Dependent/Matched/Paired Sample T-tests

What are the steps for conducting a matched/paired sample t-test?

1) establish your null hypothesis 2) Assess normality (k-s test, measures of central tendency etc.) on the mean differences NOT on each group seperately. 3) If Normality assumptions are met by the mean diff within the groups then RUN A DEPENDENT SAMPLES T- TEST **(variances do not have be checked from same population) 5) look at your test statistic to determine whether the rest is statistically significant or not. 6) Report results: t = ____, df= ___, and p = _____. and report result in the context of the problem. DON't forget... IN THE POPULATION.

What are the assumptions of the matched or paired sample t-test?

1) that the scores on the dependent variable are normally distributed IN THE POPULATION. 2) Independence within groups/independence within the sample ( we confidence that the researchers did not systematically link participants within their sample other than the pairing that defines this test.

What are some examples of a matched/paired samples t-test?

A sample of married couples is studied. Both individuals of each couple, complete a relationship satisfaction scale. The researchers want to know of there is a difference between moves and husbands mean reports IN THE POPULATION. OR A sample of clients is assessed with the BDI pre and post treatment. Time 1 and time 2.

What is the alternative hypothesis for a paired sample t-t test in words (one-tailed test)?

Ha: IN THE POPULATION the mean difference between he paired scores is greater than (or less than) zero. So our conclusion is that the mean post-test score is greater/lower than the mean pre-test score IN THE POPULATION. i.e According to the pre- and post treatment reports on the BDI, post treatment scores are significantly lower than the pre test scores. So we conclude that the mean of the post test scores is lower than the mean of pretest scores IN THE POPULATION.

What is the alternative hypothesis for a paired sample t-t test in words (two-tailed test)?

Ha: In the population the mean difference between the paired scores is not equal to zero. i.e according to pre- and post- treatment scores on the BDI, there is a significant difference between pre- and post treatment scores. Our conclusion is that the means differ IN THE POPULATION.

What is the alternative hypothesis for a paired sample t-t test in statistical notation?

Ha: Mean diff does not = 0 (two -tailed) Ha: Mean diff > 0 or < 0 (one tailed)

What is the null hypothesis for a paired sample t-t test in statistical notation?

Ho: Mean diff = 0 Or

What is the purpose of the matched or paired sample t-test?

It is also known as a dependent sample t-test. Use this test when the members of the samples are related, on a one-to-one basis in some way.