Do preceding questions influence the reporting of childbearing intentions in social surveys?

For demographers fertility intentions are a long standing source of both interest and scepticism. Scepticism has been expressed because fertility intentions regularly fail to precisely predict fertility and because they are liable to change across the life course. Here we demonstrate an additional consideration: simply changing the questions that precede fertility intentions questions can have a significant influence on responses. We illustrate this risk using a series of randomised experiments with different preceding questions; first, on mortality and risk in two convenience samples of UK undergraduate students. Secondly, we will present provisional results from a ground-breaking longitudinal experiment where the manipulated preceding questions are on close family and friends. As far as we are aware this later study is the first time that question ordering experiment looking at fertility intentions has been embedded in a representative survey, and the first longitudinal measurement of preceding-question effects using the same individuals.