t Test for Dependent (or Matched, or Correlated) Samples.
This, also, is sometimes called a t test for repeated measures, when there are two measures taken on each subject (as in a pre-test post-test situation).
Some of examples where this test would be indicated include:
A study of husbands' and wifes' political attitudes.
An examination of pre-intervention vs. post-intervention outcomes.
A comparison of two treatments on pairs of individuals matched on intelligence.
A brief, yet detailed, description along with a Java applet for computing the test is found in the following link.
Use this test to analyze the following scenario (provided in Howell, 2007, p185-7).
Everitt investigated the use of family therapy as a treatment for anorexia. Seventeen girls participated in the therapy. Each was weighed before and and after treatment. A .txt file of their weights can be found HERE.
State a reasonable research and null hypotheses for this study. Perform the test and write-up the results using an appropriate statistical clause. I may be helpful to plot the data beforehand.
If you are up for a challenge, see if you can program this test in Excel.