Re: [R] ordering of factor levels in regression changes result

On Feb 3, 2012, at 4:16 PM, Tulinsky, Thomas wrote: I was surprised to find that just changing the base level of a factor variable changed the number of significant coefficients in the solution. I was surprised at this and want to know how I should choose the order of the factors, if the

[R] ordering of factor levels in regression changes result

I was surprised to find that just changing the base level of a factor variable changed the number of significant coefficients in the solution. I was surprised at this and want to know how I should choose the order of the factors, if the order affects the result. Here is the small example. It is