Here I am using simple linear regression to predict z-ave using the full sample. Parcel centrality and within-network dispersion was re-calculated using Mahalanobis distance. The Mahalanobis distance was chosen since it measures the distance between a point and a distribution, accounting for the correlations among the variables (gradient dimensions).
Measures
clear_129_md: Parcel 129 (Area_STSv_posterior_L, FPCN) centrality across the three gradients during clear
clear_284_md: Parcel 284 (Area_PFcm_R, SMN) centrality across the three gradients during clear
g1_clear_258: Parcel 258 (Area_IFJa_R, FPCN) gradient 1 score
g1_main_108: (Middle_Insular_Area_L, FPCN) gradient score
g3_replace_82: Parcel 82 (Area_posterior_9-46v_L, FPCN) gradient 3 score
g3_suppress_44: Parcel 44 (Medial_Area_7A_L, VN) gradient score
clear_fpcn_md: Within-network dispersion of the FPCN during clear
##
## Call:
## lm(formula = z_ave ~ clear_129_md + clear_284_md + g1_clear_258 +
## g1_main_108 + g3_replace_82 + g3_suppress_44 + clear_fpcn_md,
## data = sub_nets)
##
## Residuals:
## Min 1Q Median 3Q Max
## -0.7737 -0.2982 -0.0353 0.2586 0.8778
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -5.2817 1.6073 -3.286 0.00212 **
## clear_129_md 0.1699 0.1203 1.412 0.16563
## clear_284_md 0.1302 0.1011 1.288 0.20510
## g1_clear_258 24.6444 5.2856 4.663 3.45e-05 ***
## g1_main_108 17.9143 8.3063 2.157 0.03709 *
## g3_replace_82 4.4796 2.8324 1.582 0.12163
## g3_suppress_44 6.0478 3.1010 1.950 0.05818 .
## clear_fpcn_md 3.5337 1.0841 3.259 0.00228 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.4348 on 40 degrees of freedom
## Multiple R-squared: 0.6896, Adjusted R-squared: 0.6353
## F-statistic: 12.7 on 7 and 40 DF, p-value: 1.899e-08##
## Call:
## lm(formula = z_ave ~ g1_clear_258 + g1_main_108 + clear_fpcn_md,
## data = sub_nets)
##
## Residuals:
## Min 1Q Median 3Q Max
## -0.99793 -0.37630 0.04637 0.31924 0.91738
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -4.975 1.772 -2.808 0.00741 **
## g1_clear_258 31.404 5.457 5.754 7.74e-07 ***
## g1_main_108 26.242 8.951 2.932 0.00533 **
## clear_fpcn_md 4.073 1.193 3.413 0.00139 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.4871 on 44 degrees of freedom
## Multiple R-squared: 0.5715, Adjusted R-squared: 0.5423
## F-statistic: 19.56 on 3 and 44 DF, p-value: 3.303e-08Model 1, the more complex model is significantly better than Model 2 for predicting z-ave.