Methodology for the development of normative data for Spanish-speaking pediatric populations

OBJECTIVE: To describe the methodology utilized to calculate reliability and the generation of norms for 10 neuropsychological tests for children in Spanish-speaking countries. METHOD: The study sample consisted of over 4,373 healthy children from nine countries in Latin America (Chile, Cuba, Ecuador, Guatemala, Honduras, Mexico, Paraguay, Peru, and Puerto Rico) and Spain. Inclusion criteria for all countries were to have between 6 to 17 years of age, an Intelligence Quotient of >= 80 on the Test of Non-Verbal Intelligence (TONI-2), and score of <19 on the Children's Depression Inventory. Participants completed 10 neuropsychological tests. Reliability and norms were calculated for all tests. RESULTS: Test-retest analysis showed excellent or good-reliability on all tests (r's>0.55; p's<0.001) except M-WCST perseverative errors whose coefficient magnitude was fair. All scores were normed using multiple linear regressions and standard deviations of residual values. Age, age(2), sex, and mean level of parental education (MLPE) were included as predictors in the models by country. The non-significant variables (p > 0.05) were removed and the analysis were run again. CONCLUSIONS: This is the largest Spanish-speaking children and adolescents normative study in the world. For the generation of normative data, the method based on linear regression models and the standard deviation of residual values was used. This method allows determination of the specific variables that predict test scores, helps identify and control for collinearity of predictive variables, and generates continuous and more reliable norms than those of traditional methods.