Community-cat caregivers in Spain (ONECAT-SOC, 2026): de-identified analytic dataset and reproducibility code
Description
This deposit is the reproducibility package for Paper 1 of the ONECAT-SOC study, a cross-sectional, anonymous online survey of people who care for or help manage free-roaming community-cat colonies across Spain (fieldwork 10 July–5 August 2026; final n = 1,066 adults). It contains a de-identified, survey-derived analytic dataset (one row per respondent) and the analysis code needed to reproduce the study's focal results. The dataset provides the variables used in the article: a standardized care-intensity (workload) index and its inputs (total cats attended, number of colonies, breadth of tasks, days per week); perceived-burden and perceived-institutional-support scales (support scored requiring at least three of four valid items, with a permissive one-item variant); a serious conflict/victimization indicator and its components (threats/harassment, complaints/police, removal or disappearance of cats, vandalism, poisoning or abuse, declared or suspected); organizational and investment measures (solitary versus shared care, association arrangement, weekly hours, monthly personal expenditure); institutional-arrangement, support and recognition descriptives (accreditation, direct and administrative supports with mutually exclusive "no support" coding, responsibility attribution, official and social recognition); sociodemographic covariates (age, education, insularity, urbanization); and a pseudonymous municipality-level cluster identifier used for cluster-robust inference (province-level fallback clusters where municipality could not be derived). A Python script (ONECAT_P1_reproduce.py) reads the CSV and reproduces the focal associations reported in the article's Table 2 and the main institutional-support result. A codebook (variable dictionary) and a README with methods summary and instructions are included. Estimates are associations, not causal effects; the sample is self-selected (non-probability), so it cannot be treated as representative of Spain's caregiver population. The study received a favorable opinion from the Research Ethics Committee for Human Research of the University of Las Palmas de Gran Canaria (CEIH-2026-36, 9 July 2026); participation was voluntary and followed informed consent. This deposit is de-identified: it contains no names, contact details, postal codes, municipality/province/island names, free-text responses, or timestamps; territory is represented only by a binary insularity indicator, an ordinal urbanization band, and a pseudonymous cluster id. Raw individual responses are not shared. A companion data article describes the fuller de-identified survey dataset. License: CC BY 4.0.
Files
Steps to reproduce
Requirements: Python 3.10+ with pandas, numpy and statsmodels (pip install pandas numpy statsmodels). 1. Place all files in one folder: ONECAT_P1_derived_analytic_data.csv, ONECAT_P1_codebook.csv and ONECAT_P1_reproduce.py. 2. From that folder run: python ONECAT_P1_reproduce.py The script loads the de-identified derived dataset (n = 1,066), keeps rows with a cluster id, and fits the article's focal models: ordinary least squares for perceived burden and perceived institutional support, and modified Poisson (log link, robust variance) for serious conflict/victimization, each adjusted for age, education, insularity and urbanization, with cluster-robust standard errors by the pseudonymous municipality cluster (muni_cluster). It also fits the solitary-care main effect on burden and the workload x solitary interaction on conflict. Expected output (associations per +1 SD of the care-intensity index): - Perceived burden (OLS, n = 918): b = +0.406 [0.323, 0.490], p < .001 - Institutional support, >=3 of 4 items (OLS, n = 864): b = +0.092 [0.002, 0.181], p = .044 - Serious conflict/victimization (modified Poisson, n = 919): PR = 1.052 [1.016, 1.089], p = .004 - Solitary care -> burden: b = +0.132 [0.014, 0.250], p = .028 - Workload x solitary interaction (conflict): PR = 1.127 [1.044, 1.217], p = .002 Variable construction is already applied in the CSV: the workload index is the first principal component of four standardized indicators (log total cats, colonies, task breadth, days per week); scales are item means (support requires >=3 of 4 valid items); the conflict indicator is any of five severe categories; and the "no support" categories of the D20 and E27 multiple-response blocks were recoded to be mutually exclusive of positive options. Estimates are associations, not causal effects. Full item wording, coding and missing-data handling are in the article's Supplementary Methods and in ONECAT_P1_codebook.csv.
Institutions
- Universidad de Las Palmas de Gran CanariaCanary Islands, Las Palmas de Gran Canaria
- Instituto de Productos Naturales y AgrobiologíaCanary Islands, San Cristóbal de La Laguna