[Rasch] Funny codes problem in Bigsteps

Mike Linacre (RMT) rmt at rasch.org
Tue Aug 29 07:58:33 EST 2006


Stuart:

Looks like category "1" only appears in records with extreme low scores. So 
that category was dropped from the estimation as inestimable in the frame 
of reference of the other categories. A solution is to include dummy 
records with both 1's and 2's. Include as many dummy records as quantify 
your theory about the relationship between those categories.

Mike L.

Stuart wrote:
>all the teachers who do not have mentors will get the minimum extreme 
>measure, but that will be better than arbitrarily giving them some low 
>number. The control file has codes=12345.




More information about the Rasch mailing list