0

I am working with paneldata that looks something like this:

enter image description here

I am going to perform a t-test in SAS 9.4 to find out if there is a significant change in var1 from 2014 to 2016, and I am assuming that I have to use a paired t-test, since I have several an observation in both 2014 and 2016 for each individual (ID).

My question is, can this be done in SAS, when I am using panel data like the one I have shown? Or do I need to create a a wide dataset with one variable containing the data from 2014 and one variable containing the data from 2016? I know that I have to do that in STATA, but maybe I don't have to change my entire dataset to do this in SAS?

BeeBee
  • 55
  • 6
  • Please ensure to 1. Post image as text, 2. Post anything you've tried to solve this issue on your own. As a newbie you get a few passes, but please do try in the future. Instructions on how to ask a question can be found here [ask] – Reeza Nov 12 '18 at 20:33

1 Answers1

0

You will have to transpose your data to to a paired t-test. You can use PROC TRANSPOSE though. 

 *sort for transpose;
 proc sort data=have; by id year; run;

 *reformat from long to wide;
 proc transpose data=have out=want prefix=Year_;
 by ID;
 ID Year; 
 Var Var1;
 run;

 *Paired T-Test;
 proc ttest data=want;
 paired Year_2014*Year_2016;
 run;

PS. Please include your data as text not an image in the future. We cannot write code off an image and I'm not typing out your data, so at present this is untested but should work.

Reeza
  • 20,510
  • 4
  • 21
  • 38