Comparison between the English and Bahasa Malaysia language versions of the Visual Functioning Questionnaire (VFQ-25) for use in patients with cataracts

After construct validation was computed, Cronbach’s alpha was computed for the revised questionnaire and we obtained Cronbach alpha values of 0.904, 0.898 and 0.608 for the three components of the English version of the questionnaire – difficulty with activities, responses to vision problems and general health and vision. As for the Bahasa Malaysia version, computation of Cronbach’s alpha revealed values of 0.921, 0.927 for the first two components - difficulty with activities, responses to vision problems. Cronbach’s alpha was not done on Component 3 (general health and vision) of the Bahasa Malaysia version as Component 3 has only one item.

We then proceeded to perform the Item Response Theory analysis via the use of the GPCM model. We analyzed separately each of the components (factors) that were produced via factor analysis earlier for both the English and Bahasa Malaysia questionnaires.

Table 5 illustrates the parameter estimates and fit statistics for the English questionnaire. All three components indicated good to excellent fit of the items towards each of the components with the root mean square error of approximation (RMSEA) values of less than 0.06 and P value (for the Chi-square Goodness of Fit Test) > 0.05 for all items. In addition, the items displayed high discrimination with values ranging from 0.74 to 1.79 for Component one: “Difficulty with activities”; values ranging from 0.90 to 2.08 for Component two: “Responses to vision problems”; and 0.65 to 1.25 for Component three: “General health and vision”.

Figure 1 displays the item characteristic curves (ICC) with curves comprising of P1 and up to P5 or P6 representing the Likert scales utilized in each item – e.g. Likert scale 1 for P1 curve, Likert scale 2 for P2 curve and so on. Those scoring P1 (lowest scale) were more likely seen to the left of each ICC indicating low scoring amongst those with low ability (low “difficulty with activities”) while conversely those scoring P5 or P6 (highest scale) were more likely seen towards the right of the ICCs pointing to high scoring amongst those with high ability (high “difficulty with activities”). Visually, the curves for all of the items indicated that they are well constructed items as there are adequate ordering of these curves within each item from the lowest scale on the left to the highest scale on the right. In addition, the items between the extremes (P2 curve to either P4 or P5) are bunched right in between the first (P1) and the last scale curve (P5 or P6). For instance, a poorly constructed item would have their peaks either to the left of the P1 curve or to the right of the last curve (P5 or P6).

As per Table 6, similar results were seen in the Bahasa Malaysia questionnaire with most items displaying good to excellent fit (RMSEA values of less than 0.06) towards the two analyzed components - Component one: “Difficulty with activities”, Component two: “Responses to vision problems”. Three items (B2Q12, B2Q13, B2Q14) in Component one: “Difficulty with activities” and a further three items (B3Q21, B3Q24, B3Q18) in Component two: “Responses to vision problems” have RMSEA values between 0.06 and 0.08 and thus considered to be fair fit towards their respective components (factors). Moreover, the items illustrated high discrimination with values ranging from 0.78 to 2.29 for Component one: “Difficulty with activities” and values ranging from 1.08 to 4.35 for Component two: “Responses to vision problems”.