AI Summary & Key Takeaways
AI InsightConclusion: For the evaluation of a scale’s construct validity, it is recommended to use the Rasch analysis in tandem with factor analytic methods, as the Rasch analysis explores a scale’s construct validity in terms of item and person fit, DIF and unidimensionality which is the only aspect of the factor analysis.
📌 3 Key Takeaways
Özet
Objectives: This study aims to develop measurement tools for assessing patients’ functional status with rheumatoid arthritis (RA) in terms of upper and lower extremity function and to evaluate the tools’ construct validities with classical and modern psychometric approaches. Patients and methods: Between April 2010 and April 2012, a total of 300 patients with RA (77 males, 223 females; mean age: 52.3±11.5 years; range, 18 to 82 years) who answered items from a range of widely used instruments were included. After examining initial dimensionality with exploratory factor analysis (EFA), confirmatory factor analysis (CFA), and Rasch analysis were used to evaluate the tools’ construct validities. The data-model fit was evaluated with goodness-of-fit (GoF) statistics in CFA, while the tools were examined in terms of item and person fit, unidimensionality and differential item functioning (DIF) from the perspective of Rasch analysis. Results: According to EFA, two dimensions were identified and named as “self-care-mobility-household activities related to lower extremity” and “self-care-mobility-household activities related to upper extremity” taking into account the factor loadings and the clinical classifications. While the clinical classification was tested with CFA, all items were loaded on their pre-defined dimensions with the factor loadings of ≥0.40 and GoF statistics were within the acceptable ranges. When the “self-care-mobility-household activities related to upper extremity” and “self-care-mobility-household activities related to lower extremity” tools were evaluated via the Rasch analysis, both tools were found to fit the Rasch model expectations, with a mean item fit statistics of -0.528 logit (standard deviation [SD]: 1.365) and -0.213 (SD: 1.168; mean person fit statistics of -0.412 logit (SD: 1.160) and -0.303 logit (SD: 0.859), respectively. Conclusion: For the evaluation of a scale’s construct validity, it is recommended to use the Rasch analysis in tandem with factor analytic methods, as the Rasch analysis explores a scale’s construct validity in terms of item and person fit, DIF and unidimensionality which is the only aspect of the factor analysis.