Integrated Choice and Latent Variable Models for evaluating Flexible Transport Mode choice
Polotis, I., Papaioannou, P., Basbas, S., 2012, in Research in Transportation Business & Management
doi:10.1016/j.rtbm.2012.06.007
Location |
Thessaloniki, Greece |
Population |
Other (specify) |
Sample size |
450 |
Factor analysis type |
principal components, varimax rotation |
Stepwise regression |
no |
Removal of insignificant variables |
no |
Reviewed by |
LCM |
Abstract
Flexible Transport Modes and Services are primarily demand oriented initiatives that presuppose the self motivation of each individual in order to be preferred in a constant way. Given that fact, the examination of the potential success of such interventions should take into account not only quantitative data but also qualitative /behavioural parameters that participate in the mode choice procedure. This examination should follow specific guidelines in order to have common accepted evaluation procedures. In this paper, an Integrated Framework for the ex ante evaluation of a Flexible Transport Mode Schemes, is presented. The proposed framework is implemented in a real life problem: the introduction of Flexible Transport Mode scheme for commuting trips. Following the theories and concepts of the Framework, ICLV (Integrated Choice and Latent Variables) models were developed, in order to estimate the importance of a set of variables into mode choice process, for four alternative to the car modes The models that were developed though the usage of Structural Equation Modeling techniques are hybrid binary choice models and the discrepancy function that was used was the Bayesian estimation. The analysis showed that latent variables can significantly contribute in the process of interpreting the mode choice decision.
Factors
Models
Dependent variable |
Choice of car vs. bus |
Model type |
ICLV |
Sample size |
282000.0 |
R2 |
nan |
Adjusted R2 |
|
Pseudo R2
(nan)
|
nan |
AIC |
nan |
BIC |
nan |
Log-likelihood at zero |
nan |
Log-likelihood at constants |
nan |
Log-likelihood at convergence |
nan |
Choice model |
Variable |
Coefficient |
p-value |
DTinmode |
-0.025
|
0.211 |
DToutmode |
-0.024
|
0.549 |
Dpercost |
-0.019
|
0.527 |
Personal beliefs |
-0.7
|
0.0 |
Traveler characteristics |
-0.14
|
0.046 |
Latent variable model |
Variable |
Coefficient |
p-value |
env1 |
-0.122
|
0.015 |
env2 |
-0.144
|
0.004 |
env3 |
-0.036
|
0.368 |
com1 |
-0.18
|
0.0 |
com2 |
-0.034
|
0.571 |
com3 |
0.116
|
0.097 |
flex1 |
-0.201
|
0.0 |
flex2 |
-0.658
|
0.0 |
flex3 |
-0.491
|
0.0 |
age |
-0.136
|
0.052 |
behstage |
0.172
|
0.455 |
income |
-0.499
|
0.013 |
numcar |
-0.044
|
0.142 |
Variable |
Coefficient |
p-value |
Intercept |
0.53
|
0.077 |
Dependent variable |
Choice of car vs. taxi |
Model type |
ICLV |
Sample size |
179000.0 |
R2 |
nan |
Adjusted R2 |
|
Pseudo R2
(nan)
|
nan |
AIC |
nan |
BIC |
nan |
Log-likelihood at zero |
nan |
Log-likelihood at constants |
nan |
Log-likelihood at convergence |
nan |
Choice model |
Variable |
Coefficient |
p-value |
DTinmode |
-0.019
|
0.057 |
DToutmode |
-0.026
|
0.194 |
Dpercost |
-0.07
|
0.243 |
Personal beliefs |
-0.07
|
0.317 |
Traveler characteristics |
-0.53
|
0.0 |
Latent variable model |
Variable |
Coefficient |
p-value |
env1 |
-0.025
|
0.532 |
env2 |
-0.025
|
0.405 |
env3 |
0.001
|
0.98 |
com1 |
-0.035
|
0.243 |
com2 |
-0.043
|
0.282 |
com3 |
-0.018
|
0.719 |
flex1 |
-0.043
|
0.282 |
flex2 |
-0.039
|
0.33 |
flex3 |
-0.056
|
0.351 |
age |
-0.54
|
0.0 |
behstage |
0.228
|
0.079 |
income |
-1.251
|
0.0 |
numcar |
-0.122
|
0.0 |
Variable |
Coefficient |
p-value |
Intercept |
0.51
|
0.0 |
Dependent variable |
Choice of solo car vs. carpool |
Model type |
ICLV |
Sample size |
712000.0 |
R2 |
nan |
Adjusted R2 |
|
Pseudo R2
(nan)
|
nan |
AIC |
nan |
BIC |
nan |
Log-likelihood at zero |
nan |
Log-likelihood at constants |
nan |
Log-likelihood at convergence |
nan |
Choice model |
Variable |
Coefficient |
p-value |
DTinmode |
-0.023
|
0.021 |
DToutmode |
-0.019
|
0.342 |
Dpercost |
-0.032
|
0.424 |
Personal beliefs |
-0.24
|
0.0 |
Traveler characteristics |
-0.38
|
0.0 |
Latent variable model |
Variable |
Coefficient |
p-value |
env1 |
-0.127
|
0.011 |
env2 |
-0.126
|
0.012 |
env3 |
0.008
|
0.841 |
com1 |
-0.115
|
0.021 |
com2 |
-0.206
|
0.003 |
com3 |
-0.087
|
0.147 |
flex1 |
-0.17
|
0.001 |
flex2 |
-0.145
|
0.016 |
flex3 |
-0.234
|
0.003 |
age |
-0.122
|
0.015 |
behstage |
0.331
|
0.052 |
income |
-0.645
|
0.0 |
numcar |
-0.119
|
0.003 |
Variable |
Coefficient |
p-value |
Intercept |
0.59
|
0.001 |
Dependent variable |
Choice of car vs. metro |
Model type |
ICLV |
Sample size |
408000.0 |
R2 |
nan |
Adjusted R2 |
|
Pseudo R2
(nan)
|
nan |
AIC |
nan |
BIC |
nan |
Log-likelihood at zero |
nan |
Log-likelihood at constants |
nan |
Log-likelihood at convergence |
nan |
Choice model |
Variable |
Coefficient |
p-value |
DTinmode |
-0.029
|
0.004 |
DToutmode |
-0.012
|
0.549 |
Dpercost |
-0.122
|
0.0 |
Personal beliefs |
0.1
|
0.153 |
Traveler characteristics |
0.31
|
0.0 |
Latent variable model |
Variable |
Coefficient |
p-value |
env1 |
0.009
|
0.764 |
env2 |
0.016
|
0.594 |
env3 |
-0.008
|
0.79 |
com1 |
0.053
|
0.185 |
com2 |
0.058
|
0.246 |
com3 |
0.002
|
0.968 |
flex1 |
0.034
|
0.395 |
flex2 |
0.068
|
0.174 |
flex3 |
0.074
|
0.139 |
age |
0.375
|
0.004 |
behstage |
0.141
|
0.407 |
income |
0.356
|
0.011 |
numcar |
0.067
|
0.094 |
Variable |
Coefficient |
p-value |
Intercept |
0.02
|
0.92 |