Is there a rule to how much data you must have per dummy variable?

I am doing a multiple regression project. I had to split up one of my explanatory variables into 10 dummy variables; some of these dummy variables have as low as 3 observations under it. Will this be a problem and is there a rule regarding this. For example is there a rule that i need atleast 5 observations per dummy variable.

BOB2010-12-02T21:15:43Z

Favorite Answer

*When I completed my MS thesis, and if I remember correctly, I believe my advisor recommended between 10 and 20 observations per variable* Good luck!