Doing a statistical t-test to assess if 2 groups are significantly different than each other.

Question

For most frequently used t-tests are paired t-tests and independent-sample t-tests. In a paired t-test one subject is tested twice and that is good as it removes individual differences in an independent sample t-test only 1 piece of data is taken for a subject. I use SPSS to compute and help students with t-tests.

Answer 1

A paired t-test is not appropriate for comparing 2 independent groups. A 2-sample t-test would be used in that situation. An example of a 2-sample test scenario would be heights of 15 year-old females vs 15 year-old males who attend the same school.

A paired t-test is appropriate when comparing data on the same experimental units, e.g. change in student test scores or size of a cancerous tumor, where an initial reading was obtained prior to application of a test agent and the second at some time afterwards.

For the students the test agent could be extra tutoring between tests and for the cancer it could be use of radiation therapy after getting an initial, baseline, measurement of tumor size.

Be very careful that you use the right test given the situation!