TABLE 5.

Test statistics of all eight models and results of the four χ²-difference tests.

Test statistics	Hypothesis 1, positive models				Hypothesis 2, negative models
	German sample		English sample		German sample		English sample
	Cor. allowed	Cor. zero	Cor. allowed	Cor. zero	Cor. allowed	Cor. zero	Cor. allowed	Cor. zero
N	1517	1517	1539	1539	1517	1517	1539	1539
χ2	235.783	314.497	284.340	380.434	64.026	77.137	110.226	165.051
df	54	64	54	64	25	31	25	31
p	<0.001***	<0.001***	<0.001***	<0.001***	<0.001***	<0.001***	<0.001***	<0.001***
BIC	36248	36260	38058	38089	33527	33497	32428	32443
χ2-diff.	81.985		97.191		13.11		57.248
df-diff.	10		10		6		6
p	<0.001***		<0.001***		0.04*		<0.001***

“Cor. allowed” describes the models allowing for correlations between all latent variables. “Cor. zero” describes the models in which all correlations of interest were set to zero. The row “χ²” contains robust chi-square values for all models, the row “BIC” contains the models’ robust Bayesian Information Criteria. “χ²-diff.” is the test statistic for the scaled χ²-difference test. “df-diff.” is the corresponding df-difference. Note that as the χ²-difference test is a function of two standard (not robust) test statistics, the χ²-difference test statistic cannot be calculated by simply subtracting the scaled χ² test statistics above. *p < 0.05. ***p < 0.001.