Beck Depression Inventory (BDI) (Beck et al., 1996), one of the most widely used self-report tests of depressive symptoms. Since then, it has been validated for various groups including clinical and nonclinical settings, and has been translated for use with Turkish populations. Beck Depression Inventory has 21 items (depressive symptoms) each of which is represented by four items. These four items are numbered from 0 to 3 like a rating scale.  For instance the scale for item 5 is [0=I don't feel particularly guilty, 1=I feel guilty a good part of the time., 2=I feel quite guilty most of the time., 3= I feel guilty all of the time.] and,  for item 6 is [0=I don't feel I am being punished., 1=I feel I may be punished., 2=I expect to be punished., 3=I feel I am being punished.]. Majority of studies used BDI treated the data as if it comes from rating scales. However, unlike many psychological measurement tools with a rating scale, items do not to share the same rating scale (e.g., Likert agreement). Although there are four points in each, the anchor values for each item are different. Yet, if each item is designed to have a different rating scale, then partial credit model (PCM) is indicated, and it requires strong evidence for us to use rating scale model (RSM). Unfortunately there is no study in the literature that shows one model is better than the other for BDI. Therefore I aim to determine which of the model works better while determining the best factorial structure. Rasch analysis of BDI data from 390 high school students shows that two factorial structure shows better item fits when PCM model is used.  Somatic domain is definitely works better with PCM. Cognitive\affective domain is better with PCM, yet still two misfiting items exist.