The effect of attitudes on reference-dependent preferences: Estimation and validation for the case of alternative-fuel vehicles

Mabit, Cherchi, Jensen, Jordal-Jørgensen, 2015, in Transportation Research Part A: Policy and Practice

doi:10.1016/j.tra.2015.08.006
Location Denmark
Population General
Sample size 2093
Factor analysis type exploratory factor analysis, unknown rotation
Stepwise regression no
Removal of insignificant variables yes
Reviewed by LCM

Abstract

Several recent studies in transportation have analysed how choices made by individuals are influenced by attitudes. Other studies have contributed to our understanding of apparently non-rational behaviour by examining how choices may reflect reference-dependent preferences. This paper examines how reference-dependent preferences and attitudes together may explain individual choices. In a modelling framework based on a hybrid choice model allowing for both concepts, we investigate how attitudes and reference-dependent preferences interact and how they affect willingness-to-pay measures and demand elasticities. Using a data set with stated choices among alternative-fuel vehicles, we see that allowing for reference-dependent preferences improves our ability to explain the stated choices in the data and that the attitude (appreciation of car features) explains part of the preference heterogeneity across individuals. The results indicate that individuals have reference-dependent preferences that could be explained by loss aversion and that these are indeed related to an individual's attitude towards car features. The models are validated using a large hold-out sample. This shows that the inclusion of attitudes improves the models' ability to explain behaviour in the hold-out sample. While neither reference-dependent preferences nor the attitude affect the average willingness-to-pay measures in our sample, their effect on choice behaviour has implications for policy recommendations as segments with varying attitudes and reference values will act differently when affected by policy instruments related to the demand for alternative-fuel vehicles, e.g. subsidies. © 2015 Elsevier Ltd.

Factors

Variable Pattern loading
Road position of the car (importance) (Other) 0.52
Joy of driving the car (importance) (Mode-liking) 0.48
Car comfort (importance) (Comfort) 0.67
Car design (importance) (Other) 1.08
It is important for me to drive a car that I like (Mode-liking) 0.87
I notice the type of car others drive (Other) 1.7
Variable Pattern loading
nan () nan

Models

Dependent variable Vehicle choice
Model type MNL
Sample size 14694.0
R2 0.221
Adjusted R2
Pseudo R2 (nan) nan
AIC nan
BIC nan
Log-likelihood at zero nan
Log-likelihood at constants nan
Log-likelihood at convergence -7910.0
Variable Coefficient p-value
Conventional 0.0 nan
Hybrid ASC 0.51 0.0
Bio-diesel ASC 0.2 0.002
Electric ASC 0.75 0.0
Electric ASC*male -0.34 0.0
Hybrid/Bio-diesel/Electric ASC*worker 0.2 0.005
Acceleration -0.15 0.549
Acceleration*male -0.94 0.001
Acceleration*age 30-44 -0.98 0.036
Annual cost -0.52 0.0
Annual cost*dist>25 -0.35 0.004
Range 0.96 0.0
Range*age 30-44 0.91 0.0
Range*dist>25 0.5 0.001
Price -1.72 0.0
Price*age>60 0.23 0.028
Price*children -0.22 0.012
Price*dist<4 0.27 0.005
Price*high income 0.36 0.0
Price*single -0.66 0.0
Service 0.17 0.0
Service*diesel 0.14 0.009
Dependent variable Vehicle choice
Model type MNL
Sample size 14694.0
R2 0.227
Adjusted R2
Pseudo R2 (nan) nan
AIC nan
BIC nan
Log-likelihood at zero nan
Log-likelihood at constants nan
Log-likelihood at convergence -7855.0
Variable Coefficient p-value
Conventional 0.0 nan
Hybrid ASC 0.57 0.0
Bio-diesel ASC 0.32 0.0
Electric ASC 0.92 0.0
Electric ASC*male -0.33 0.0
Hybrid/Bio-diesel/Electric ASC*worker 0.21 0.004
Acceleration -0.18 0.484
Acceleration*male -0.94 0.001
Acceleration*age 30-44 -0.98 0.036
Annual cost -0.51 0.0
Annual cost*dist>25 -0.35 0.004
Range 0.95 0.0
Range*age 30-44 0.92 0.0
Range*dist>25 0.5 0.001
Price -1.7 0.0
Price*age>60 0.27 0.007
Price*children -0.16 0.057
Price*dist<4 0.23 0.012
Price*high income 0.34 0.0
Price*single -0.57 0.0
Service 0.15 0.0
Service*diesel 0.19 0.001
n0 (reference effect in the purchase price) 0.27 0.0
n*diesel (reference effect in the purchase price) 0.14 0.009
n*loan (reference effect in the purchase price) -0.13 0.012
Dependent variable Vehicle choice
Model type HCM
Sample size 14694.0
R2 0.224
Adjusted R2
Pseudo R2 (nan) nan
AIC nan
BIC nan
Log-likelihood at zero nan
Log-likelihood at constants nan
Log-likelihood at convergence -7869.0
AFC
Variable Coefficient p-value
Constant 2.04 0.0
Age > 44 -0.1 0.0
Commute distance < 4 km -0.16 0.0
Commute distance > 25 km 0.06 0.0
Dummy for diesel reference vehicle -0.13 0.0
Dummy for freuent car user -0.2 0.0
Dummy for car financed by loan -0.09 0.0
Sigma_lv 0.62 0.0
Variable Coefficient p-value
Conventional 0.0 nan
Hybrid ASC 0.14 0.271
Bio-diesel ASC -0.1 0.424
Electric ASC -0.17 0.23
Electric ASC*male -0.35 0.0
Hybrid/Bio-diesel/Electric ASC*worker 0.16 0.028
Acceleration -1.02 0.036
Acceleration*male -0.92 0.001
Acceleration*age 30-44 -0.95 0.046
Annual cost -0.16 0.317
Annual cost*dist>25 -0.32 0.007
Range 0.67 0.001
Range*age 30-44 0.91 0.0
Range*dist>25 0.48 0.001
Price -0.6 0.0
Price*age>60 0.26 0.012
Price*children -0.18 0.028
Price*dist<4 0.15 0.089
Price*high income 0.33 0.0
Price*single -0.61 0.0
Service 0.18 0.0
Service*diesel 0.14 0.009
Hybrid ASC*Appreciation of car features 0.25 0.0
Bio-diesel ASC*Appreciation of car features 0.21 0.002
Electric ASC*Appreciation of car features 0.59 0.0
Acceleration*Appreciation of car features 0.51 0.046
Annual cost*Appreciation of car features -0.24 0.028
Range*Appreciation of car features 0.2 0.11
Price*Appreciation of car features -0.73 0.0
Dependent variable Vehicle choice
Model type HCM
Sample size 14694.0
R2 0.228
Adjusted R2
Pseudo R2 (nan) nan
AIC nan
BIC nan
Log-likelihood at zero nan
Log-likelihood at constants nan
Log-likelihood at convergence -7823.0
AFC
Variable Coefficient p-value
Constant 2.04 0.0
Age > 44 -0.1 0.0
Commute distance < 4 km -0.16 0.0
Commute distance > 25 km 0.06 0.0
Dummy for diesel reference vehicle -0.13 0.0
Dummy for freuent car user -0.2 0.0
Dummy for car financed by loan -0.09 0.0
Sigma_lv 0.62 0.0
Variable Coefficient p-value
Conventional 0.0 nan
Hybrid ASC 0.24 0.057
Bio-diesel ASC 0.11 0.368
Electric ASC 0.15 0.368
Electric ASC*male -0.33 0.0
Hybrid/Bio-diesel/Electric ASC*worker 0.17 0.021
Acceleration -1.1 0.021
Acceleration*male -0.93 0.001
Acceleration*age 30-44 -0.96 0.046
Annual cost -0.14 0.368
Annual cost*dist>25 -0.33 0.007
Range 0.67 0.002
Range*age 30-44 0.93 0.0
Range*dist>25 0.48 0.001
Price -0.6 0.0
Price*age>60 0.29 0.002
Price*children -0.13 0.11
Price*dist<4 0.13 0.134
Price*high income 0.32 0.0
Price*single -0.55 0.0
Service 0.16 0.0
Service*diesel 0.19 0.001
n0 (reference effect in the purchase price) 0.52 0.0
n*diesel (reference effect in the purchase price) 0.13 0.009
n*loan (reference effect in the purchase price) -0.17 0.001
Hybrid ASC*Appreciation of car features 0.22 0.001
Bio-diesel ASC*Appreciation of car features 0.14 0.036
Electric ASC*Appreciation of car features 0.49 0.0
Acceleration*Appreciation of car features 0.55 0.028
Annual cost*Appreciation of car features -0.24 0.021
Range*Appreciation of car features 0.19 0.134
Price*Appreciation of car features -0.73 0.0
n*Appreciation of car features -0.15 0.0

The Attitudes and Travel Database is produced with support from the Center for Teaching Old Models New Tricks at Arizona State University, a University Transportation Center sponsored by the US Department of Transportation through Grant No. 69A3551747116.

sha256:a08d9e369743bf7e6d1c40d27347318209b40a7fb1543813fdcf31b898918815