Test of Significance for paired data

[GreggB]

I have pre and post-test data that is expressed in percentiles. Is there a test of significance (maybe similar to McNemars) that would test to see if the gains (or losses) are significant at a certain confidence level? Essentially, did Instructor A's students make significant gains, significant losses, or neither?

instructorID         studentID    Pre    Post

         A                     1              42     80

         A                     2              99     90

         B                     3              14    20

         B                     4               50   50

[ballardw · Accepted solution]

I would start with something like which would require the data to be sorted by instructorid.

Proc ttest data=have;

   by instructorid;

   paired pre*post;

run;

Which will show summary of the differences between the variables for each instructor as well as a ttest for the difference being different from zero.

There are a number of options depending on specific questions.