ESRA 2025 Preliminary Program
All time references are in CEST
Best practices for multilingual surveys |
| Session Organisers |
Mrs Janelle van den Heuvel (Statistics Netherlands)
Dr Vera Toepoel (Statistics Netherlands)
|
| Time | Wednesday 16 July, 11:00 - 12:15 |
| Room |
Ruppert 111 |
In today's increasingly multicultural societies, a growing number of languages are spoken by potential respondents. Traditionally, most surveys have been distributed solely in the official languages of the country. However, this practice might no longer be sufficient as it excludes a significant portion of the population that does not speak this language. This exclusion results in biased data that does not accurately reflect the views and needs of the entire population. The critical question remains: how can we effectively reach these individuals while keeping those without a migration background engaged?
To tackle this issue, many countries are now distributing surveys in multiple languages. There is still much uncertainty about the best way to implement multilingual surveys though. We have a lot of questions, among others, about the following topics:
• Approaching respondents who do not speak the official language
• Communicating that the survey is available in multiple languages
• Choosing the appropriate modes for multilingual surveys
• Training interviewers to interact with multilingual respondents
• Effective strategies for implementing multilingual surveys
• The future of multilingual surveys
Join us in this session to share your experiences, insights, and solutions. Let us strive towards an inclusive approach to obtain more representative data in our increasing diverse society.
Keywords: multilingual surveys, representativity, nonresponse, hard-to-reach population,
Papers
How to include non-native speakers in household surveys? Experiences of a multilingual approach at Statistics Netherlands
Mrs Janelle van den Heuvel (CBS) - Presenting Author
Dr Vivian Meertens (CBS)
Mrs Michelle Creemers (CBS)
Mrs Edith Adriaanse (CBS)
Mr Wim Heijboer (CBS)
At statistics Netherlands (CBS), we see a rising number of nonresponses due to language barriers in the CAPI mode of several household surveys. As it is CBS’s duty to produce statistics on the Dutch population, it is of importance to include all relevant groups. Within this project, we aim to see whether an inclusive approach by using the native tongue of the target group is a meaningful way to include them. We took a deep-dive review into various communication and observation methods at CBS and looked into its influence on the response rate in the groups who experienced a language barrier. Several focus groups were conducted with representatives of the different target groups as well as in-depth interviews with experts in the field. Furthermore, two experiments were conducted in which we researched the influence of communication materials in the native language of the target group in the CAWI and CAPI mode. In this presentation, we aim to shed light on our multi-language project and present the results from both these communication experiments as the consultations in the expert- and focus groups. Furthermore, we will present the conclusion that we draw based on these results and aim to answer the question whether attending people in their native tongue is a fruitful approach.
The implications of offering questionnaires in multiple languages.
Dr Sebastian Lundmark (University of Gothenburg) - Presenting Author
Dr Björn Rönnerstrand (University of Gothenburg)
Ms Frida Sandelin (RISE)
In many Western countries, and especially countries like Sweden and Germany, the share of immigrants has increased rapidly over the last four decades, leading to many new languages being spoken and read by the population. Simultaneously, these countries have seen rapidly deteriorating willingness to complete questionnaires. To combat this, lowering sample persons’ thresholds to complete self-administered questionnaires may increase response rates. One such threshold may be the language in which the questionnaire can be completed.
Through three preregistered experiments administered to simple random samples of people living in Sweden, the impact of offering several languages in which to complete self-administered mixed-mode questionnaires (paper-and-pencil and web) was assessed. The results showed that merely offering multiple languages in which to complete the questionnaire does not appear a cure-all for increasing response rates, not even among those who have immigrated from countries where the offered language was the majority language. In the first experiment, the sample persons being offered twelve languages were, surprisingly, less likely to complete the questionnaire than those offered the questionnaire only in Swedish. In the second experiment, offering three languages statistically significantly increased response rates compared to offering it just in Swedish, and was so both for immigrants and for those who were born in Sweden. The third experiment, directly replicating the second experiment, will be completed by January 2025 with its results still unknown but to be presented at the conference.
Minority Languages in The Patient-Reported Indicator Survey (PaRIS), (Organisation for Economic Co-operation and Development)
Miss Lucija Bosnjak (Ipsos UK) - Presenting Author
The Patient-Reported Indicator Survey (PaRIS), an OECD initiative, aims to evaluate healthcare from the patients’ perspective. As the first international survey focusing on individuals with chronic conditions managed in primary care, inclusivity is a core element, achieved through a multilingual approach incorporating both major national and minority languages. The patient and provider questionnaires were translated or adapted for use across 22 countries, resulting in 31 country/language combinations, including five English adaptations. These questionnaires were designed for online self-administration, with adaptations for paper self-completion and telephone interviews where necessary.
To further enhance accessibility, the survey included patient questionnaires in minority languages. This presentation highlights the challenges of providing minority languages in push-to-web survey designs, where language needs are unknown and there is no interviewer mediation. It explores two translation strategies that were used: full translation, which fully converts questionnaires into a minority language when a significant population speaks it, ensuring accessibility but requiring substantial resources and risking underutilisation; and translation on demand, offering flexible, cost-effective translations as needed with language cards and real-time phone interpretation, covering more languages but potentially causing inconsistencies and delays in data collection.
The presentation will discuss adaptations for local contexts and the selection process for minority languages. It will cover case studies from selected countries involved in PaRIS, illustrating approaches to offering minority languages, communicating multilingual availability, and addressing specific issues such as making language options clear on the landing page, both online and via helpline services.
The discussion will highlight the importance of effectively navigating these translation strategies and addressing the complexities of multilingual survey implementation to enhance data reliability and inclusivity within a push-to-web context. Ensuring all voices are heard, particularly from minority language groups, is crucial for achieving a comprehensive understanding of the
