How to test the effect of pre-post, treatment-control intervention with three age groups, two dependent variables and gender effects?

The hypothesis of my research project is “a certain training improves the ability in Subject A” .

Sample details

Groups: 1A (25), 1B (21), 2A (25), 2B(27), 3A(23), 3B (23). The number within brackets represent the number in each group.

Control/experimental: 1A, 2A, 3A are experimental groups and 1B, 2B, 3B are control groups.

Group labelled (1A, 1B); (2A,2B); (3A,3B) fall under 3 different age groups respectively.

Data collected was as follows:

1. Pre-test and post-test scores for each group taken using subject A ability test
2. Subject A and Subject B achievement scores for each grade (5-6 scores taken at different intervals in an academic year of class unit tests given by the teacher).

Statistical analysis

With limited knowledge in experimental statistics, pre-test and post-test results for all the groups were taken to test Subject A ability. I conducted a paired t-test for each group separately.

Subject B unit test scores were also taken just to have a check on the effect of training on one other subject achievement, different from subject A.

Questions :

Apart from paired t-test what other statistical tool will help in line with my hypothesis?
1. I wish to compare the Subject A ability scores between control and experiment groups
2. Compare the Subject A ability scores on the 3 different age groups
3. Compare the unit test scores conducted in class separately for subject A and Subject B
4. Compare ability scores on the subject A for male and female in each group and overall

PS: I will be using SPSS for statistical analysis.

• You might want to find a more descriptive title to attract more attention to your question. – Gala Jun 24 '13 at 6:28
• See stats.stackexchange.com/questions/3466/… for a previous question on a related topic. It should at least address all your questions related to subject A. – Gala Jun 24 '13 at 6:29
• Thank you for your reply, I could not ask a specific questions as there were many sub questions. Thank you for the link given , I will read through and get back. – vittal Jun 24 '13 at 7:40

For the most part it is similar to a standard pre-post treatment-control design. As such most of the responses to this question would be relevant. The main difference is that you have three age groups.

I don't quite understand how you can have unit scores taken at different intervals and still have a pure pre-post design.

Nonetheless, one way of implementing your analysis is as a 3 x 2 x 2 mixed ANOVA with 3 levels of age, 2 levels of condition, and 2 levels of time.

This can readily be tested using a GLM - repeated measures in SPSS with time as your repeated measures factors and age and condition as your between subjects variables.

Furthermore, it sounds like you have two different dependent variables. One for Subject A and one for subject B. If you are happy you could just run two separate ANOVAs, one for each DV.