Effect of survey mode on response patterns: comparison of face-to-face and self-administered modes in health surveys

Abstract

While face-to-face interviews are considered the gold standard of survey modes, self-administered questionnaires are often preferred for cost and convenience. This article examines response patterns in two general population health surveys carried out by face-to-face interview and self-administered questionnaire, respectively. Data derives from a health interview survey in the Region of Southern Denmark (face-to-face interview) and The Danish Health and Morbidity Survey 2010 (self-administered questionnaire). Identical questions were used in both surveys. Data on all individuals were obtained from administrative registers and linked to survey data at individual level. Multiple logistic regression analyses were used to examine the effect of survey mode on response patterns. The non-response rate was higher in the self-administered survey (37.9%) than in the face-to-face interview survey (23.7%). Marital status, ethnic background and highest completed education were associated with non-response in both modes. Furthermore, sex and age were associated with non-response in the self-administered mode. No significant mode effects were observed for indicators related to use of health services, but significant mode effects were observed for indicators related to self-reported health-related quality of life, health behaviour, social relations and morbidity (long-standing illness). The same factors were generally associated with non-response in both modes. Indicators based on factual questions with simple answers categories were overall more comparable according to mode than indicators based on questions that involved more subjective assessments. Other measures such as efficiency and cost-effectiveness of the mode should also be considered when determining the most appropriate form of data collection.