Premium
The response scale for the intellectual disability module of the WHOQOL: 5‐point or 3‐point?
Author(s) -
Fang J.,
Fleck M. P.,
Green A.,
McVilly K.,
Hao Y.,
Tan W.,
Fu R.,
Power M.
Publication year - 2011
Publication title -
journal of intellectual disability research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.941
H-Index - 104
eISSN - 1365-2788
pISSN - 0964-2633
DOI - 10.1111/j.1365-2788.2011.01401.x
Subject(s) - scale (ratio) , item response theory , cronbach's alpha , likert scale , psychology , reliability (semiconductor) , confirmatory factor analysis , point (geometry) , psychometrics , construct validity , field (mathematics) , polytomous rasch model , statistics , clinical psychology , developmental psychology , mathematics , structural equation modeling , geography , power (physics) , physics , geometry , cartography , quantum mechanics , pure mathematics
Abstract Objective To deal with the question of whether a 5‐point response Likert scale should be changed to a 3‐point scale when used in the field testing of people with intellectual disabilities (IDs), which was raised after the pilot study of World Health Organization Quality of Life (WHOQOL)‐DIS, a module being developed with the World Health Organization measure of quality of life for disabilities. Methods Three possible ways were used to generate hypothetical data by merging a 5‐point scale into a 3‐point scale. The analyses were based on both item response theory and classical measurement theory. The partial credit model for polytomous response was performed for item evaluation; the confirmatory factor analysis was used to check construct validity, the Cronbach's alpha for domain reliability, and correlation analyses for the relationship between the 5‐point scale and the generated 3‐point scale. Results Most items with a 5‐point response scale had disordered response options and/or unequal‐length intervals between successive response options; these deficiencies were removed or improved without decline of validity and reliability in the hypothetical data of 3‐point scales. Conclusion Instead of the 5‐point scale, a 3‐point scale could be used for IDs in the field test of developing the module WHOQOL‐DIS.