Skip to main content
Fig. 7. | Biology of Sex Differences

Fig. 7.

From: Univariate and multivariate sex differences and similarities in gray matter volume within essential language-processing areas

Fig. 7.

Structure of the LR models after including TIV as an additional predictor. A, B Nomograms illustrating the relative contribution of TIV and each of the features included in the raw and PCP datasets to the Pclass scores. The values of three discrimination indexes (R2, C index, and Somers’ D) of each of these two models are reported within the plots. C Ordinal relationships (quantified through the absolute value Spearman’s rho correlation index) between the coefficient values of the four LR models fitted in this study and depicted in A and B of Figs. 6 and 7). Note that these associations were calculated excluding the coefficient value associated to TIV (which is only included in two of these four LR models) and that, because the sign of these associations is arbitrary (i.e., it arises from the different sex category used as reference in the distinct LR models), absolute rho values are reported. D Values of the regression coefficients in each of the four LR models fitted in the present study. Highlighted in green are those coefficients reaching statistical significance (p < 0.05) in each model (see details in Additional file 1: Table S3D). E UpSet plot illustrating the intersections between the predictors reaching statistical significance in the fitted LR models. In this plot: (1) the color of the line-joined circles denotes whether the features listed in each column reached statistical significance (green) or not (white) in a particular model, thus identifying which models are part of each intersection; (2) the height of the bars of bars on the top illustrates the number of features included in each intersection (the cardinality of each intersection); (3) the color of the bars denotes the number of models that included the listed features as significant predictors. Thus, for example, the first intersection includes three brain features that reached statistical significance as predictors in all four models, the second one includes two brain features that reached statistical significance in the PCP, raw + TIV, and PCP + TIV models (but not in the raw model), and so on

Back to article page