I am currently doing research on the moderation effect of the Aim on factors (independent variables) that influence technology acceptance (dependent variable). My research is as follows:
- Group A had to answer questions about Aim A
- Group B had to answer questions about Aim B
Additional information:
- The questions the groups answered used a Likert-scale and were identical.
Each questionnaire item naturally return a different variable name for each group. For example, acceptance (A) is returned in group A as A.A and in group B as A.B.
Can I now pool these different naming variables in one overarching variable to test for the moderation effect of the Aim on the relation between the independent variables that were measured and the dependent variable ?
Thanks in advance for helping me!
The best you can do is to run separate analyses by sub-sample and force the effects for the independent variable(s) to be the same for both sub-samples.
Hello Ellen,
If you randomized the groups, then you may confidently combine the two sets and use aim type (a/b) as a moderator.
If you had intact groups, you'll have to grit your teeth (metaphorically speaking) and trust that the responses are not being influenced by other, unmeasured variables in such a way that they (responses across groups) are not directly comparable. For some degree of support, you could compare the groups on common variables which you may have collected (including demographic attributes). To the degree that you don't see differences, that helps bolster the argument of groups being comparable (or, at least, not wildly different so far as you're able to discern).
Good luck with your work.
- Badges
- Science topic