1
$\begingroup$

I have survey data from students who were asked to rate (yes/no) if a math equation was useful for a set of four educational purposes. Students were ask to answer this survey for five different math equations and they make up two groups based off of when they are taught in traditional course sequences.

I'd like to see how students' responses group together, so I'm planning on using some type of cluster analysis. However, I'm worried about the boolean nature of the data and the fact that students were asked about the same set of educational purposes for five different math equations. Maybe there is a relationship between subsets of student responses?

My question is, would it still be appropriate to do a cluster analysis given that my data is boolean and that there's a possible relationship between the features? If not, is there something more appropriate to get an understanding of what is the range of student thought across equations and educational purposes?

$\endgroup$
1
$\begingroup$

You can cluster with binary variables. For example hierarchical clustering with Jaccard index. But that needs to be carefully chosen.

I don't think you have a clustering problem. You seem to already have two groups, and you'll want to test if their answers differ significantly or not.

$\endgroup$
1
  • $\begingroup$ Thanks for the insight! I don't think I looked at it this way. $\endgroup$ – Linh Chau Dec 19 '19 at 20:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.