Figure 3.

Cholesky decomposition of the nested factors EF model for the cross-wave data (individual tasks not shown for simplicity). Each Wave 2 latent variable is decomposed into additive genetic (A), shared environmental (C), and nonshared environmental (E) variance shared with Wave 1 (paths to Wave 2 from A1, C1, and E1), and that unique to Wave 2 (paths from A2, C2, and E2). Squaring the path estimates provides the proportion variance accounted for by each variable (i.e, to obtain the percentages depicted in Figure 2). Common EF, Updating-Specific, and Shifting-Specific were estimated in the same model, but are orthogonal. Numbers on arrows are standardized regression coefficients predicting Common EF, Updating-Specific, and Shifting-Specific factor with the genetic and environmental latent variables. Numbers in brackets are standard errors. Boldface type and solid lines indicate statistical significance (p < .05), determined with chi-square difference tests. Note that when the chi-square difference tests disagreed with the standard errors, only the results for the difference tests were interpreted. W1 = Wave 1; W2 = Wave 2; Common EF = Common Executive Function; Updating-Sp. = Updating-Specific; Shifting-Sp. = Shifting-Specific.