commute by car two times or more per week and face congestion of 10 or more minutes per
trip for at least two times a week. Car commuters have been selected for the sample since they
are likely to be confronted with road pricing once implemented. The only selection criterion
for the other half (249 respondents) of the sample is that respondents had to possess a car.
Taking this other group into account makes it also possible to compare effects of a road
pricing measure between different user classes. After a pricing measure was shown
respondents who had a job (i.e. 422 of the 512 respondents) were asked to indicate the
probability that they would change to another residential location closer to work. A second
question aimed at the probability of searching for another job closer to the residential location.
The response scale for both questions consisted of 7 categories (Likert-scale) ranging from
‘highly unlikely’ to ‘highly likely’.
To investigate the relative influence of trip and more location related variables in the actual
residential location choice (second goal), data from a stated choice experiment among 564
respondents is used. Again the respondents were commuters, who drive to work by car two
times or more per week and face congestion of 10 or more minutes per trip for at least two
times a week. To every respondent 9 hypothetical choice situations were shown, consisting of
two alternatives. The total design of the experiment consisted of 27 choice situations.
Therefore, three blocks of 9 screens were randomly assigned to the respondents. The
experiment was generic (i.e. non-alternative specific). This means that both alternatives
consisted of the same attributes and that alternatives were not labelled (for example not one
alternative always having higher toll costs). The alternatives within the experiment were:
number of bedrooms, the monthly rent or mortgage costs of the house, the location ((large)
city, medium sized city, small village/rural area), the travel time (free flow and time in
congestion) and travel costs (road pricing and fuel costs). Every attribute systematically
varied at 3 levels. The actual values shown were tailored to the specific situation of the
respondents such as the actual commuting distance. The number of bedrooms presented in the
choice screens was made dependent on the type of housing. The monthly housing cost in the
experiment furthermore, was varied around the actual housing cost. Additionally, a distinction
was made between rent and mortgage costs. Fuel costs for respondents who get fuel cost
compensation in the current situation were set to zero within the experiment. Finally, the set-
up of the experiment aimed at making differentiations in monthly cost on average comparable
to travel cost variations (including monetarized travel time, toll and fuel costs).