# CREATE Survey Dataset: Repository Guide

## Dataset overview

- **Data file:** `CREATE_DATA_CSV ZENODOCLEAN.csv`
- **Unit of observation:** one anonymous adult survey respondent
- **Records:** 366
- **Variables:** 161
- **Geographic scope:** Puerto Rico
- **Format:** wide CSV, one row per respondent
- **Language:** Spanish response labels; most variable names are abbreviated English labels

The dataset contains original survey responses and project-specific recodes and indices related to daily activities, primary transportation mode, travel time, roadway safety, vehicle access, concern about natural hazards, community hazard experience, mobility impacts, roadway recovery time, detour burden, emergency transportation options, Blue Economy activities, roadway access quality, coastal classification, household size, and vehicle availability.

## Important use notes

- **Recommended workflow:** Import the CSV using UTF-8 encoding; preserve column names exactly; inspect blanks and “No se” separately; use binary/ordinal recodes only when they match the intended research question.
- **Geographic comparison:** MUN_COAST is municipality-level. BARRIO_COAST is neighborhood-level and is the more specific exposure classification used in the natural-hazard vulnerability analyses. Do not treat both variables as interchangeable.
- **Sampling limitation:** The survey used convenience sampling. Results describe participants and should not be presented as statistically representative of Puerto Rico without appropriate weighting and design information.
- **Conditional questions:** Some travel-time, access, safety, and quality questions were shown only when a relevant transportation mode was selected. Blanks may therefore mean “not applicable,” not refusal.
- **Top-coded categories:** For calculations using HH SIZE or NUMBER VEHICLES, the value 5 represents “five or more.” Means calculated directly from these variables are conservative lower-bound estimates.
- **Multiple-response field:** MOBILITY_OPTION_DISASTER contains semicolon-delimited choices, including occasional free-text responses. Prefer the separate option indicators for standard quantitative analysis.
- **Unknown categories:** “No se” is Spanish for “Do not know.” Retain it as a separate substantive response in descriptive work; the ordinal roadway score columns leave it blank.
- **Indices:** Composite indices summarize several items but do not replace item-level analysis. Review item completeness and internal consistency before inferential modeling.
- **Small cells:** Several modes and geographic groups have small sample sizes. Suppress or combine very small cells when reporting results to protect confidentiality and avoid unstable percentages.
- **Quality control:** The public data export should be accompanied by the survey instrument, this codebook, the Year 1 report, and a version/date statement.

## Files recommended for the repository

1. The de-identified CSV data file.
2. `CREATE_Survey_Data_Codebook.xlsx`, containing the complete 161-variable dictionary and coding schemes.
3. The survey instrument or questionnaire.
4. The CREATE Year 1 report.
5. A data availability statement and final Zenodo citation/DOI.

## Confidentiality

The file contains anonymous IDs and no direct personal identifiers. Users should not attempt to identify respondents and should avoid publishing very small geographic or demographic cells.

## Encoding

Open the CSV as UTF-8. The current export contains some corrupted accented characters in Spanish labels, such as `Veh.culo`; these are display artifacts in the export and do not change the numerical recodes.