Some Comparisons of Methods of Fitting the Dosage Response Curve for Small Samples