Discreet Choice Experiment on car sales in Norway

Published: 18-10-2018| Version 1 | DOI: 10.17632/m26wv676yn.1
Contributors:
Steffen Kallbekken,
William Brazil,
Håkon Sælen,
James Carroll

Description

The data is from an online survey administered to a representative population of former and prospective car buyers in Norway. This dataset contains the discrete choice experiment (DCE) data collected as part of the survey, stored in NLogit format. Codebook: Option - 1 (car one); 2 (car two); 3 (neither) Choice - respondent choice Treated - 0 (non-monetary framing); 1 (monetary framing) Safety - % of max Euro NCAP rating Energy - litres per 10km Capacity - litres of boot capacity Cost - car price in NOK ID - respondent identifier This research set out to examine the role that monetary running cost information can play in terms of highlighting the fuel efficiency of new vehicles. Specifically, this study involved the distribution of a split sample (control/treatment) discrete choice experiment to a representative sample of the Norwegian car buying population, via an online survey undertaken in late 2017. This survey was distributed to over 1000 individuals representing a cross section of the Norwegian population in all regions of the country. Prior to the distribution of the survey, a series of focus groups identified safety rating and luggage space as the most important attributes to include in the experiment, in addition to the research parameters of interest: purchase price and energy efficiency. Attribute levels were selected to reflect those currently present in the Norwegian automobile market, see Table 1. A fractional factorial design, utilising the JMP software package, generated 32 unique choice pairs. To prevent respondent fatigue, these pairs were split across four survey blocks, so that each respondent faced only eight choices. These eight choices were presented in either the control or treatment format, with each respondent only receiving choices in a single format to avoid any framing contamination effects. Therefore, there were eight versions of the survey in total, four control and four treatment blocks. In the control version of the experiment, the attributes were displayed in a simplified version of how they are currently displayed on new cars in Norway. In the treatment version, the energy consumption variable was augmented with a monthly running cost estimate, displayed in terms of Norwegian Kroner (NOK). Both the treatment and control images also contained a graphic with the vehicle’s environmental rating (A-G), as mandated under current EU and Norwegian legislation. The rating is based on CO2-emssions, which is proportional to fuel consumption when fuel type is constant. In this study, all vehicles considered used gasoline. The findings from our analysis of the data suggest that with the addition of running cost estimates, individuals’ WTP for more efficient vehicles can be significantly increased, in the case of this research by up to 28%.

Files