Orthogonal-polynomial contrasts


New Member
Hi there,

During my studies, I always learned that, when you have an independent ordinal variable, you could include it into your regression model, but you would have to create (k-1) dummies, as you would have done for a nominal variable. Yet, I was reading Fox and Weisberg's book (An R companion to applied regression), and they talk about orthogonal-polynomial contrasts, that you can compute with the ordered command in R.

So my question would be: as an independent variable, is it more appropriate to use orthogonal-polynomial contrasts for ordinal variable? Could we still use dummy variables?

If my question is to vague, just let me know, and I will be gladly clarify it.

Thank you!