I am to do a project for an inferential statistics class and I would like some basic guidance.

I am to propose a research question. My proposed research question is "Does drinking stimulants such as coffee improve a student's grades?"

It was suggested to me that my population could be the students at my school. I will impliment simple random sampling by devising two groups of students, one being a group who often consume stimulants like coffee and another who seldomly or never consume a stimulant such as coffee. Upon gathering the data of each group I will need to compare the average grades for each group using independent samples t-test for both respective groups. My Hypothesis test will be a two independent samples test for mean.

My question is what are the limitations? (How could bias and/or confounding variables could limit this study?)

Also what graph do you think would be best served to visually display the data?