71 The Otology Questionnaire Amsterdam 4 Construct validity Inter-item correlation: The inter-item correlation (ICC) matrix for construct validity was already performed by Bruinewoud et al 2018 and showed good correlation between the individual items in the impact subgroup.5 No items had a too high correlation with another item (>.9; i.e. most likely identical items) nor a too low correlation (<.2; i.e. no correlation; Table 2). Confirmatory factor analysis (CFA): The CFA was performed using collapsed data; that is, answer categories containing less than 5% of the answers filled in were combined with the adjacent category. In items 1, 2, 3, 6 and 7, the answer categories 4 and 5 were combined (4; ‘Agree’, 5; ‘Strongly agree’). In Item 4, answer categories 3, 4 and 5 were combined (3; ‘Neutral’, 4; ‘Agree’, 5; ‘Strongly agree’). There were no residual correlations above 0.2, which indicates that there were no local dependencies. Hypothesis testing: All hypotheses were confirmed. As expected, moderate to high correlation coefficients were found between the impact items of the OQUA and the total score of the GHSI (0.672) and the general subscale of the GHSI (0.721). In addition, in line with expectations, weak to no correlation between the OQUA and the social (0.005) and physical subscale (0.177) of the GHSI was confirmed. Table 2: Inter-item correlation matrix showing good correlation between the individual items. Item 1 2 3 4 5 6 7 8 9 1 1,000 ,647 ,547 ,542 ,628 ,481 ,424 ,608 ,551 2 ,647 1,000 ,499 ,641 ,513 ,431 ,491 ,533 ,633 3 ,547 ,499 1,000 ,540 ,574 ,498 ,426 ,559 ,463 4 ,542 ,641 ,540 1,000 ,468 ,467 ,494 ,492 ,600 5 ,628 ,513 ,574 ,468 1,000 ,613 ,610 ,737 ,548 6 ,481 ,431 ,498 ,467 ,613 1,000 ,700 ,695 ,492 7 ,424 ,491 ,426 ,494 ,610 ,700 1,000 ,625 ,519 8 ,608 ,533 ,559 ,492 ,737 ,695 ,625 1,000 ,635 9 ,551 ,633 ,463 ,600 ,548 ,492 ,519 ,635 1,000 > 0,9 means a too high correlation (i.e. most likely identical items); <0,2 means a too low correlation ( i.e. no correlation).
RkJQdWJsaXNoZXIy MjY0ODMw