Table 4 Model fit statistics

From: Genetic risk for schizophrenia and autism, social impairment and developmental pathways to psychosis

Model

χ2, p

DF

RMSEA

CFI

SRMR

BIC

Fully connected model

26.17, p < 0.001

7

0.065 (0.040–0.093)

0.879

0.030

22,063

Without PRSscz&asd → IQ

3.53, p = 0.62

5

0 (0–0.046)

1

0.012

22,054

Without PRSscz&asd → SF baseline

3.67, p = 0.82

7

0 (0–0.030)

1

0.012

22,041

Without PRSscz → SF follow-up

4.17, p = 0.84

8

0 (0–0.027)

1

0.012

22,035

Without PRSscz&asd → fMRI social

4.71, p = 0.91

10

0 (0–0.017)

1

0.013

22,022

Without PRSasd → PEs

4.90, p = 0.94

11

0 (0–0.010)

1

0.013

22,016

Without PEs → SF follow-up

5.87, p = 0.92

12

0 (0–0.014)

1

0.014

22,011

Without IQ → SF follow-up

6.09, p = 0.94

13

0 (0–0.007)

1

0.014

22,004

Without sex → SF baseline&fMRIsoc

6.91, p = 0.96

15

0 (0–0)

1

0.015

21,992

Without IQ → SF baseline

6.73, p = 0.82

11

0 (0–0.026)

1

0.014

22,018

Without IQ → PEs

8.55, p = 0.74

12

0 (0–0.029)

1

0.016

22,013

Without IQ → fMRI social

6.38, p = 0.78

10

0 (0–0.029)

1

0.016

18,299

Without sex → SF follow-up

10.73, p = 0.47

11

0 (0–0.041)

1

0.021

18,297

  1. χ2 difference tests showed that models did not significantly worsen when removing connections up to model (M), which had significantly worse fit indices than model (L) (p < 0.05)
  2. PRS polygenic risk score, scz schizophrenia, asd autism spectrum disorder, SF social functioning, PEs psychotic experiences, fMRIsoc fMRI social cognition