Conferences

Having a young child with learning problems or delayed development is likely to put considerable strain on the parents and many parents will require some kind of outside support. The Portage Program is a home-intervention program for children who have special educational needs. A developmental checklist is used to assess the children’s existing repertoire of skills in six developmental domains. For each domain the “items” consist of descriptions of often complex behaviors. A basic assumption underlying the checklist is that items are ordered on a developmental continuum which corresponds to the chronological age in case of normally developing children.

The present paper focuses on the psychometric characteristics of the revised version of the checklist in the Dutch setting. Parents of children from 0 to 5 years old who are developing normally, were recruited in several ways and sent a checklist, of which 736 were returned. The children were assigned to groups based on chronological age. For each domain, each age group received a different subset of items, targeted at their supposed developmental level. These item sets are called ‘booklets’. Within a booklet the items were arranged in supposed order of difficulty. Each booklet overlapped with the previous and subsequent booklets. Within a booklet item means can be compared directly, but between booklets we have to account for the age differences. Using the booklet overlap we can put the item difficulties on the same scale, but this can be done more easily within an IRT modeling framework. The scaling models we will use are the Rasch model and the one-parameter logistic model (OPLM), both implemented in the OPLM program (Verhelst, Glas & Verstralen, 1995). With OPLM a set of test items can be calibrated on a common scale, and several item oriented statistical tests become available. Moreover the OPLM program is suited to incomplete data designs.

At present the analyses are still in a preliminary stage. We conclude tentatively, that it is possible to find a reasonably fitting scale for each of the content domains, but at the cost of losing a large number of many items. It might still prove to be better to develop separate scales for the age-groups instead of trying to scale all items on the same dimension. Also the item difficulties were in general increasing with the original ordering, but there were many inversions.