Table 3.

P-values from chi-square difference tests to determine which of the nested path models exhibited the best fit for each pain complaint

		Stability model	Normal causality model	Reverse causality model
	Model comparison	$\mathbf{\Delta }\mathit{\chi }$	$\mathbf{\Delta }\mathit{\chi }$	$\mathbf{\Delta }\mathit{\chi }$
		p-value	p-value	p-value
Neck pain	Normal causality	0.01	-	-
	Reverse causality	0.65	-	-
	Reciprocal causality	-	0.46	$\mathbf{<}$0.01

Note: Each model is compared to a more complex model. A significant difference test indicates that the more complex model is preferred, i.e. the additional paths are justified. P-values under the threshold for statistical significance are given in bold