Evaluating the Psychometric Properties of 8-item-CPQ_11-14 using Item Response Theory

Objective:  This study aimed to evaluate the psychometric properties of the 8-item-CPQ_11-14 RSF:8 and ISF:8 and identify differential item functioning across gender using Item Response Theory. Method:  The questionnaire was administered to 649 12-year-old Hong Kong school children (male=49.2%; female=50.8%). Unidimensionality of the scale was checked by confirmatory factor analysis (CFA), principle component analysis (PCA) and local dependency (LD) statistics. Graded response model (GRM) was fitted to the data. Contribution of each item to the scale was assessed by item information function (IIF). Reliability of the scale was assessed by test information functions (TIF). Differential item functioning (DIF) across gender was identified by Wald test and expected score function. Result:  Both CPQ_11-14RSF:8 and ISF:8 did not deviate much from the unidimensionality assumption. Results from CFA indicated acceptable fit of the one-factor model using various fit indices (RSF:8: NFI, CFI, GFI≥0.90, RMSEA=0.09, SRMR=0.04; ISF:8: NFI, CFI≥0.84, GFI=0.94, RMSEA=0.10, SRMR=0.04). PCA indicated that the first PC explained >30% of variation with high factor loadings (most items>0.30) for both RSF:8 and ISF:8. LD statistics<10 indicated the absence of local dependency. GRM showed good fit of data (RMSEA<0.05). Flat and low IIFs were observed in the oral symptoms items suggesting these items contributed little information to the scale. Comparing the TIFs, RSF:8 demonstrated higher reliability than ISF:8. Apart from the oral symptoms items, the item “Concerned with what other people think” in RSF:8 demonstrated a uniform DIF (p<0.001). Girls deemed themselves more emotionally sensitive to other’s view on them. However the expected score functions were not much different between boys and girls. Conclusion:  With the use of Item Response Theory, CPQ_11-14 RSF:8 demonstrated better psychometric properties than ISF:8 when assuming a unidimensional structure. Items with differential item function across gender were identified but its impact on the overall score was minimal.