I sat in an immaculate office in the Rwandan capital Kigali with a bureaucrat who crossed out words on my questionnaire with a black bic pen. ‘Your research team cannot ask about “ethnic groups”.’ He dug the pen into the paper. I sighed inwardly, although this was the expected response. I wasn’t an independent researcher who could slip under government radar – I was working with a registered NGO. ‘Then, could I use the word “group”, in general?’ I felt confident my participants would understand group to mean ethnic. ‘No’, the bureaucrat shook his head. You can’t use the word group. I started to fidget. I was supposed to research ethnic relations without using the word ‘ethnic’ or ‘group’. ‘What word do you suggest, then?’ He tapped his pen against my questionnaire. ‘Perhaps, “people”. For example in question number 15, “I would be willing to allow my daughter to marry people?”.’ Later that day my Rwandan research assistant spoke in low consoling

tones. ‘We will write “people” and say “people”, but as we explain the question, participants will understand what we mean. Everyone speaks in code here.’ Recent arrests of Rwandan researchers working for an international organization weighed on my mind. ‘But if this code is so easy to understand, what is the difference – aren’t our researchers still at risk for discussing ethnicity?’ He shrugged. ‘Our participants know us, so I don’t think there will be a problem. And I don’t think we have another choice.’