5.4 Content balancing

To select items for an instrument, the first step is to make sure that the items are distributed evenly over the \(D\)-score scale (see chapter @(ch:information)). The items that have difficulty estimates from a Rasch model, all load to the construct and are therefore all assumed to measure child development well. The Rasch model is a one dimensional model, and this model is well suited for developmental data (see Booklet I: chapter 4 for the rationale behind the choice for the Rasch model to model child development). Nevertheless, child development is studied to occur on multiple domains: motor development; language development; cognitive development etc. Having this content validity reflected in the \(D\)-score measure is important. In Booklet II: 6.3 Domain coverage and scores we show the extent to which the different developmental domains are covered in the \(D\)-score.

In constructing a new instrument to measure the \(D\)-score, the domain coverage may also be important. So to make sure that the new instrument has a good content validity, select items in such a way that all domains are represented. The domain representation by the selected items can be checked by plotting the item information by \(D\)-score for the selected items, and color the item information for the different domains. This will show to what extend, all domains are covered by the current selection of items.

item	label	domain
ddigmd057	Lifts head to 45 degrees on prone position	motor
mdtgmd002	Prone (on tummy), can lift head up to 90 degrees	motor
iyosec001	Does your child stop crying or calm down when you come to the room after being out of sight, or when you pick him or her up?	sem
by1mdd054	picks up cube (check hand preference)	motor
mdtgmd010	Sits by self well	motor
gricgd022	Finds toy under cup	cog
sgrred035	Uses spoon	life
by3cgd051	Blue Board Series: 1 piece	cog
by3cgd059	Attends to story	cog
mulcgd200	Matches by shape	cog