Does the same question asked of the same population yield the same answer in face-to-face interviews when other parts of the questionnaire are altered? If not, what would be the implications for proxy-based poverty measurement? Relying on a randomized household survey experiment implemented in...