How to Handle Categorical Variables
"When the (1−p) confidence interval, also known as the error bar, crosses the level indicated by the base level category then that estimate is not significant."
Could you please elaborate what does it mean by "confidence interval crosses the level indicated by the base level category"?
Thanks.
Comments
Here's an example from p.15 of the GLM text.
The base level for the vehicle type variable was chosen to be "sedan". The confidence interval for "van" crosses 0 which is the estimate for the sedan base level. So we reject at whatever significant level we're using the idea that vans and sedans have different behavior. Whereas for the SUV and truck vehicle type levels, the confidence intervals do not cross 0 so we assume the estimates for those are significant.