Feasibility of Using a Multilingual Web Survey in Studying the Health of Ethnic Minority Youth

Abstract

BackgroundMonolingual Web survey is a common tool for studying adolescent health. However, national languages may cause difficulties for some immigrant-origin youths, which lower their participation rate. In national surveys, the number of ethnic minority groups is often too small to assess their well-being. ObjectiveWe studied the feasibility of a multilingual Web survey targeted at immigrant-origin youths by selection of response language, and compared participation in different language groups with a monolingual survey. MethodsThe Adolescent Health and Lifestyle Survey (AHLS), Finland, with national languages (Finnish/Swedish) was modified into a multilingual Web survey targeted at a representative sample of 14- and 16-year olds (N=639) whose registry-based mother tongue was other than the national languages. The survey was conducted in 2010 (16-year olds) and 2011 (14-year olds). The response rate of the multilingual survey in 2011 is compared with the AHLS of 2011. We also describe the translation process and the e-form modification. ResultsOf the respondents, 57.6% answered in Finnish, whereas the remaining 42.4% used their mother tongue (P=.002). A majority of youth speaking Somali, Middle Eastern, Albanian, and Southeast Asian languages chose Finnish. The overall response rate was 48.7% with some nonsignificant variation between the language groups. The response rate in the multilingual Web survey was higher (51.6%, 163/316) than the survey with national languages (46.5%, 40/86) in the same age group; however, the difference was not significant (P=.47). The adolescents who had lived in Finland for 5 years or less (58.0%, 102/176) had a higher response rate than those having lived in Finland for more than 5 years (45.1%, 209/463; P=.005). Respondents and nonrespondents did not differ according to place of birth (Finland/other) or residential area (capital city area/other). The difference in the response rates of girls and boys was nearly significant (P=.06). Girls of the Somali and Middle Eastern language groups were underrepresented among the respondents. ConclusionsA multilingual Web survey is a feasible method for gathering data from ethnic youth, although it does not necessarily yield a higher response rate than a monolingual survey. The respondents answered more often in the official language of the host country than their mother tongue. The varying response rates by time of residence, ethnicity, and gender pose challenges for developing tempting surveys for youth.