I want to see whether patients have a higher risk of seizures, the bigger their brain tumor.

My dependent variable is therefore "seizure (yes, no)" and my independent variable is "tumor size" ranging from 1 to 5.

There is a significant correlation if I just type

logistic risk size

creating a logistic regression for the variable, but if I run it as

logistic risk i.size

there is no change, but I can see the value gets more significant the larger the tumor.

I'm quite suspicious this change is significant but I think my problem is I can't get it to recognize it as an ordinal variable.

Perhaps the solution is simply not to treat it as ordinal but as continuous, and therefore my original result is true?

  • $\begingroup$ How is the tumor size variable created? What do the categories 1 to 5 mean? $\endgroup$ – COOLSerdash Apr 22 at 9:29
  • $\begingroup$ It is 1 variable with a dropdown (1 choice) from 5 mm to 45mm if I recall correctly, in increments totaling 5 options from low to high size. $\endgroup$ – Paze Apr 22 at 10:37
  • $\begingroup$ Given your explanation, treating it as categorical is the right thing to do. $\endgroup$ – Dimitriy V. Masterov Apr 30 at 4:09

Your Answer

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

Browse other questions tagged or ask your own question.