How to code a categorical variable for logistic regression with overlap in the categories/subgroups?

Asked Jan 21 '19 at 20:21

Active Jan 21 '19 at 20:21

Viewed 81 times

Suppose I have a categorical variable consisting of four levels: a, b, c, and d. When these levels are mutual exclusive, I would use dummy coding - so three dummies with for example level a as reference.

Is this dummy coding scheme still valid if two or more categories can overlap for a certain observation? For example, a person who has both level a and level c. Or does this mean I need to make four dummies instead?

asked Jan 21 '19 at 20:21

Amonet

I think it should work. But I would rather encode it into four levels instead of three. – Pedrinho Jan 21 '19 at 20:39
That's what my intuition tells me also, but it makes me wonder if four levels / dummies is correct. – Amonet Jan 22 '19 at 07:40
Just do it. I cannot think of a reason why it should not be correct. – Pedrinho Jan 22 '19 at 13:17

How to code a categorical variable for logistic regression with overlap in the categories/subgroups?

0 Answers0

Linked