Interview versus self-completion questionnaires in discrete choice experiments

Since the NOAA (National Oceanic and Atmospheric Administration) panel (1993) on Contingent Valuation (CV), it has been accepted that Willingness to Pay (WTP) data should ideally be collected using only face-to-face interviews and not self-completion surveys. However, there has been little testing of the accuracy of Discrete Choice Experiment (DCE) data collected using self-completion methods, which may actually produce more reliable results because of a reduction in interviewer bias and because respondents are given more time to think. This research, conducted at the South African National Arts Festival (NAF), compares the findings of face-to-face and self-completion surveys using a choice experiment eliciting the willingness of attenders to pay for various attributes of live theatre performances. Results show that attribute coefficients are consistently lower for the self-completion data than for the interview data and, for the model including interaction terms, have lower SEs for the majority of the coefficients. WTP estimates are also lower and, given ticket prices, more realistic, when using the self-completion data.