Table 4.

Model comparison.

Model	PTSS R2	PTSS f2	Q2() Average
A: Hypothesized	.438	NPS = .011 WM = .114 DISTRESSING EVENT = .431	0.067
B: Null	.374	NPS = .000 DISTRESSING EVENT = .428	−0.086
C: Comparison	.407	NPS = .008 WM = .056 DISTRESSING EVENT = .373	−0.109

Note. A: Hypothesized model in which treatment burden (Neurological Predictor Scale; NPS) is associated with post-traumatic stress symptoms (PTSS) both directly and indirectly via two mediators: (1) the experience of a distressing event and (2) fronto-limbic white matter (WM) microstructure. WM is indexed by radial diffusivity (RD) values from the left and right cingulum bundle and inferior fronto-occipital fasciculus. PTSS indicators include avoidance, re-experiencing, and arousal. Covariates include whether a typically developing control was a sibling of a patient (sibling status) and age at testing. B: Comparison model excluding WM tests whether treatment burden and distress alone are associated with PTSS as effectively as when WM is included. C: Anatomical control model testing whether the effect of WM on PTSS is specific to fronto-limbic tracts. WM is represented by RD values from control tracts, the left and right optic radiation. The best-fitting model was identified based on a combination of out-of-sample relevance (Q²) for the final endogenous variable, PTSS. Superior models demonstrated higher explained variance (R²), higher effect size (f²), and positive Q². Model comparisons were conducted according to PLS-PM guidelines outlined by Hair et al. (2017, 2022).