When you have two sets of values that are paired, using a compare means test would throw away the information that the values are paired. Instead, construct a new attribute that computes the differences for each case. Then use a test mean to decide if the difference is significantly different from zero (or whatever value you hypothesize).
This is shown below for an example in which we have a random sample of student test scores before and after a unit of instruction.
Key Curriculum Press/Key College Publishing www.keypress.com/fathom |