Replication Data and Code for “From Purchase to Participation: Sequential Choice Experiments on PEV Smart-Charging”

Description

This replication package contains the data, code, and survey materials used to produce the results in the manuscript submitted to Utilities Policy. The analysis is based on a discrete choice experiment (DCE) conducted with 769 valid responses in 2024.

Steps to reproduce

Workflow Description 1. Data Collection o Data collected in Qualtrics from this survey.  In addition to the above link, the Qualtrics file is included (Electric_car_charging_survey (1).qsf). o Raw survey responses were downloaded from Qualtrics o Incomplete responses were deleted in excel using the filter function (Usable_data_combined.csv) o The experimental design for the first DCE is stored separately in (File for merging_big_data_8choices.csv). o The experimental design for the second DCE is stored separately in (File4merging_4 choices_ex2.csv) 2. Data Transformation o First DCE  The two sources (Usable_data_combined.csv; File for merging_big_data_8choices.csv) were merged using the script (Data cleaning and merger ex 1).  This script outputs a long-format file called merged_file.csv that contains all choice tasks, alternatives, and respondent data. o Second DCE  The two sources (Usable_data_combined.csv; File4merging_4 choices_ex2.csv) were merged using the script (Data cleaning and merger ex 2).  This script outputs a long-format file called merged_data_2.csv that contains all choice tasks, alternatives, and respondent data. 3. Data Cleaning o Basic filters and exclusions applied in Excel using the filter function (file: merged_data.csv) based on the quality criteria described in Section 5.1 of the manuscript:  Illogical choice combinations (e.g. day/night price and single meter)  Extreme/outlier values (e.g. age > 100)  Straight-line responses  Protest responses, ect o Nonvalid responses coded as ex1_junk=1, reflected in the text of the manuscript o Filtered sample demographic breakdown.R: Table 3 4. Model Estimation (R / Apollo) o We use the R to estimate each model described in the paper:  LC_price_reward.R : Table 4 – latent class probabilities with merged_data_2.csv to create merged_data_with_class_probabilities.csv, which is needed for the next step.  LC_WTP_EX2.R: Table 5 and Table 6 – merged_data_with_class_probs_both_DCEs.csv, which is needed for the next steps.  Figure 3 data.R: creates data for Figure 3: Figure 3 matrix.R  minimum_categorical latent class model.R: Table 1A LC region.R: Table 2A LC_WTP_EX2_region.R: 3A  opt_out_reasons.R: Table 4A EV_driver_interactions.R: Footnote 4 restricted model comparison_opt_out_first DCE.R: Footnote 5. o These scripts produce log-likelihoods, coefficient estimates, WTP figures, and joint probabilities in R output. 5. Other Tables and Figures o Tables 1 and 2 copied from an excel spreadsheet and formatted in Word o The numerical output from the R models copied into Microsoft Word and formatted manually for Tables 3–5 as well as all tables in the appendices, except for Table 1A, which is already mentioned. o Figures 1 and 2 screenshot from the actual survey, with a tooltip (made in Publisher) and cursor imposed on it

Institutions

• University of Antwerp

  Flanders, Antwerp

Categories

Funders

• Energy Transition Fund (Belgium)

Licence