Has anyone attempted to replicate the cognition factor scores (General Ability, Executive Function, Learning/Memory) from Thompson et al. 2019? I have been using the factor loadings reported in the original manuscript and borrowing from the open source code (GitHub - ABCD-STUDY/BPPCA_Neurocognition_Thompson_etal_2018: The Structure of Cognition in 9 and 10 year-old Children and Associations with Problem Behaviors: Findings from the ABCD Study’s Baseline Neurocognitive Battery.) but have been unable to reconcile the factor scores I’ve calculated with the factor scores available from DEAP (neurocog_pc1.bl, neurocog_pc2.bl, and neurocog_pc3.bl). The correlations between my calculated scores and the DEAP scores are greater than 0.9 but not quite 1.
Here are my calculations:
combined.data$thompson_PC1 <- (0.754*combined.data$PicVocab)+(0.213*combined.data$Flanker)+(0.471*combined.data$List)+(0.205*combined.data$CardSort)+(0.015*combined.data$Pattern)+(0.012*combined.data$Picture)+(0.82*combined.data$Reading)+(0.306*combined.data$RAVLT)+(0.5*combined.data$LMT)
combined.data$thompson_PC2 <- (0.065*combined.data$PicVocab)+(0.712*combined.data$Flanker)+(0.148*combined.data$List)+(0.71*combined.data$CardSort)+(0.813*combined.data$Pattern)+(0.135*combined.data$Picture)+(0.12*combined.data$Reading)+(0.125*combined.data$RAVLT)+(0.299*combined.data$LMT)
combined.data$thompson_PC3 <- (0.19*combined.data$PicVocab)+(0.067*combined.data$Flanker)+(0.493*combined.data$List)+(0.232*combined.data$CardSort)+(0.085*combined.data$Pattern)+(0.863*combined.data$Picture)+(0.122*combined.data$Reading)+(0.712*combined.data$RAVLT)+(0.068*combined.data$LMT)