1
$\begingroup$

I am conducting a pre- and post-test to see if student perceptions change following an treatment. I have two variables that could be treated as continuous, which will be measured both times - motivation and self-efficacy - for a treatment and control group (third variable, I suppose). I'm interested in knowing if there is a higher motivation and self-efficacy combination over time.

1. Chi-square based on expected values

One way I've thought to look at this is reducing it to a 2 x 2 table for "high" and "low" on each variable like this from the pre-test scores:

enter image description here

Then I could use a chi-square test for the treatment group's second test with the "expected values" being from the first test. This would show me if there is a change but not necessarily where it is. It also wouldn't tell me the change for specific students - in other words, if every student moved but the totals were still the same I wouldn't know.

2. Multiple t-tests

A second option would be to perform two t-tests. I would first get the change-scores for each student. Then, compare the change in motivation between the control and treatment using an independent t-test, and do the same for self-efficacy. This would tell me whether the treatment had an stronger impact on the variables, but not take into account both at the same time.

What type of analysis could I perform to tell if there is a positive change in these two variables following the experiment?

$\endgroup$

1 Answer 1

1
$\begingroup$

Option 1 means dichotomizing variables that aren't already dichotomous, which is usually a bad idea.

If you're sure you want to do significance testing, Option 2 is a pretty good choice, although I would replace the $t$-tests with Wilcoxon rank-sum tests, since however you're measuring motivation and self-efficacy, I doubt the scores are normally distributed. I'm not sure if there's an easy way to account for possible dependence between the dependent variables.

$\endgroup$

Your Answer

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

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