Publicatie

Item non-respons bij CQ onderzoek onder patiënten van allochtone herkomst.

Roset, B., Stronks, K., Sixma, H., Rademakers, J., Arah, O. Item non-respons bij CQ onderzoek onder patiënten van allochtone herkomst. TSG: Tijdschrift voor Gezondheidswetenschappen: 2008, 86(8), p. 471-477.
Download the PDF
Background: The development of consumer experience measures (the so-called consumer quality index or CQ-index) in The Netherlands has not taken into account the data quality, instrument reliability and validity among ethnic minorities. To estimate the quality of data for these groups, ethnic differences in item non-response were assessed. Possible explanations for any ethnic differences were explored. Method: To measure the impact of ethnic background on possible differences in item non-response, multilevel linear regression models were fit to the data of 23058 respondents who filled out the hospital version of CQ-index. The association of ethnicity with overall percentage of non-response on core items was explored, while accounting for the possible explanatory effects of education and Dutch language reading proficiency. Results: There were ethnic differences in item non-response, with item non-response being higher for the Turkish, Moroccan and other non-western ethnic minorities. This effect could only be explained by their education and Dutch language reading proficiency for Moroccan respondents. Discussion: Higher item non-response for ethnic minorities on core aspects of the hospital CQ index results in lower data quality for these respondents. This is probably explained by unmeasured factors, although measuring respondents’ Dutch language proficiency through a written question in Dutch is problematic. Further research into these and related issues should include thorough cognitive testing and extensive response analysis to ensure the reliability, validity and usability of the CQ-index among multi-ethnic groups in the Netherlands.(aut. ref.)