Independent t-test for two samples

#1
Hello,

I am doing a bachelor project about empathy's class efficiency at school. To do so, we gave an "empathy course" to a class of 10 years old children.
We then gave to the children of this class a test to measure their level of empathy and compared those results to the ones of another class who did not have an empathy class.
I want to compare the means of those two classes but I am not sure about which test to use. I was thinking using the "Independent t-test for two samples" (the
two samples being : 1) the class who took the empathy classes and 2) the class who had not empathy class).

What do you think of using this test?

Thank you very much

jamo
 

Karabiner

TS Contributor
#2
If you have no baselin data (empathy before the empathy course) from both classes, then your results will be nearly
uninterpretable with regard to possible classe differences. Any differences could have existed befor the course.

Moreover, if you want to compare children from to school classes, then the children within a class are clearly no
independent observations, they will heavily influence each other. So you either have n=2 (classes), which is too few
for statistical tests, or you have to perform a t-test with biased results. So maybe you should talk to your supervisor.

With kind regards

Karabiner