Los puntos clave no están disponibles para este artículo en este momento.
Item analysis techniques commonly assume that item difficulties are stable over varying test contexts. The current emphasis on information-efficiency and individualized selection of items has made context stability a critical assumption. The simple item difficulties and Rasch log easiness values of 15 verbal analogy items were compared over seven internally consistent analogy tests. The results indicate that 60% of the items varied over contexts and that context difficulty was a factor for 27% of the items. The general applicability of item-parameter-invariant models is questioned by the finding of contextual interaction for parameter estimates.
Whitely et al. (Thu,) studied this question.