z-scores corrected for age + k-means
Good afternoon every body,
I am writing to you in the context of my PhD. My thesis focuses on predictors of reading achievement in children with developmental language disorders.
My first question concerns the management of composite scores. I wasn't happy with the use of average z-scores (a high z-score compensated for a low z-score in some children, and could mask a deficit in some children) so until now, I've created a z-score using the mean and standard deviation of my control group. However, this does not take into account the age of the children and may penalize the youngest. What do you think is the most reliable technique for correcting z-scores for age? Should I use weighted regression for age?
My other question concerns cluster analysis. I'd like to see if there are any language and pre-reader profiles within my group of children with developmental language disorders. I've started my analyses with k-means. The software automatically optimizes the number of clusters according to the BIC index. But you can also choose to optimize according to AIC or Silhouette, or to optimize manually by comparing the BIC/AIC/-2LL values to determine the optimal number of clusters. Could you recommend some papers to help me better understand how to proceed with k-means?
Kind regards !
Comments
Hi Prisca,
I am not sure what the best way is to correct z-scores for age, but I am sure you are not the first one to confront this problem (so I would look online and in the literature to see what the experts recommend). And for the clusters, my strong preference is to go with BIC.
Cheers,
E.J.