OBJECTIVE: To assess whether an oral health-related quality of life (OHRQoL)measure showed differential item functioning (DIF) by ethnicity. METHODS: A simple random sample of 12- and 13-year-old schoolchildren enrolled in the Taranaki District Health Board's school dental service, New Zealand. Each child (n = 430) completed the Child Perception Questionnaire (CPQ(11-14)) in the dental clinic waiting room, prior to a dental examination. The dataset included age, gender, ethnicity, and deprivation status. The general principle of the analytic plan was that equal scores from each CPQ(11-14) item were expected from both non-Mäori and Mäori groups regardless of their ethnic group. Ordinal logistic regression was performed. The dependent variables were the CPQ(11-14) items. The ethnicity group and each CPQ(11-14) domain score were the independent variables. Non-uniform DIF was assessed through adding an interaction term for each CPQ(11-14) sub-scale. RESULTS: Non-uniform DIF was found in two items, one in the Functional Limitations sub-scale and another in the Social Well-being sub-scale. Uniform DIF was found in one item of the Emotional Well-being sub-scale. CONCLUSION: Both non-uniform and uniform DIF by ethnicity was found in three of 37 items of the CPQ(11-14) questionnaire, showing it is important to perform DIF analysis when applying OHRQoL measures.
OBJECTIVE: To assess whether an oral health-related quality of life (OHRQoL)measure showed differential item functioning (DIF) by ethnicity. METHODS: A simple random sample of 12- and 13-year-old schoolchildren enrolled in the Taranaki District Health Board's school dental service, New Zealand. Each child (n = 430) completed the Child Perception Questionnaire (CPQ(11-14)) in the dental clinic waiting room, prior to a dental examination. The dataset included age, gender, ethnicity, and deprivation status. The general principle of the analytic plan was that equal scores from each CPQ(11-14) item were expected from both non-Mäori and Mäori groups regardless of their ethnic group. Ordinal logistic regression was performed. The dependent variables were the CPQ(11-14) items. The ethnicity group and each CPQ(11-14) domain score were the independent variables. Non-uniform DIF was assessed through adding an interaction term for each CPQ(11-14) sub-scale. RESULTS: Non-uniform DIF was found in two items, one in the Functional Limitations sub-scale and another in the Social Well-being sub-scale. Uniform DIF was found in one item of the Emotional Well-being sub-scale. CONCLUSION: Both non-uniform and uniform DIF by ethnicity was found in three of 37 items of the CPQ(11-14) questionnaire, showing it is important to perform DIF analysis when applying OHRQoL measures.
Authors: I-Chan Huang; Walter L Leite; Patricia Shearer; Michael Seid; Dennis A Revicki; Elizabeth A Shenkman Journal: Value Health Date: 2011 Sep-Oct Impact factor: 5.725
Authors: Marziyeh Doostfatemeh; Seyyed Mohammad Taghi Ayatollahi; Peyman Jafari Journal: Health Qual Life Outcomes Date: 2020-10-21 Impact factor: 3.186