Rasch analysis for advanced assessment solution

We perform multi-mini-interview for a screening body and perform the G-theory analysis on the data and we now want to compare the outcome some with Rasch analysis. TA three-facet Rasch model need to be run on all data to separate the effects of candidate ability, assessor leniency/stringency and the marking scale. This technique is an application of Item-Response Theory and has the capability to both identify these sources of error and partially adjust for them in a way that is fairer on candidates (Roberts et al. 2010). By calculating estimates of different ‘facets’, namely candidate ability, interviewer stringency/leniency and question difficulty a so-called ‘fair average score’ can be computed that reflects candidates’ abilities more accurately. The ‘fair score’ is the score that would have been obtained if all the measures of the other elements on all other facets had been located at the average measure (Linacre 2009). Analysis was performed using FACETS [login to view URL] Winsteps.