From: A Rasch analysis of the self-administered Foot Health Assessment Instrument (S-FHAI)
Step | Psychometric property | Statistical approach and criteria | Results (original S-FHAI) |
---|---|---|---|
1 | Rating scale functioning: Does the rating scale function consistently across the items? | • Average measures for each category and threshold on each item should advance monotonically • Zstd values < 2.0 in outfit mean square (MnSq) values for step category calibrations [27] | Four items had less than 10 responses per category. Two categories were combined, and then only two items did not meet the criteria |
2 | Internal scale validity: How closely do the item responses match the responses expected according to the Rasch model? | • Item goodness-of-fit values with MnSq values between 0.6 and 1.4 [28] | All items met the criterion |
3 | Internal scale validity: Is the scale unidimensional? | • Principal component analysis, with the first component explaining ≥ 50 % of total variance and any additional component explaining < 5 % (or eigenvalue < 2.0) of the remaining variance [26] | First component explained 54.2 % of total variance, and second component explained 5.9 % (eigenvalue 3.25) of total variance |
4 | Person-response validity: How closely do the individual responses match the responses expected according to the Rasch model? | • Person goodness-of-fit statistics with infit MnSq < 1.4 and Zsrd value ≤ 2.0 [29] • ≤ 5 % of the sample fails to demonstrate acceptable goodness-of-fit values [29] | 34 % of participants failed to demonstrate acceptable goodness-of-fit values |
5 | Person-separation reliability: Can the S-FHAI distinguish between two distinct foot-health groups in the sample? | • Person-separation index ≥ 2.0 [30] | 0.37 |
6 | Differential test functioning (DIF): Are the item difficulty calibrations stable in relation to gender and age? | • Mantel–Haenszel statistic for polytomous scales using log-odds estimators in WINSTEPS software (P < 0.01) [31, 32] | Gender: all item difficulty calibrations were stable Age: six items had DIF in relation to age |