7.2 External validity

7.2.1 Discriminatory validity

Discriminatory validity indicates the extent to which the D-score can distinguish children with non-normal development from children that are developing normally. We may evaluate this by identifying children with lagging development, for example, indicated by reflex or tonus problems, and study whether the D-score can discriminate those children from the general population. Section 9.3 presents some examples.

7.2.2 Convergent and divergent validity

Convergent validity is the extent to which the D-score relates to similar constructs. We measure it by the correlation between the D-score and the total score on Bayley-III or Denver.

The correlation with the other construct should be 0.6, or higher for good convergent validity. Unfortunately, at present, only limited data is available for the DDI, so we cannot assess convergent validity for the D-score at this point.

Divergent validity is the extent to the D-score is uncorrelated with measures of a different construct.

Relation between body height and the D-score in the SMOCC data.

Figure 7.3: Relation between body height and the D-score in the SMOCC data.

Figure 7.3 shows both convergent and divergent validity at work. The figure shows that, as expected, there is a strong and almost linear relation between body height and the D-score. However, after correction for the child’s age, the relationship between height and D-score almost disappears. Thus, growth and development are entirely different concepts.

We can also evaluate the strength of the relations between the D-score and proxy measures of child development, such as stunted height growth (see section 1.3). The low correlation between DAZ and HAZ suggests that stunting is a poor proxy for child development.

7.2.3 Predictive validity

Predictive validity refers to the degree to which the D-score predicts the score on a criterion that is measured later. For the D-score, we may compare to measures for IQ at the school-age as a possible criterion.

Vlasblom et al. (2019) found strong evidence that individual milestones of the DDI measured during the first years of life predict later intellectual functioning at ages 5–10 years. It is to be expected that the D-score, which builds upon these individual items, will also predict limited intellectual functioning, perhaps even better.