• http://dx.doi.org/10.18235/0002023

Inter-American Development Bank (2023). Replication data for: SPOON: Continuous Program to Improve Nutrition: Baseline Survey Results in Baja Verapaz, Guatemala. IDB Open Data. https://doi.org/10.60966/p7wx1c1d

1. English

What is this dataset?

The SPOON dataset focused on Baseline Surveys in Baja Verapaz, Guatemala is a compilation of baseline resources from a child and caregiver nutrition intervention in Guatemala. They bring together household, caregiver, and observations survey modules, randomized treatment assignment files, and anthropometric indicators (for example, _zbmi, _zlen, HFA 2–3 SD variants).

What questions can researchers answer with these datasets?

• Do nutrition counseling and related program components improve child growth (length/height-for-age, BMI-for-age) and care practices?
• Which caregiver behaviors (feeding practices, knowledge, food preparation) mediate impacts on anthropometric outcomes?
• How do household characteristics (assets, shocks, composition) and caregiver traits (for example, grit, Rosenberg self-esteem, decision-making power) shape program effectiveness?

What files and modules are included?

• Survey modules: caregiver_all_el/, household_el/, observations_el/, knowledge_el/, practice_el/
• Anthropometrics: hfa_2sd/, hfa_3sd/, bmi_2sd/, _zlen/, _zbmi/
• Assignment & analysis assets: 3-treatment_status/, new_household_assignments/, 7-analysis_merge/, 2-baseline_prep/
• Documentation & tables: tables3-5/, tablesS4-S11/, tableS4/, instrument files (for example, 20180823_Instrumento SPOON Guatemala)
• dataset_metadata.csv summarizing all resources

What do the anthropometric files contain?

Standardized WHO-style indicators: - Length/height-for-age (for stunting risk): hfa_2sd/, hfa_3sd/, _zlen/ - BMI-for-age (for thinness/overnutrition risk): bmi_2sd/, _zbmi/ These are typically provided as z-scores and binary cutoffs (±2SD or ±3SD) suitable for regression or prevalence estimates.

What caregiver and household variables are available?

• Caregiver: demographics, education, psycho-social indices (grit, Rosenberg self-esteem), decision-making, time use, knowledge, and practices.
• Household: composition, assets, shocks, food security, WASH, and socio-economic markers.
• Observations: enumerator-recorded conditions of the home and environment to validate self-reports.

How is treatment defined in the SPOON datasets?

Randomized (or clustered) treatment status is provided in 3-treatment_status/ and new_household_assignments/, enabling intention-to-treat and treatment-on-the-treated analyses. The 7-analysis_merge/ resources help replicate analytic merges used in tables (tables3-5/, tablesS4-S11/).

How do I merge the datasets?

Use the assignment files as the nucleus, then merge household/caregiver/observations and anthropometric indicators by the shared IDs documented in the resource dictionaries or merge scripts (for example, 7-analysis_merge/). Always confirm key names and types before joining.

What are standard analyses people run on SPOON datasets?

• ITT / DiD models using baseline treatment assignment and child outcomes (for example, _zlen, _zbmi, hfa_*).
• Mediation by caregiver knowledge/practices and household features (WASH, food security).
• Heterogeneity by child age/sex, caregiver education, and household poverty markers.
• Robustness with alternative SD cutoffs (2SD vs 3SD) and different clustering levels.

What should I know about timing and instruments?

Instrument files (for example, 20180823_Instrumento SPOON Guatemala) describe modules and question flow. Use them to align survey timing with anthropometrics and to interpret skip patterns, constructed indices, and composite scores.

What are the dataset limitations?

• Baseline focus: without later rounds, causal claims rely on randomized assignment and cross-sectional contrasts; check documentation for follow-up availability.
• Measurement error: anthropometrics require careful QC; re-check outliers flagged via 2–3 SD thresholds.
• Context specificity: results are most applicable to communities/settings represented in the Guatemala sample.

What formats are provided?

Primarily CSV tables plus documentation (instruments, code/tables directories) and a dataset metadata CSV cataloging resources, formats, sizes, and access URLs.

How do these relate to other SPOON resources?

These baseline spoon datasets align with SPOON program publications and technical reports. Pair the assignment files with anthropometrics and survey modules to reproduce summary tables (for example, tables3-5/, tablesS4-S11/).

Any tips for reproducible workflows?

• Start from dataset_metadata.csv to inventory resources.
• Standardize ID keys; keep a single source of truth for merges (for example, 7-analysis_merge/ scripts).
• Log data cleaning (winsorization/outlier rules) for _zbmi, _zlen, hfa_*.
• Pre-register model specs (outcomes, covariates, clusters) to keep the analysis auditable.