Hi Amr
First define the response variable in your study, but the most common tests are as follows:
For continous data:
For categorical data:
I hope this help you.
A pre-test/post-test study like that is an intervention problem. You may compare the pre-test and post-test means using a t-test even though this is not ideal since time series data are necessarily correlated and not independent as required. Building an intervention model is the ideal thing. However the sample size of 20 appears too little for such a purpose.
I agree with Vicente Esparza Villalpando's list. However, each of these assumes that you can match the individual pre-test to the individual post-test. That is, the identity for each observation must have been recorded.
If the values are continuous, you determine the normality of the distribution by plotting histogram of the difference (Post-Pre score -- SPSS or any other software will do this for you) before using paired t test. If the difference is not normal, use Wilcoxon signed-ranks test as suggested by Esparza above.
thank you so much for your answers. My question now is pretest posttest one group is a quasi experimental design and why some researcher does not consider it as a quasi experimental design
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Modeling
- Analyses of Variance