My research is to evaluate change of emotional responses of participants, after performing two types of game activities. Two groups of participants, Group A and Group B, will be assigned to Activity A and Activity B respectively. The research design does not cater for a control group.

I plan to measure the baseline emotions (using the Mood Questionnaire) [1] of all the participants. The arousal and valence components will be measured using the Self Assessment Manikin (SAM) questionnaire [2]

So now, I have pre-test responses from Mood Questionnaire (baseline emotional responses), and the post-test responses from SAM.

I am not sure how to proceed with the data from each instrument. For instance, how do I obtain a change in emotional responses from each subject? How do we normalize the baseline emotions in this case?

Mood Questionnaire available at [1] https://www.metheval.uni-jena.de/mdbf.php

SAM questionnaire available at [2] http://irtel.uni-mannheim.de/pxlab/demos/index_SAM.html


1 Answer 1


Normally, I'd recommend using the same mood measure pre and post manipulation. This allows you to assess the size and direction of change. It also likely removes more error variance from the post-test measure (which would increase statistical power), because using the same measure is likely to increase the correlation between pre and post.

That said, if you've used different measures pre and post, then you could perform an ANCOVA with group as independent variable, baseline mood as the covariate, and post-test mood as the dependent variable.

A slight possibility is that you could map scores on both tests to a common scale. This might allow you to make qualified claims about whether emotions increased or decreased. That said, this would require that you have access to additional data on people who completed both tests in a non-experimental context. You could then use the means and correlations between the two tests to develop a mapping to a common scale.

  • $\begingroup$ Appreciate for the kind clarification and direction. Actually, these are the data that I have at hand right now, pre-test from Mood Questionnaire and post-test from SAM. I will proceed with the idea of the ANCOVA, that you suggest. (1) I am stuck at the stage of mapping the items from Mood to SAM, though I will be looking at the Arousal related items in Mood (Calm-Nervous) and Valence related items in Mood (Good-Bad). (2) What kind of additional data do I need for the same subjects in the non-expt context? I do not have a control group. Thx $\endgroup$
    – user39531
    Commented May 26, 2014 at 1:15
  • $\begingroup$ In general, you'd be looking for pre-test covariates in your ANCOVA that predict your post-test dependent variables well. Greater prediction by covariates means greater statistical control for error variance which generally means more statistical power. $\endgroup$ Commented May 26, 2014 at 3:43
  • $\begingroup$ Can you please provide me with an example what pre-test covariates I should be looking for. Thanks $\endgroup$
    – user39531
    Commented May 26, 2014 at 3:47
  • $\begingroup$ @user39531 An example would be you have a measure of positive mood using test A and test B; Test a was measured pre-test and test b was measured post-test. So you use Test A as the covariate and Test B as the DV. $\endgroup$ Commented May 26, 2014 at 6:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.