Skip to content

lm without intercept

1 message · Peter Dalgaard

#
On Feb 18, 2011, at 14:20 , Jan wrote:

            
No. That'll be a piece of pragmatic advice caused by the experience that people too often get it wrong and e.g. try to remove anything with a high p-value from a model.

If your model meaningfully implies a zero intercept, as it does in some physical and biological models (no displacement at time zero, no growth at zero sunlight, etc.) then of course you can fit a model that incorporates this knowledge.