===============================================================================
DATA & CODE DEPOSIT — README
Flood risk perception and behavioural responses in conditions of repeated
floods: evidence from Bagnacavallo (Ravenna, Italy)
===============================================================================

This folder contains the survey data, documentation, and analysis code
supporting the manuscript. The survey (N = 100) was administered as a
complement to three Focus Group Discussions and served an exploratory role
rather than aiming for statistical representativeness.

-------------------------------------------------------------------------------
CONTENTS
-------------------------------------------------------------------------------

1. survey_raw_anonymised.csv
   The raw survey export as collected (Google Forms output), with the original
   Italian question texts as column headers. Direct personal identifiers have
   been removed (see ANONYMISATION below) and replaced by a respondent_id.
   This is the untransformed source data. Response values are kept in their
   original Italian (e.g. "Sì"/"No", "Per nulla"/"Poco"/"Abbastanza"/"Molto")
   because they are the raw responses; the codebook translates them.

2. survey_analysis_anonymised.csv
   The cleaned analysis dataset. Same 100 respondents, but with short English
   variable names and all derived/recoded variables added (numeric Likert
   recodes, the flood-damage dummy, mean willingness index, age bands, etc.).
   Response values remain in their original Italian; the codebook gives their
   English meaning.

3. survey_data.dta
   Identical content to file 2, in Stata format. This is the cleaned dataset
   produced by survey_cleaning.do and read by survey_figures.do.

4. survey_cleaning.do
   Stata script — DATA CLEANING. Imports survey_raw_anonymised.csv, renames
   variables to English (by column position), converts types, recodes the
   Likert items, builds the derived variables (flood-damage dummy, mean
   willingness index, age bands, ordered risk-attribution recode), and
   standardises the free-text municipality names. It saves the cleaned dataset
   as survey_data.dta and survey_analysis_anonymised.csv. Comments are in
   English. Runs as-is with no path editing required, provided
   survey_raw_anonymised.csv is in the working directory.

5. survey_figures.do
   Stata script — FIGURES. Reads survey_data.dta and produces the five figures
   reported in the manuscript:
     - Figure 5: distribution of perceived flood predictability
     - Figure 6: distribution of perceived flood impact severity
     - Figure 7: perceived vulnerability to future floods (who is at risk)
     - Figure 8: individual willingness to change habits, by behavioural domain
     - Figure 9: individual vs perceived community willingness
                 (pluralistic-ignorance gap, by flood-damage group)
   All percentages and means are computed directly from the data. Comments are
   in English.

6. Codebook_Bagnacavallo.docx
   Full data documentation: definition of every variable, coding scheme for
   each (Likert scales, dummies, categorical recodes), the meaning of the
   Italian response values in English, and the ordered list of all
   transformations applied during processing. Read this alongside the .do files.

-------------------------------------------------------------------------------
HOW TO REPRODUCE
-------------------------------------------------------------------------------

1. Place all files in the same folder and set it as the Stata working
   directory (cd "path/to/folder").
2. Run survey_cleaning.do  -> creates survey_data.dta and
   survey_analysis_anonymised.csv from the raw file.
3. Run survey_figures.do   -> creates fig5-fig9 (.png).

The scripts have been checked to reproduce the figures reported in the
manuscript, including the pluralistic-ignorance gap (no-damage group
3.01 vs 2.00; damaged group 2.82 vs 1.98).

-------------------------------------------------------------------------------
ANONYMISATION
-------------------------------------------------------------------------------

The original working files collected direct personal identifiers for optional
recontact consent only: respondent name and surname, email address, and
telephone number, plus a submission timestamp. ALL of these have been removed
from every file in this deposit and replaced by an anonymous respondent_id
(integers 1-100). No cell in the shared files permits re-identification. This
is required for GDPR compliance and is consistent with the informed-consent
terms under which participants took part. The original identifiable files are
NOT included in this deposit and should not be shared.

-------------------------------------------------------------------------------
CONTACT
-------------------------------------------------------------------------------

For questions about the data or code, please contact the corresponding author.
