Is there a way to compare two different instruments in the same Rasch analysis using Winsteps?
 
Let us assume a sample of persons been administered both instruments:
 
If so
1. analyze the two instruments separately, and produce PFILE=s.
2. cross-plot the person measures, e.g., with the Winsteps "Plots", "Compare" plot.
3. does the empirical trend line on the plot have the slope of an identity line (i.e., the line is: y = x +a) ?
 
4. If so, all the data can be analyzed together as one long "super test".
To produce comparable results for the two instruments:
Write out anchor files: IFILE= and SFILE=
Then anchor IAFILE= and SAFILE= and delete the items for each instrument in turn, IDFILE=, to produce comparable results for the instruments.
 
5. If not, then you are in a "Celsius to Fahrenheit" situation.
The instruments cannot be combined into one analysis.
Use USCALE= and UIMEAN= to adjust the measures in one instrument so that the relationship between the measures on analyses of the two instruments becomes y=x, the identity line.
Then the person and item hierarchies can be compared directly using their PFILE= and IFILE=