Table 3
Negative Binomial Regressions for Number of Person-Trips: All Purposes and Recreational Purposes
Variable
Beta
SE
t-stat
P-value
Negative binomial regression
Poisson regression
Dependent variable: number of person-trips for all purposes
Constant
1.672
0.012
138.00
0.000
HH members per vehicle owned
0.255
0.004
59.20
0.000
Income per household member
–3.16E-06
2.77E-07
–11.40
0.000
Population density
–4.42E-06
6.10E-07
–7.25
0.000
Vehicle age
–0.014
0.001
–20.80
0.000
Weekend day indicator
0.100
0.007
13.50
0.000
SUV indicator
0.045
0.012
3.85
0.000
Pickup indicator
–0.164
0.011
–14.90
0.000
Minivan indicator
0.302
0.011
28.50
0.000
Over-dispersion parameter
0.351
0.004
97.10
0.000
Log-likelihood Function for number of person-trips for all purposes
Constant only
–121,113.3
–160,951.3
Convergence
–117,983.1
–148,642.7
Pseudo-R2
0.026
0.076
Dependent variable: number of person-trips for recreational purposes
Constant
0.485
0.021
22.80
0.000
HH members per vehicle owned
0.204
0.008
25.90
0.000
Income per household member
–3.14E-06
4.73E-07
–6.60
0.000
Population density
–6.12E-06
1.09E-06
–5.60
0.000
Vehicle age
–0.018
0.001
–15.60
0.000
Weekend day indicator
0.590
0.014
41.40
0.000
SUV indicator
0.005
0.021
0.20
0.814
Pickup indicator
–0.292
0.019
–15.20
0.000
Minivan indicator
0.308
0.020
15.70
0.000
Over-dispersion parameter
1.009
0.011
93.40
0.000
Log-likelihood function for number of person-trips for recreational purposes
Constant only
–82,460.37
–10,6536.4
Convergence
–80,612.10
–99,753.51
Pseudo-R2
0.022
0.064
Number of observations: 41,538 vehicles
Model form: Number of person-trips ~ negative binomial with expected value
exp(β′X) and non-negative over-dispersion parameter
