RELIABILITY & VALIDITY PRPP-ASSESSMENT IN CHILDREN 167 7 A traditional statistical procedure was conducted for convergent validity (hypothesis 1 and 2) using SPSS for Windows version 26.0 (New York, IBM corporation, 2019). To make rank-correlation possible, COPM and PRPP-mastery data was ranked in three categories: inadequate, sufficient, good. For the COPM performance categories were based on the Dutch grading system: inadequate=0-4, sufficient=5-6, good=7-10. For the PRPP mastery score, categories were based on knowledge of the assessment and set on inadequate=0-40, sufficient=41-84, good=85-100. Rank-correlations were calculated using Kappa, where a kappa of 0.4-0.6 is seen as moderate, 0.6-0.8 as substantial, and above 0.8 as almost perfect[45]. For structural and known group validity, all data was entered into excel then analyzed using FACETS 3.71.4. A four-faceted Rasch model was generated with facets measuring children, tasks, raters, and the PRPP-Assessment items [49]. The RASCH-generated hierarchy of the four facets was visually presented using the ‘Wright map for the many-facets Rasch analysis’ (in FACETS called the vertical rulers) to give an overview of how each facet is calibrated and to ensure that all facets perform sufficiently predictably to be considered a unidimensional construct [50]. Rasch analysis converts ordinal-level ratings to interval-level measure, the so-called logits. When interpreting Rasch analysis, the mean of the person abilities, Table 3 Hypotheses formulated a priori Aspect of construct validity Hypotheses Convergent validity 1. The rank-correlation between the COPM-performance score (parent rated) and the PRPP-assessment mastery score (stage 1) calculated by kappa, shows a moderate positive correlation (0.50 < r > 0.70). 2. The rank-correlation between the COPM-performance score (child rated) and the PRPP-assessment mastery score (stage 1) calculated by kappa, shows a moderate positive correlation (0.50 < r > 0.70). 3. The child-recognition in the PRPP-reports expressed by parents is positive (‘yes’) in at least 80% of the cases Structural validity 4. Item-fit statistics are considered acceptable (mean-square range 0.5 – 1.7, infit and outfit statistics are between -2 and +2) 5. Expert-statements on the difficultness of descriptors for children with mitochondrial disorders fit with the Rasch calculated hierarchy of hardest to easiest descriptors (items) for at least 80% Known group validity 6. The difference in efficiency of cognitive strategy application between the three profiles is noticeable based on the Rasch generated interaction between items and profile
RkJQdWJsaXNoZXIy MjY0ODMw