Revisions: From Purchase to Participation: Sequential Choice Experiments on PEV Smart-Charging

Description

This dataset contains all materials required to replicate the analyses reported in Fowler, Van Passel, and Lizin (2026), resubmitted to Utilities Policy. The study uses two sequential discrete choice experiments (DCEs) embedded in a single online survey of 769 Belgian drivers, conducted in July–August 2024 in Dutch, French, and English. The first DCE elicits preferences for purchasing and installing a flexibility-capable residential EV charger; the second captures willingness to enroll in a retailer-managed smart charging service agreement. Preferences are analyzed using two-class latent class models estimated in R with the Apollo package, and respondent-level class probabilities are cross-classified across the two experiments to characterize the gap between hardware adoption and managed-charging participation. The repository includes: (1) the anonymized respondent-level dataset and long-format choice task files for both DCEs; (2) coded experimental design matrices for all five blocks of each DCE; (3) all R estimation scripts, including latent class models, the class membership function, WTP and delta-method calculations, the cross-classification under independence and comonotonicity assumptions, and the bootstrap procedure; (4) the complete survey instrument in Dutch, French, and English, including choice task instructions, hover-over tooltip text, instructional video mp4 files, all choice cards, demographic and attitudinal items, and the barrier checklist used in Appendix 4; and (5) ethical approval documentation. A README file describes the full replication workflow, file naming conventions, and software requirements (R 4.5.1, Apollo). The dataset is published under a CC-BY 4.0 license. Researchers may freely use, adapt, and redistribute the materials with appropriate attribution.

Steps to reproduce

Workflow Description 1. Data Collection • Qualtrics file included (Electric_car_charging_survey (1).qsf). • Raw survey responses downloaded from Qualtrics • Incomplete responses deleted in excel using the filter function (Usable_data_combined.csv) • The experimental design for the first DCE stored separately (File for merging_big_data_8choices.csv). • The experimental design for the second DCE stored separately (File4merging_4 choices_ex2.csv) 2. Data Transformation • First DCE o The two sources (Usable_data_combined.csv; File for merging_big_data_8choices.csv) merged using the script (Data cleaning and merger ex 1.R) o Script outputs a long-format file: merged_file.csv that contains all choice tasks, alternatives, and respondent data. • Second DCE o Two sources (Usable_data_combined.csv; File4merging_4 choices_ex2.csv) merged using (Data cleaning and merger ex 2.R). o This script outputs a long-format file (merged_data_2.csv) that contains all choice tasks, alternatives, and respondent data. 3. Data Cleaning • Basic filters and exclusions described in manuscript were applied in Excel using the filter function (file: merged_data.csv) based on the quality criteria described in Section 5.1 of the manuscript: o Nonvalid responses coded as ex1_junk=1, reflected in the text of the manuscript o Filtered sample demographic breakdown: Table 3: Panel A  Panel B data found using data filers in Microsoft Excel 4. Model Estimation (R / Apollo) • We use the R to estimate each model described in the paper: o LC_price_reward_conditional.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. o LC_WTP_EX2_conditional.R: Table 5 and Table 6  This creates merged_data_with_class_probs_both_DCEs_conditional.csv, which is needed for the next steps. o bootstrap_probabilities.R: creates data for Figure 3: Table 7 Pannel B o minimum_categorical latent class model.R: Table 1A o LC_WTP_EX2_region.R: 3A o opt_out_reasons.R: Table 4A o EV_driver_interactions.R: Footnote 11 o restricted model comparison_opt_out_first DCE.R: Footnote 13. o LC_WTP_EX2_include protests.R and LC_price_reward_conditional_with_protest.R: Footnote 7 o dominance_check_revise.R: Appendix 2 o DCE1_3_classes.R and DCE2_3_classes.R: Footnote 6 • These scripts produce log-likelihoods, coefficient estimates, WTP figures, and joint probabilities in R output. 5. Tables and Figures • Tables 1 and 2 copied from an excel spreadsheet and formatted in Word • The numerical output from the R models copied into Microsoft Word and formatted manually for Tables 3–7 as well as all tables in the appendices, except for Table 1A, which is already mentioned. • Figures 1 constructed in Power Point: • R: version 4.5.1 • Other software: o Microsoft Excel (for data filtering) o Microsoft Word (for tables) Microsoft Power Point (for figure 1)

Institutions

• Hasselt University

  Flanders, Hasselt
• University of Antwerp

  Flanders, Antwerp

Categories

Funders

• Energy Transition Fund (Belgium)

Licence