I created a survey and posted on the university meme site, got 200 responses. The goal of this survey was to determine relationship between GPA and hours/week worked in a job for full time uni students. I already filtered out the data to exclude things like jobs which you can study on job vs not, no freshmen, etc. Here is just an initial scatterplot.enter image description here

What are some tests I can do to test for relations between gpa and hrs/week worked for 1. Both stem and non-stem majors and 2. for differences in dirstribution for non-stem and stem majors?

There's a paper which found students who work 12-15hrs/week have slightly higher grades than those who don't, so I thinking about looking at how they analyzed data. Also thinking about checking out 2D Kolmogorov Smirnov test, if it's applicable.

  • $\begingroup$ Just by eye, it seems both scatterplots show contrentations of points at lower right, which might indicate negative correlations. Have you computed correlations? Also, GPAs seem lower for STEM majors. Have you done a two-sample t test? $\endgroup$
    – BruceET
    Jul 21, 2019 at 14:35
  • $\begingroup$ Nope, because while I am a statistics major (undergrad), I only begin taking real stat classes starting this fall :/. I'll try the things you said. also, would an anova test of gpa of students who work {<20hrs/week, >20hrs/week} make sense? I've tried it just using scipy and it seems to show p-value of .03 $\endgroup$ Jul 22, 2019 at 8:37
  • 1
    $\begingroup$ For groups <20 hrs/wk vs >20 hrs/wk: seems to be a 2-sample t test, maybe Welch 2-sample t test because there's no guarantee of equal variances. But for such small sample sizes I'd check normality. (Maybe t test on ranks if far from normal.) $\endgroup$
    – BruceET
    Jul 22, 2019 at 9:11


Your Answer

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