Notable Discrete Choice Modelling projects prior to 2003. *Future and Simple was not directly involved in these studies.

Bay Area Rapid Transit (BART) - San Francisco California

The first major test of discrete choice methods was a study on commuter uptake of San Francisco's BART, which opened in 1975. Notes Nobel laureate Dan McFadden of his 1972 study: "The official forecast in 1973 was that it would carry 15 percent of all commute trips in the Bay Area. We predicted 6.3%. The actual share in 1975 turned out to be 6.2% percent, so we did well." McFadden’s 2000 Nobel Prize was largely due to the theory and applications he pioneered on this study.

citation

Australian Defence Force - Personnel Retention

A few years ago, the Australian Defence Force started using choice modelling in order to optimise the package benefits and responsibilities offered to personnel in different roles - thereby maximising retention of highly trained people. According to Jordan Louviere of University of Technology Sydney, the implementation of the project's recommendations saw Royal Australian Air Force pilot attrition rates fall from 21% to 3%. Jordan also found that no change in package and conditions would significantly increase or decrease the pool of Navy submariners – there is simply a small number of people psychologically disposed to do the job.

The Harvard University Project on Faculty Appointments

Research undertaken by Harvard University’s Graduate School of Education (Project on Faculty Appointments, Dr. Cathy Trower, Project Manager), aimed to retain Harvard's pre-eminent position, by identifying the optimum composition of job packages, thereby continuing to attract the best candidates.

The Harvard studies found that geographical location dominates choices for both tenure and non-tenure track job offers, and fairly large salary differences are required to offset undesirable locations (desirable locations can pay less, but less desirable locations must pay more to attract an individual). Likelihood of tenure and a department’s national ranking are next, then university ranking, with work balance relatively less important for tenure track offers. For non-tenure track offers, university ranking and work balance are less important, while contract length seems unimportant.

The choice models also allowed Harvard to profile the best targets for each job offer.

Electricity Deregulation - Benchmark Pricing and Service Levels

DCM has been used as the core scientific basis to establish regulatory benchmarks for electricity services in both South Australia and New South Wales in the lead up to privatisation. The danger in setting minimum service levels too high is that the costs of compliance will lead to entrants in the deregulated markets charging higher prices - for levels of service that customers aren’t very interested in. In a regulated market, consumers have not been exposed to a range or choice of services and prices, so historical data is not informative. DCM projects identified the appropriate levels of uptime and fault response for businesses and households.