Donald Burrill wrote: > OTOH, if you also have <time-since-diagnosis> in its original form (not > categorized, but as numbers from 1 (not 0?) to, say, 25: then you could > use that as a linear predictor (and still include quadratic and cubic > orthogonal components if you wish), along with <age>.
If you have the raw data (not categorized) then using this as a continuous predictor is more powerful. You can still look at interactions (and probably should as Donald Burrill suggests). Also, the categorized variables may also inflate the Type I error rate (approaching 100% for large samples if there are ceiling effects in the categorized predictors). Thom . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
