9.4 Better measurement
The D-score metric is a generic measure of child development. It summarises child development by one number. We found that D-score fairly represents development domains over the entire scale. Due to its generic nature, the D-score is less suitable for measuring a specific domain. It may then be better to use a specialised tool that accesses motor, cognitive or communication faculties. For example, think of sub-scales from the Bayley, ASQ, Griffiths, and so on. Note that also in those cases, one still has the option of calculating a D-score.
The opposite scenario may also be of interest. Suppose we want to measure generic development AND identify any areas of slow growth. Extending the measurement by adding more items from domains with a higher failure rate will then increase precision in areas of suspected delay.
Since we based the D-score on a statistical model, we may create instruments customised to the exact needs of the study. Population-based studies may require a short measure consisting of a handful of items per child, and aggregate scores over many children to achieve precision. Intervention studies aim for a precise estimate for the intervention effect. If group sizes are small, we may administer a more extended test to achieve the same precision and vice versa. At the other end of the spectrum, for clinical purposes, we want a precise estimate for one particular person, so here we will administer a relatively long test. The good news is: As long as we pick items from the statistical model, the D-score in those three cases are all values on the same scale.
Our ongoing work targets tailoring instruments to a study design and discusses all of these options. And more.