Factors Affecting the Vallidity of Stated Preference
SWANSON J, Steer Davies Gleave, UK
This paper describes a study funded by the Seedcorn research programme of the UK Department of Environment Transport and the Regions (DETR). The study was designed to look for empirical evidence of how well forecasts produced using Stated Preference (S P)
This paper describes a study funded by the Seedcorn research programme of the UK Department of Environment Transport and the Regions (DETR). The study was designed to look for empirical evidence of how well forecasts produced using Stated Preference (S P) actually perform. In the field of transport this is a harder task than it might first appear. Transport projects can take years to build, alter which many of the assumptions underlying the forecasts may no longer hold. Forecasts are almost never produced solely on the basis of SP, but will usually rely on other types of data and forecasting techniques, so that it may be difficult to ascribe the success or failure of the forecasts to the SP component. And of course some projects will never be built because the forecasts appeared not to justify them, and in these cases no empirical evidence exists.
It was therefore proposed that the study should look outside the transport field at work done in general market research. We hoped to find examples where the time lapse between forecasts and product delivery was short, and where the forecasting methodology was based substantially or entirely on SP so that clearer inferences could be drawn about the relationship between outcome and the methodology. We also felt that the market would be much larger than in transport, with the potential for many more examples, and chose to concentrate on the United States, which is probably the biggest single market, with the longest track record of experience.
Coupled with this, we also proposed to review the work undertaken in recent years by experimental psychologists and economists into how people make choices, and how preferences are formed and expressed. This was prompted by earlier work undertaken by us (Ampt et al., 1995) which suggested that there was scope for improving SP practice by taking a more careful view of how choices are actually made.
For both parts of the work the methodology was to review the literature and correspond with practitioners in the US. No new survey work was undertaken.
In fact we quickly found that there is little or no published material comparing forecasts and actual outcomes. We were told that plenty such evidence exists, but that it is commercially confidential. This led us to change the emphasis of the project. The literature on Conjoint Analysis, which is to general market research what SP is to transport, is large, and a vigorous debate has been taking place among academics and practitioners about best practice. We have reviewed this literature and drawn from it recommendations which, we propose, could be used to improve the practice of SP in transport.
This rest of this paper falls into four sections. In section two we give a review of the development of conjoint, and of ideas from conjoint which we believe to be useful. We then turn to the work of the psychologists and economists, and their insights into choice processes. The fourth part discusses analysis and interpretation of SP results, and finally we offer recommendations for improving the practice of SP.
Association for European Transport