Starting with the initial pool of 60 FAR- and ER-related items of the original KPI, a series of analyses was performed to develop the AEQ, involving the following steps. First, items that have caused too many missing data, because they are work-related, and items with low face validity were excluded. Second, the KMO measure of sampling adequacy was used to determine items that are appropriate for principal components analysis (PCA, Tabachnik and Fidell, 2001). Third, factor structure of the remaining KPI-items was calculated separately for the affective, cognitive and behavioral responses using three PCAs. For each PCA, examination of eigenvalues, Catell’s scree test, and parallel analysis were used to determine the number of factors that were appropriate to interpret (Horn, 1965). As we expected low to moderate correlations between the factors, subsequent oblique rotations (Promax) were used. Fourth, only items with factor loadings greater than 0.4 were used. The items should not show a cross loading (i.e. a second loading greater than 0.35). Item-total statistics guided the removal of items that did not correlate with the overall score of the questionnaire. Finally, the PCAs were repeated. Fifth, the internal consistency was measured on the AEQ scales using the Cronbach’s a, which indicated the extent to which a set of test items can be treated as measuring a single latent variable. Values of Cronbach’s a >.7 were considered adequate (Tabachnik and Fidell, 2001; Jensen, 2003). A series of Pearson product moment correlations were calculated to examine the scale intercorrelations between the subscales of the AEQ. Finally, for calculating criterion-related validity, correlations were computed between the AEQ sub-scales and typical outcome variables (e.g. pain intensity, disability). Univariate analyses of variance with CPC grades of pain severity as a between subjects factor (with grades 1–4) were used to examine whether patients with CPG grades three and four show more FARs and patients with CPG 2 show higher ERs. In the case, an AEQ sub-scale was significantly related to age, gender or pain duration, these variables were used as covariates. To examine the construct validity, correlations were inspected between AEQ sub-scales and conceptually similar and distinct measures.