Question

A friend of mine does program evaluations for counseling centers that deal with juvenile delinquents. As such, he does many simple statistical analyses that compare some “pre-intervention” behavioral trait to that of the behavioral trait “post-intervention”. For example, say that some intervention is aimed at reducing aggression in adolescent boys. My friend might then quantify pre-intervention aggression (also known as “baseline” aggression) by administering something like The Aggression Scale developed by Orpinas and Frankowski (2001), which yields a value between 0 and 66, with greater numbers indicating higher aggression. Then, after the adolescent boys complete the intervention program, my friend would then again administer The Aggression Scale to determine the effects of the intervention program on aggression (i.e., “post-intervention”). The data might look something like in the below table:

Participant	Pre-Intervention Aggression	Post-Intervention Aggression
1	35	29
2	42	41
3	45	43
4	39	27
5	45	40
6	43	34

Using the above data, answer the following questions

1. Which type of t-test (one-sample, independent samples, related samples) is appropriate for determining if aggression decreased from pre-intervention to post-intervention? Why?

2. State the alternative and null hypotheses (using the statistical notation for stating mathematical relationships) that represents the prediction that aggression will decrease from pre-intervention to post-intervention in these adolescent boys.

3. Calculate standard error of the mean difference and t_obt.

4. Using an alpha of 0.05, what is t_crit? Would you reject or fail to reject the null hypothesis? Why?

Answer 1