McFadden 1976 estimated a multinominal logit model of work-trip mode choice for
ID: 3173426 • Letter: M
Question
McFadden 1976 estimated a multinominal logit model of work-trip mode choice for the San Francisco Bay area. A portion of his results are given in table 4.19. The work-trip modes available to the 771 commuters sampled were:
1 automobile be all alone
2 bus with walk access
3 bus with automobile access
4 carpool
The numbers in parentheses or the mode associated with each variable
Explanatory variable. Coefficient estimate. t statistic
Mode cost cents/postage wage cents per minute (1–4) -0.0284 -4.31
Auto in vehicle time minutes (1,3,4 ) -0.0644 -5.65
Worktime minutes (2, 3) -0.0259 -2. 94
Transfer wait time (minutes) (2,3) -0.0689 -5.28
Number of transfers (2.3) -0.105 -0.77
Headway of first bus minutes (2.3) -0.0318 -3.18
Number of drivers in household (1) 1.02 4.81
number drivers in the household (3) 0.99 03.29
number drivers in the household (4) 0.872 4.25
Head of household dummy variable (1) 0.627 3.37
Employment density at work location (1) -0.00160 -2.27
Home near or in CBD (1) -0.502 -4.18
Autos per driver (1) 5.00 9.65
Autos per driver (3) 2.33 2.74
Autos per driver (4) 2.38 5.28
Auto alone constant (1) -5.26 -5.93
bus with auto access constant (3) -5.49 -5.33
Carpool constant (4) -3.84 -6.36
Source McFadden 1976 table 3, P.38
d) what will be the effects on a workers work-trip mode choice who lives near downtown and works in a high density employment area.
Which of the two effects is stronger.
Explanation / Answer
d)Employment density at work location : This variable has low estimate and t-statistic value is relatively smaller than other variables. Hence the effect of this variable is not going to affect much the target.