Fig. 3: Multi-domain structural model (Word reading subset).

Genetic-relationship matrix structural equation modeling (GSEM) of literacy, phonological awareness, language and PWM abilities (N = 6383). The path diagram (a) depicts the genetic factors of the best-fitting model (IPC model) describing variation in spelling accuracy at age 7 (Spelling a 7, NB), phonemic awareness at age 7 (Phon aware 7, AAT), listening comprehension at age 8 (Listening compreh 8, WOLD), non-word repetition at age 8 (NW repetition 8, CNRep), and word reading speed at age 13 (W reading s 13, TOWRE). The phenotypic variance was dissected into common (AC) and specific (AS1-AS5) genetic factors, according to an Independent Pathway model, as well as residual factors (E1-E5), based on a Cholesky decomposition (Factor loadings not shown, Supplementary Table 17, and Supplementary Fig. 4). Observed measures are represented by squares and factors by circles. Single-headed arrows (paths) define relationships between variables. Dotted and solid paths represent factor loadings with p > 0.05 and p ≤ 0.05 respectively. The variance of latent variables is constrained to unit variance; this is omitted from the diagram to improve clarity. The variance plot (b) depicts the standardized genetic variance components for the model in a. The correlogram (c) shows genetic (r_g, lower triangle) and residual (r_e, upper triangle) correlations for the model in a. Point estimates and their SEs are reported in Supplementary Table 20. All measures were rank-transformed. IPC model – Combined Independent Pathway/Cholesky model.