educabr: Harmonised Brazilian Education Series • educabr

Harmonised historical series on Brazilian education — enrollment and educational attainment, compiled and reconciled across decades of heterogeneous official and academic sources.

educabr provides curated long-run series on Brazilian formal education — enrollment by stage and network, years of schooling, literacy and attainment — compiled and reconciled from heterogeneous official and academic sources (Censo Escolar, PNAD, Censo Demográfico, Anuário Estatístico do IBGE, INEP CENSUP, Kang/FGV-IBRE, Walter & Kang, among others) into a single tidy schema with explicit per-row provenance.

🇧🇷 educabr reúne séries históricas tratadas sobre educação formal no Brasil — matrículas, anos de estudo, atingimento educacional — em um único schema tidy, com proveniência explícita. Veja a vinheta em português.

Quick links

📖 Reference site — function reference, articles, news
📊 Live dashboard — interactive multi-source comparison on shinyapps.io
📝 Get started — 10-minute tour of the API
🐛 Issues

Installation

educabr is not yet on CRAN. Install the development version from GitHub:

# install.packages("remotes")
remotes::install_github("mancano-tales/educabr")

Usage

The package exposes three top-level functions plus the bundled dashboard launcher.

`get_enrollment()` — school enrollment

Who is enrolled at a given stage in a given year. Annual flow, school perspective.

library(educabr)

# Gross enrollment rate for ensino fundamental, Brazil, all years
get_enrollment(
  level     = "fundamental",
  indicator = "rate",
  geo_level = "BR"
)

# Tertiary enrollment by network, all sources kept side by side
get_enrollment(
  level    = "superior",
  network  = c("publica", "privada"),
  modality = "total",
  year     = c(1990, 2024)
)

# Enrollment rates by race/colour, 1960 onward
get_enrollment(
  level     = "fundamental",
  indicator = "rate",
  dimension = "race",
  year      = c(1960, 2010)
)

`get_schooling()` — educational attainment

How much the population has already studied. Accumulated stock, person perspective.

# Mean years of schooling for adults, Brazil
get_schooling(geo_level = "BR")

# Same series broken down by race and by sex
get_schooling(dimension = "race")
get_schooling(dimension = "sex")

`run_dashboard()` — interactive Shiny app

educabr::run_dashboard()

Three navbar tabs:

Enrollment — Kang/FGV series by stage, year, race
Tertiary Education — multi-source comparison 1907–2024 (Kang, Maduro Jr., IBGE Século XX, INEP CENSUP), with optional reconstructed totals for the 2000–2008 in-person + EAD split
Educational Attainment — Walter & Kang mean years of schooling by BR / region / UF, with race and sex breakdowns

Every chart has a “View R code” button that prints a self-contained snippet (educabr + ggplot2 + plotly) you can paste into RStudio to reproduce the interactive chart locally.

Datasets

Three internal datasets back the public functions:

Dataset	Rows	Coverage	Source(s)
`enrollment_kang_fgv`	6,238	BR + UF, EF1/EF2/EF/EM/ES, 1933–2010 (+ race breakdown 1960–2010)	Kang, Paese & Felix (2021); Kang & Menetrier (2024); Kang, Menetrier & Comim (2024)
`enrollment_tertiary`	1,341	BR, ensino superior, 1907–2024, by network/institution/modality	IBGE Século XX; Maduro Jr. (2007); Kang/Paese/Felix (2021); INEP CENSUP Synopsis / Microdata / Power BI
`schooling_kang_fgv`	2,287	BR + region + UF, mean years of schooling, 1925–2015, with race and sex breakdowns	Walter & Kang (2024)

End users should call get_enrollment() / get_schooling(). The datasets are exposed for inspection but the public API normalises schema differences, applies per-source filters and translates labels.

Schema

All get_*() functions return a tibble in the canonical tidy-long schema documented in inst/dict/schema.yaml: one row per observation, alternative sources for the same indicator as separate rows (column source), aggregations as an explicit factor level ("total") rather than NA. Controlled vocabularies live in inst/dict/vocabularies/, and PT-BR labels for every factor level in inst/dict/i18n.yaml.

educabr is complementary to educabR (Sidney Bissoli), which organises access by official source (get_ideb(), get_enem(), get_censo_escolar(), …). Here the axis is by theme and historical series: single indicators compiled across multiple sources over long time spans.

Design inspirations: geobr (coherent function family), PNADCperiods (embedded dashboard + methodological delivery), and the brverse ecosystem.

Citation

Cite both the package (for the harmonisation work) and the originating sources for any data you actually use. The educabr_cite() helper builds APA / BibTeX entries for every bundled source:

citation("educabr")                              # cite the package

educabr_cite("kang_paese_felix_2021")            # cite one source
educabr_cite(c("walter_kang_2023",               # cite many
              "inep_microdados_censup"))
educabr_cite()                                   # all bundled sources

# Typical workflow — query first, then cite only what you used
d   <- get_enrollment(level = "fundamental", indicator = "rate")
educabr_cite(unique(d$source), style = "text")

A Zenodo DOI will accompany the first tagged release.

Contributing

Issues and PRs welcome. To contribute new data, open an issue first describing the source file, author, geographic/temporal coverage and licence — before submitting a PR.

License

Code under GPL (>= 3). Data redistributed under CC BY 4.0 (except where the original source imposes restrictions; see inst/dict/vocabularies/sources.yaml). See LICENSE.md.

educabr