"Dang, Jeff" wrote: > > Edstat, > > I have personally found that a lot of health researchers like to aggregate > normally distributed, continuous outcomes into dichotmous outcomes. In some > cases, this is done because the researcher is more familiar with dicohotmous > outcomes (disease/no disease) and seeks to interpret their results in terms > of odds ratios within a logistic regression. > > In some cases, this can be problematic because you lose information. For
In _all_ cases (see the MacCallum et al. paper that others mentioned) you lose information and hence power. > instance those near the cut-off point are forced into one group or another. > Thus, you exaggerate the differences for some individuals. Dichotomization can introduce spurious significance. Even if the continuous measure is the by-product of a genuine dichotomy it is unlikely that the cut-off will be in the right place to make dichotomization appropriate. 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/ . =================================================================
