#  I am planning to monitor the proportion of children meeting a benchmark on the Z-scores from the CREDI Short Form. Is this appropriate? 

 



We currently do not recommend this. The CREDI team recently discovered that the CREDI Short Form scores are underdispersed. This is not generally a problem for estimating the average level of development in a population and does not cause issues for users interested in hypothesis testing (e.g., examining differences in Short Form scores between groups). However, this underdispersion problem poses important challenges when looking at Z-scores generated from Short Form data, particularly when attempting to calculate the proportion of children above or below a benchmark. Because of the underdispersion problem, it is difficult or impossible to observe very extreme Z-scores (e.g., below -2 or above +2). This is a major problem for those wanting to examine the number or proportion of children falling below a given benchmark (e.g. the proportion of children with scores below -2 SD) because the estimate of children below this benchmark is very much undercounted in the current Z-scores. As such, we are currently recommending not to use existing Short Form Z-scores for monitoring purposes until we update the methodology we use to calculate scores using a revised reference population. We estimate that this process should take several months and will notify users when updates have been made. In the meantime, users who are interested in learning more about this issue and our planned methodological solutions are welcome to contact us. While less interpretable, it would be appropriate to monitor the average Short Form scores by age group until we have developed a more acceptable solution that allows users to calculate proportions based on Z-scores.