Statistical analysis and interpretation in a study of prelingually deaf children implanted before five years of age

This article describes the major statistical analyses used in a large-scale study of prelingually deaf children implanted before 5 yr of age. Two major challenges posed by the data were the need to reduce the number of outcome measures to a reasonable number while retaining the important information underlying the multiple measures, and the need to determine the unique contributions of multiple sets of predictor variables. The approach taken-principal components analysis of the outcome measures followed by hierarchical multiple regression-provided a compromise between statistical complexity and ease of interpretation.