Assumptions/Violations/Concerns for Basic T-Test for Difference between Means?

[techsassy]

Hello, I work for a college which is interested in exploring whether or not students in Fall 2019 had a better/worse GPA, on average, compared to students in Fall 2020, by course.

Looking at, say, Psychology 101, is it valid for me to do a t-test comparing the mean GPA for Psychology 101 in Fall 2019 vs. Fall 2020, using the entire population of students who took these courses? Are there any special assumptions/violations that I should be concerned about, or is this a pretty safe statistical exercise?

I've taken many statistics courses throughout my life, but it's been many years and I am rusty, so any guidance is much appreciated. Thank you!

 

[hlsmith]

You may want to make sure there are not the same student in both samples. Also, if there are differences between the groups, you may need to control for those in a regression model (e.g., what year they are, age, major, etc.). Since if there so happens to be imbalances in background characteristics, you could spuriously associate those with calendar year.

 

[techsassy]

Thank you for your thoughts, hlsmith!