We have two groups, and one of the groups is offered a group therapy treatment, while the other group is on the waitlist. Before starting the treatment, both groups were asked to complete some measures. The problem is that the groups' mean scores on these measures differ statistically significantly from each other at this time point. How is it possible to control for this difference while comparing the groups' scores on the same measures after the treatment is completed?
Assuming that the "measures" you are talking about are not the primary outcomes of the studies, then you would adjust for them in the statistical analysis by including them in the the regression model. If your sample size is small, or there are a lot of measures, then consider a propensity score adjustment. If the measures are the baseline values of the outcomes of interest, then again thy simply get included in the analysis as covariates.
Hi
1. What are the research questions/hypotheses of your research.
2. Selecting responses from respondents assumes that the respondents are almost equally likely, i.e. the selection of respondents should be done randomly.
3. Studying two groups of of two groups of patients, one group was offered therapy treatment and the other is not. This is not a questionnaire, it is an experiment to study the effect of the therapy treatment of patients.
4. In order to develop the right statistical test of significance, the sample size need to be given.
Regards,
Zuhair
@ Z. A. Al-Hemyari
1. The research question is "Does group therapy influence the levels of depression?"
2. The respondents were selected randomly, that is why there are now group differences in some of the questionnaires at baseline. If they were not seleceted randomly, the scores would not differ from each other.
3. Exactly, the study consists of two groups, one experimental group, and one control group. The experimental group is offered group therapy, the control group is on the waitlist. And the study investigates the effect of therapy, which is measured through the scores on a depression scale. But the problem is that, the groups did not have the same level of depression at baseline, so interpreting the change on the scales is problematic.
4. Both groups consist of 50 participants.
In light of this information, any ideas on how to control for the levels of depressive severity at baseline?
@ Adrian Esterman
The scales are the primary outcome of the study. And I would like to compare the experimantal and control groups on these scales (between-group analyses). I am not planning on using regression analysis for this study, I am planning on using independent groups' t test, but since the groups differ on the severity of depression at baseline, I need some statistical method to control for this difference. That is my problem.
Both groups consist of 50 participants, so yes, the sample size is small, but 100 participants would have enough statistical power to detect between-group differences.
How do I add the measures as covariates when I plan on using t tests? Is there a way to do that? Should I use ANOVA, and add the baseline scores as covariates?
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS