10.5 D-score from existing instruments

There is a vast base of historic child developmental data using existing instruments. The problem is that each device defines its own summaries, so we cannot compare scores across tools. Different instruments have different domains, various age forms, different stopping rules, diverse age norms, and so on. Yet, the milestones in these instruments are often very similar. Most tools collect data on milestones like:

  • Can the child stack two blocks?
  • Can the child roll over?
  • Can the child draw a cross?
  • Can the child stand?
  • Can the child say “baba?”

With the D-score methodology in hand, we are ready to exploit the overlap in milestones shared by different instruments. Common items can act as bridges, so - with the appropriate item-level data - we may attempt calculating D-scores from other tools as well.

The task is to identify milestones that overlap between both instruments, filter out milestones that do not fit a joint model, and estimate the item difficulties of items that remain. Chapter II will explore this possibility in more detail.