This core of this paper is a description and discussion of a way of quantif
ying data gathered using a Delphi survey, in this case concerned with attem
pting to predict the future of interactive television for retailing in the
luxury and performance car industry. The analysis of the data was based, no
t on the standard cross-impact matrix approach to handling quantifiable inf
ormation within Delphi surveys, but on a novel alternative method which all
ows the data to be expressed in terms of probabilities (the basic equations
are probability statements using binomial probabilities and Bayes's formul
a) and takes into account, in a well-founded way, different levels of respo
ndents' expertise. The method owes something to the work of D.V. Lindley. T
o the knowledge of the authors of this paper, and despite its suitability f
or the purpose, Lindley's approach to quantifying weighted expert opinions
has never before been used in conjunction with a Delphi survey. The analysi
s for an example question from the questionnaire used in the survey is give
n in detail, and the results of four other questions are discussed. (C) 200
0 Elsevier Science Inc.