common variables except the key variables or, if there are some, that they are handled properly. ... View more