While @ Modelytics, I was told that if dummy variable takes 3 values, then there is a need for 2 variables in the regression equation. Like a dumb fuck I had never asked the reason for it. I merely accepted it.

Today while reading path models, I came across a statement which seems so bloody obvious . If you include 3 variables for each of the dummies, then design matrix does not become a full rank matrix and hence cannot be invertible and hence there are no estimates from the model.

Such a simple concept, I did not even have the knowledge to understand it as I was completely oblivious to linear algebra. Anyways , this is one of those things I learnt @ 33.

I guess my learning@33 will be quite a few :)