Skip to main content

Table 8 Fit statistics for different invariance models of CAPE-42 positive psychotic experiences across study samples

From: Cultural differences in positive psychotic experiences assessed with the Community Assessment of Psychic Experiences-42 (CAPE-42): a comparison of student populations in the Netherlands, Nigeria and Norway

CAPE-Pos frequency

χ2

(Δχ2)a

df

(Δdf)

p

(Δp)

CFI

(ΔCFI)b

RMSEA

(ΔRMSEA)b

SRMR

(ΔSRMR)

Comparison

Decision

5-dimensional modelc; n = 245d (Netherlands); n = 478d (Nigeria); n = 162d (Norway).

 M1 Configural invariance

768

426

<  0.001

0.823

0.059

0.063

–

Reject

 M2 Metric invariance

809 (41)

454 (28)

<  0.001 (0.018)

0.804 (0.019)

0.061 (−0.002)

0.074 (−0.011)

M1 vs. M2

Reject

 M3 Scalar invariance

1004 (195)

482 (28)

<  0.001 (<  0.001)

0.711 (0.093)

0.071 (− 0.010)

0.088 (− 0.014)

M2 vs. M3

Reject

2-dimensions reduced model (Strange experiences – items 5, 17, 24, 26, 28, 30, 31; Paranoia – items 2, 6, 7, 22); n = 245d (Netherlands); n = 478d (Nigeria); n = 162d (Norway).

 M1 Configural invariance

197

129

<  0.001

0.917

0.049

0.050

–

Accept

 M2 Metric invariance

213 (16)

147 (18)

<  0.001 (0.291)

0.911 (0.006)

0.048 (0.001)

0.062 (−0.012)

M1 vs. M2

Accept

 M3 Scalar invariance

334 (121)

165 (18)

<  0.001 (<  0.001)

0.775 (0.136)

0.071 (−0.023)

0.086 (−0.024)

M2 vs. M3

Reject

 M3a Partial scalar invariancee

226 (13)

159 (12)

<  0.001 (0.375)

0.910 (0.001)

0.046 (0.002)

0.065 (−0.003)

M2 vs. M3a

Accept

  1. ΔCFI < 0.01 implies that invariance assumption holds [81], aYuan-Bentler scaled test-statistic (rounded to nearest integer), bRobust RMSEA and CFI from the scaled test-statistic; cItem no. 41 dropped, due to (near-)zero variance in Dutch and Norwegian samples (i.e. [virtually] all scores = 1); dMissing data points were processed using listwise deletion, eNo intercept constraints for items 6, 7, 17