544 Discrete Choice Modeling
11.7.4 Application
The multinomial choice models are illustrated with a well-known data survey of
commuters between Sydney and Melbourne (see Greene, 2007a, and references
cited). A sample of 210 travelers between Sydney and Melbourne were asked which
of four travel modes they chose air, train, bus or car. The variables used in the
models are:
TTME=Terminal time, in minutes, zero for car,
INVT=In-vehicle time for the journey
GC=generalized cost=in-vehicle cost + a wage times INVT
HINC=household income
PSIZE=traveling party size:Table 11.10 lists descriptive statistics for the variables in the model. The left-
hand side for each panel lists the means and standard deviations for the variables
Table 11.10 Descriptive statistics for variablesVariable Mean Std. dev. Mean Std. dev.58 observations
AIR All 210 obs. that chose AIR
TTME 61.010 15.719 46.534 24.389
INVT 133.710 48.521 124.828 50.288
GC 102.648 30.575 113.552 33.198
PSIZE 1.743 1.012 1.569 .819
HINC 34.548 19.711 41.724 19.115
63 observations
TRAIN All 210 obs. that chose TRAIN
TTME 35.690 12.279 28.524 19.354
INVT 608.286 251.797 532.667 249.360
GC 130.200 58.235 106.619 49.601
PSIZE 1.743 1.012 1.667 .898
HINC 34.548 19.711 23.063 17.287
30 observations
BUS All 210 obs. that chose BUS
TTME 41.657 12.077 25.200 14.919
INVT 629.462 235.408 618.833 273.610
GC 115.257 44.934 108.133 43.244
PSIZE 1.743 1.012 1.333 .661
HINC 34.548 19.711 29.700 16.851
59 observations
CAR All 210 obs. that chose CAR
TTME .000 .000 .000 .000
INVT 573.205 274.855 527.373 301.131
GC 95.414 46.827 89.085 49.833
PSIZE 1.743 1.012 2.203 1.270
HINC 34.548 19.711 42.220 17.685