6.1 Comparing child development across studies
Figure 6.1 shows the scatterplot of the D-score by age separately for each cohort. Remember from section 2.1 that each study selected its own set of instruments to collect the data. The scatterplots demonstrate a significant advance made possible by the D-score: We can plot the developmental scores of children from different cohorts, with different ages, using different instruments, on the same vertical axis.
The five blue lines guide the eye. These lines indicate the locations of the -2SD, -1SD, 0SD, +1SD and +2SD quantiles at each age in the combined data. Section 5.4, in Chapter I motivates the idea and provides some technical details. We’ll come back to these lines in section 7.2.
By and large, the data in every study follow the blue lines. Perhaps the most obvious exception is the
GCDG-JAM-STUNTED cohort, where older children somewhat exceed the D-score range. It is unknown whether this is real, or due to a sub-optimal calibration of the instrument.
Figure 6.2 plots the same data with D-score transformed into age standardized scores (DAZ). Replacing the D-score by the DAZ emphasises the differences both within and between studies. The majority of observations lies between the -2 SD and +2 SD lines in all cohorts. Using DAZ makes is easier to spot deviating trends, e.g., for the Jamaican or Ethiopian data.