Here are some messages that will or might appear when the RaschAnalysis1 program runs.




The program expects to find that all items have been scored on a right/wrong basis, with zero points for a wrong answer and one point for a correct answer.


It determines if this condition has been met by looking at two rows in the IStats worksheet, the ones labeled "Min" and "Max". An example is shown here from a 56-item multiple choice test - one of the items, "I42", has been found to have a "Min" of 1.00:




Now it just so happens that the program is being a bit too smart. I42 was indeed scored on a right/wrong basis but, of the 137 students, not one got the item wrong. As a result, the minimum score on the item was 1 (one) and the macro wasn't happy.


So the error message wasn't quite accurate in this case, but the result is the same: I42 cannot be processed. It has to be excluded from the test. In Lertap5 this is done by adding a *exc line in the CCs sheet, as shown here:




Why c43? Because that's where I42 responses were found in the Data sheet - the Freqs report says so, have a look:




Note that excluding items where everyone got it right, or wrong, is common in Rasch analyses. Such items will not fit the Rasch model. Many other Rasch programs don't sound an error as the RaschAnalysis1 macro does; they, the other programs, make their own processing adjustments internally without saying anything. Incidentally, the same goes for students: those who get all items right, or all wrong, cannot be accommodated by the Rasch model. Dexter is an R package that uses "inf" (infinity) for a student's Rasch score (called "theta" in Dexter) when a student has a perfect score, as exemplified below with results from the 35-item BLOT test: