Person characteristic curves: we need
a. enough observations of the person
b. all the items to have the same response format (or delete those that don't)

 
If so, then
1. Transpose the data matrix: Winsteps Output File menu. Transpose. So that the persons become the items.

2. ISGROUPS=0 in the transposed file if you want to model each person to have a unique rating scale.

3. Analyze the transposed file.
4. The Graphs menu now produces person curves, not item curves.

 

Here is an example from Example0.txt