Data and code for "Can Public Procurement Promote Organic Farming? Evidence From a Cross-Country Factorial Survey Experiment"

Le résumé fourni par la source

Data and replication code for a cross-country factorial survey experiment (FSE) on stakeholder preferences for organic public food procurement policy packages. Stakeholders involved in public food procurement and public catering in Czechia, Germany, Hungary, Poland, and the United Kingdom each evaluated eight randomized procurement policy scenarios. Every scenario varied four policy attributes: * minimum organic content requirement: none, 10%, 40% or 70%;* local/regional sourcing criterion: required or not required;* support for organic product use: none, advisory services, network and value-chain development, or a comprehensive program combining both;* cost management: market-priced organic costs, or public funding covering the organic surcharge. Respondents rated each scenario on three 0–10 scales: perceived fairness, perceived effectiveness in promoting organic farming, and perceived political feasibility. The design is a full factorial of 4 × 2 × 4 × 2 = 64 scenarios, divided into eight balanced blocks of eight. Each respondent was randomly assigned to one block and saw its scenarios in randomized order, yielding a panel structure of evaluations nested within respondents. Fieldwork ran from November 2024 to July 2025 on the SurveyEngine platform. The questionnaire was developed in English and translated and adapted by native experts in each country, with several rounds of qualitative pretesting. The analysis sample comprises 1,281 respondents and 10,248 vignette evaluations (Poland: 552, Czechia: 484, United Kingdom: 87, Hungary: 79, Germany: 79). This is a targeted stakeholder sample rather than a population sample, and recruitment strategies differed across countries. Contents: Survey data as Excel workbooks, one pair per country: a vignette file recording which of the 64 designed scenarios each respondent saw, and a covariate file with consent, screening, socio-demographics, knowledge checks, all 24 scenario ratings and debriefing items. Each workbook carries the SurveyEngine data dictionary as a second sheet, in the language fielded in that country. The English master questionnaire is included as a Word document: welcome and consent screens, screening and stakeholder questions, the explanation of public procurement and organic farming shown to respondents, the attribute and level table used to build the vignettes, an example vignette, the three evaluation criteria, and the socio-demographic, knowledge-check and debriefing blocks. Three R scripts reproduce the published results: the weighted multivariate ordered probit models on the full sample (pooled, country-specific, and the SUR-style Gaussian robustness models), the same suite on the restricted sample of respondents closest to procurement who passed both knowledge checks, and the post-estimation code for the average discrete changes in the paper's main figure. Data protection: The technical identifiers (IP address, session id, invitation source and id, original survey URL) and all verbatim open-text answers have been removed, as have the records of respondents who declined consent. Estimation uses a hierarchical Bayesian framework (Hamiltonian Monte Carlo via Stan and the `brms` package), with respondent-level random intercepts correlated across the three outcomes. Contents: ├── FSE questionnaire.docx English master survey instrument│├── visionary_fse_organic_polish_fse_8_cards_full_table.xlsx├── visionary_fse_organic_polish_covariates (4).xlsx├── visionary_fse_organic_english_fse_8_cards_full_table (1).xlsx├── visionary_fse_organic_english_covariates (7).xlsx├── visionary_fse_organic_german_fse_8_cards_full_table.xlsx├── visionary_fse_organic_german_covariates (2).xlsx├── visionary_fse_organic_hungarian_fse_8_cards_full_table (2).xlsx├── visionary_fse_organic_hungarian_covariates (7).xlsx├── visionary_fse_organic_czech_fse_8_cards_full_table (1).xlsx├── visionary_fse_organic_czech_covariates (3).xlsx│├── FSE_main_3_full.R main models, full sample → outputs0/├── FSE_main_3_restricted.R robustness, restricted sample → outputs1/└── Average discrete changes.R Figure 3 post-estimation → outputs_figures/ There are two data files per country: a vignette file (`*_fse_8_cards_full_table*.xlsx`) recording which of the 64 designed scenarios each respondent saw, and a covariate file (`*_covariates*.xlsx`) holding everything else, including the ratings. Both have two sheets: `data` (the records) and `dictionary` (SurveyEngine variable and value labels in the language of that country's questionnaire). The dictionary sheet is authoritative for question wording. The file-name suffixes `(1)`, `(2)`, `(3)`, `(4)`, `(7)` are download artefacts, not version numbers. The R scripts open the files by these exact names in the `country_specs` table at the top of each script, so any renaming has to be mirrored there, and the data files must sit in the R working directory. The authors acknowledge funding from the European Union under Grant Agreement no. 101060538(VISIONARY project). The work of UK participant was funded by UK Research and Innovation (UKRI)under the UK government’s Horizon Europe funding guarantee grant numbers 10037976. Views andopinions expressed are, however, those of the authors only and do not necessarily reflect those of theEuropean Union or REA. Neither the European Union nor the granting authority can be held responsible forthem.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.