Thanks for sharing the links. We have already drawn a ROC corve in SPSS (file attached). The curve is not smooth classically. Is it possible to get a smoother curver using SPSS?
The ROC curve is a matter of fact. If it isn't smooth, that's the way it is. It shows you how your data really are, not what you would like to see
Generally, lumpy ROC curves are because there are either few individual values in your predictor or because of small numbers of cases. But the curve is showing what actually obtains in your data.
Those smooth curves are drawn under the assumption that your data is binormal in healthy and diseased populations. Generally the more the data you have, the smoother the roc curve will be.