I have a dataframe with 20 variables describing the situation of the students in a school while the do two different courses. Now, I am studying if there are multicollineality between some columns and I have the next question:

I have two categorical variables (with the same 4 categories) describing if a student has a good mod (1-bad mood, 4-awesome mood) while he does the two different courses:

 ~              goodmood_course1          goodmood_course2
 student1             0                       3
 student2             1                       1

I found that the two columns are almost the same:

    table( df$goodmood_course1 == df$goodmood_course2 )
    ... FALSE TRUE
        5     375

How can I deal with this multicollinarity? Could I remove one of the columns or would be better to do a mean of the values. Should I do a chi test with the data? Thank you so much!


In this case, I think removing one variable is fine. The number of cases that are not equal is so small that very little information is lost by dropping one variable.

Taking the mean of ordinal variables (which these are) is often done, but, technically, is not proper. Here, there is no need to violate the nature of the data by adding ordinal variables.


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.