Comparability of questionnaire and interview responses on health among people with foreign origin

• Published in: European journal of public health 2020-09, Vol.30 (Supplement_5)
• Main Authors: Lilja, E, Seppänen, A, Kuusio, H
• Format: Article
• Language: English
• Subjects: Data collection; Health; Interviews; Loneliness; Migrants; Polls & surveys; Public health; Questionnaires; Respondents; Statistical analysis; Well being
• ISSN: 1101-1262; 1464-360X
• DOI: 10.1093/eurpub/ckaa166.1312

Abstract Background In health surveys, self-administered questionnaires are a good way to collect data most cost-effectively. For some population groups, a mixture of methods might be necessary in order to reach the desired response rate. If the collection methods vary, the comparability of data might become an issue when evaluating differences between two population groups or changes over time. This study examines the comparability between questionnaire and phone interview responses in a survey on migrants in Finland. Methods This study uses the data collected in Survey on Well-Being Among Foreign Born Population (FinMonik), conducted in 2018-19. There were 6 836 respondents, out of which 6 505 filled the questionnaire and 331 were interviewed by phone. The survey included questions 'Do you ever feel lonely?' and 'Do you find that your current state of health is:'. Both were answered using a five-point Likert scale ranging from 'never'/'good' to 'all the time'/'bad'. Logistic regression was used to assess statistical differences. Results Among those interviewed by phone, the distributions were more skewed than among those who had filled the questionnaire. In the phone interviews, the emphasis was on the socially more desirable options: regarding loneliness, 'never' was selected 1.7 times more often in the phone interview compared to the questionnaire while 'good' was used to describe the state of health 1.3 times more often. On the common binary indicator 'fairly often or all the time', those who had filled the questionnaire appeared lonelier than those interviewed by phone (14.5% vs 10.4%, p = 0.032), and the difference remained significant when adjusted with a range of sociodemographic variables. However, there was no significant difference on the binary indicator of perceived health ('good or fairly good'). Conclusions The distributions differed considerably between questionnaire and interview responses, but the level of impact varied between indicators. Key messages The data collection methods should be examined carefully when comparing results from different surveys. Instruments designed for questionnaires may yield different results when asked by an interviewer.