If survey research is research that consists of interviewing large samples of persons, or asking them to complete a standardized questionnaire—and I agree that's what a survey is—then surely the social sciences would have long ago broken through to far more sophisticated methodologies for addressing theoretically relevant hypotheses. Interview and questionnaire studies are rather primitive data-collection techniques, however sophisticated the concomitant sampling techniques or data-analytic techniques might be.

Therefore, I find it a bit ironic that we are, in this symposium, finding so much to debate, and so many propositions to consider about survey research. What, after all, is there to say about survey research that hasn't already been said? Aha! you say, we're not talking about survey research per se, but survey research in Africa. Well, that is indeed so, but some of the most important things that must be said about survey research apply to survey research wherever it is done, in Dar es Salaam or Swarthmore or Syracuse. And before turning to some issues raised by the authors of the core paper in this symposium, which issues pertain particularly, if not uniquely, to survey research in Africa, let me make some general observations.

First, survey data are empirical data, and that's good, because empirical data are better than guesses, or deductions, or intuition, or even (I think) divine revelation. But survey data reveal only what people say in response to what the survey research asks.