Bangladesh + 1 more

For Rohingya, trust begins with who is asking the questions, May 2021


The Kutupalong–Balukhali expansion site in Cox’s Bazar, Bangladesh has become one of the largest refugee settlements in the world, since the arrival of more than 700,000 Rohingya in 2017. Surveys and needs assessments among the refugee population are typically designed in English, translated to Bangla, then interpreted during enumeration to Rohingya by Bangladeshi humanitarian workers or interviewers who speak Chittagonian, a local dialect. Chittagonian is often considered mutually intelligible with Rohingya, but this has been questioned by Translators Without Borders (TWB) and other humanitarian actors. One TWB study found the two languages do not use similar words for many important concepts, while the Joint-Multi Sector Needs Assessment (J-MSNA) reported difficulties among refugee communities in understanding information when not delivered in Rohingya. Moreover, written scripts for Rohingya are new and not in wide use among the population. Only a third of refugee households in Cox’s Bazar are able to read and write.

Findings from a 2019 REACH pilot assessment indicate survey results from Bangladeshi interviewers can exhibit considerable bias for perception-related questions. A recent qualitative study from the International Organization for Migration (IOM) and ACAPS suggests this bias can be reduced by working with Rohingya interviewers. In early 2021, Ground Truth Solutions (GTS) in partnership with IOM conducted its fifth round of surveys with Rohingya aid recipients to gauge their views on the humanitarian response. To ascertain whether the ethnicity of the interviewer had an effect their responses, surveys were conducted by both Rohingya and Bangladeshi interviewers across the same locations. We surveyed five camps (2E, 9, 15, 18, 20), with a sample size of approximately 120 per camp and interviewer type.

Key findings

  • Respondents interviewed by Rohingya express much lower levels of satisfaction with aid services than those interviewed by Bangladeshis. Similar effects were observed for most questions on whether a particular aid service improved over the last 12 months.

  • Differences between the interviewer types were also significant on topics around safety, respect and information provided by aid agencies. This variation aligns with the social desirability of the answer option. Rohingya interviewers are more likely to elicit views that are socially undesirable, and less likely to capture perceptions and behaviors that are socially desirable.


Ground Truth Solutions
added project to body as per case 50068