MaggieZ

The question is: "There is a belief among students that if they go to office hours, it will increase their midterm exam score by 5 points. Evaluate this belief by means of an appropriate t test." The data are presented as the midterm exam scores in one column (quantitative variable), of 100 students, and ‘went to office hours or not’ in another column (categorical variable, which can be recoded as a dummy variable with 0 as went and 1 as didn’t go), which describes each of the 100 students. I added two new columns, copying and pasting the scores of those people who went to office hours into one of the columns and the scores of those who didn’t into the second column, for the analysis. That is, one column has the title 'scores on midterm for people who went to office hours', the second one 'scores on midterm for people who did not go to office hours'. I think the H0 is then that mean1 minus mean 2 = 5, and the alternative hypothesis is mean1 minus mean2 is not 5. My question is: Do I use the paired samples t test here or an independent samples t test (with either equal or unequal variance (it's unknown whether the variances are equal or not))? Thank you!