Ölçeklerde yapı geçerliliğinin değerlendirilmesinde faktör analizi ve Rasch analizi yaklaşımları
Abstract
In order to obtain more precise observations in healthcare field, the measurement tool should be valid. Validation is the capability of the measurement tool on correct and accurate measurement of a characteristic, without intervention of other characteristics. In a scale development process, probably the most important and analytically evaluated type of validity is the (internal) construct validity. "Exploratory Factor Analysis (EFA)" and "Confirmatory Factor Analysis (CFA)" under the factor analysis, "Rasch analysis" under the modern test theory are the most common approaches in the evaluation of scale's internal construct validity. Within this study, measurement tools for the determination of disability levels of romatoid arthritis patients in terms of "self care-mobility-housework" have been developed for lower and upper extremities separately and their internal construct validities have been examined by EFA, CFA and Rasch analysis. According to EFA results, which are actualized with 32 items in total, it has been determined that the items are gathered in two dimensions; taking into account the factor loadings and the clinical classifications, first dimension has been named as "self care-mobility-house work related to lower extremity" and second dimension named as "self care-mobility-house work related to upper extremity". When the results of EFA were compared with the clinician opinion, it was observed that only one item showed misfit; as a result, it was determined to evaluate this item in the dimension which was approved by the clinical structure. While the clinical (theoretical) structure was tested with CFA, the results were found to be satisfactory and in harmony with the clinical structure according to goodness of fit statistics. In the Rasch analysis, "self care-mobility-house work related to lower extremity" and "self care-mobility-house work related to upper extremity" were handled separately and also, both were taken together ignoring the clinical structure. Within the scope of the analyses, the items showing misfit to the model, showing differential item functioning in terms of age, gender and/or duration of disease and the ones which violate the hypothesis of unidimensionality were excluded from the study. Compared to the factor analysis that only examines the dimensional structure of the scales, Rasch analysis contains various evaluations such as item fit, person fit and item bias. Thus, these methods should be used together in the evaluation of scale's internal construct validity