Checks for spurious effects with Oldham’s method confirmed validity for the correlation with reading accuracy in L-types (especially with text reading accuracy: r = −0.528, p = 0.052, whereas other components were not fully significant) and with reading speed in P-types (for all components: r ≤ −0.339, p ≤ 0.024), but not for word reading accuracy (r = −0.022, p = 0.887).
Of the seven included global scores (reading accuracy, reading speed, writing accuracy, phonemic awareness, memory and callosal transfer), a significant difference emerged for Global writing accuracy ( p = 0.029, partial eta square = 0.033), while Global reading accuracy and Global phonemic awareness almost reached significance ( p = 0.081, partial eta square = 0.023 and p = 0.071, partial eta square = 0.021, respectively).
When significant or close-to-significant (i.e., p < 0.10, considering that a large effect of a single subcomponent could have been masked by small or null effects in other subcomponents) differences emerged between the two groups of children (responders and non-responders) at the pre-test on a global score; the scores of the components involved in it were analyzed with a multivariate analysis of variance to better characterize the difference (overall multivariate effect and single effects, with alpha = 0.05, two-tailed).
A check for spurious effects with Oldham’s method confirmed validity for the correlation with Global reading accuracy (r = −3.96, p = 0.028) and the subcomponents of nonword reading accuracy and text reading speed, whereas correlations with word and text reading accuracy only approached significance.