Methodological appendix

In each of the four countries both a mass and an elite survey were undertaken using, as far as possible, a common questionnaire. In all, 2,030 respondents were interviewed. In Bangladesh a written Bangla (Bengali) translation of the questionnaire was used, and in India a written Kannada version. Both in Ghana and in Côte d'Ivoire the variety of local languages, some unwritten, was such that interviewers used either an English (Ghana) or French (Cote d'Ivoire) version as the basis for their work. In southern Ghana interviewers used mainly the Twi language; some who knew Krobo, Ga or Ewe were used for the very small number of respondents who felt more comfortable in those languages. At training sessions, interviewers practised simulations…