John Fox <jfox <at> mcmaster.ca> writes: > Actually, the effects package does exactly what you suggest for continuous > predictors. But not for lme. Dieter