Skin Color and Educational Attainment in 18 Latin American Countries, 2016-2017

Javier Castillo's research dataset analyzes the causal effect of interviewer-rated skin color on years of schooling across 18 Latin American countries. It contains data from 23,163 respondents in the 2016–2017 AmericasBarometer surveys, analyzed using a Double/Debiased Machine Learning design. The dataset was last updated on 2026-05-19.

Use Cases

• Estimating causal effects of skin color on educational outcomes based on the described counterfactual framework.
• Analyzing cross-national variation in colorism effects based on the five-category typology of country-level ethnoracial configurations.
• Studying the impact of interviewer rating heterogeneity on social science measurements based on the use of interviewer fixed effects.
• Replicating or extending the partially linear Double/Debiased Machine Learning analysis described in the methodology.

Strengths

• Data is nationally representative, covering 18 Latin American countries.
• Includes a substantial sample size of 23,163 survey respondents.
• Implements a causal inference design (Double/Debiased Machine Learning) to address confounding.
• Explicitly models and adjusts for interviewer-specific rating heterogeneity.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the underlying analysis dataset is unknown, which may limit suitability assessment.
• The primary file is a 27.9 KB DOCX document, suggesting the dataset's scope is limited to analysis outputs or a codebook rather than raw data.

Provenance

Source

AmericasBarometer 2016–2017 surveys.

Collection Method

Nationally representative surveys analyzed with a Double/Debiased Machine Learning design.

Time Range

2016-2017

Freshness

Last updated 2026-05-19 04:19:06; freshness should be verified.

Geography

18 Latin American countries