Korean National Assembly Data for Political Science Education

Provides ready-to-use datasets from the Korean National Assembly (assemblies 20 through 22, 2016-2026) for teaching quantitative methods in political science. Includes legislator metadata, bill proposals, roll call votes, asset declarations, and policy seminar records. Designed as a Korean politics counterpart to packages like 'palmerpenguins', enabling students to practice regression, panel data analysis, text analysis, and network analysis with real legislative data. Roll call vote data and spatial voting models are described in Poole and Rosenthal (1985) . Legislative data is sourced from the Korean National Assembly Open API.

assemblykor

Korean National Assembly data for political science education.

Documentation: https://kyusik-yang.github.io/assemblykor/

Overview

The goal of assemblykor is to provide a curated collection of Korean National Assembly datasets for teaching quantitative methods in political science. Think of it as a Korean politics counterpart to palmerpenguins.

The package includes seven built-in datasets covering legislators, bills, asset declarations, policy seminars, committee speeches, plenary votes, and roll call records, all drawn from public data of the Korean National Assembly (2004-2026).

Why tidyverse?

This package and all its tutorials follow tidyverse conventions. We use dplyr, ggplot2, tidyr, and the pipe operator (%>%) throughout.

Why tidyverse-first for teaching?

• Readability: tidyverse syntax reads like a sequence of data operations, reducing cognitive load for beginners (Cetinkaya-Rundel et al., 2022).
• Consistency: Packages share a common grammar and data structure, so learning one makes the next easier (Wickham et al., 2019).
• Doing things quickly: Students can perform meaningful data analysis from day one, rather than spending weeks on base R syntax (Robinson, 2017).
• Community standard: Most modern R textbooks, including R for Data Science, adopt tidyverse as the default, making it easier for students to find help and resources.

Note: This package is designed for educational purposes only. The datasets have been processed and curated for classroom use and may not reflect the most up-to-date or complete records. For research or policy analysis, please verify against the original data sources listed in the Data sources section below.

Data corrections in 0.1.4. legislators$seniority is now the seniority at each assembly (it was the lifetime number of terms), the committees of each assembly were rebuilt, vetoed bills record the result of the re-vote, speeches$member_id links to legislators, and get_proposers() no longer stops at 100 names per bill. These follow release 0.7.0 of kna. See NEWS.md for the full list and the size of each change.

Meet the data

Installation

# Install from CRAN
install.packages("assemblykor")

# Or install the development version from GitHub
# install.packages("remotes")
# remotes::install_github("kyusik-yang/assemblykor")

Usage

library(assemblykor)

data(legislators)
data(bills)
data(wealth)
data(seminars)
data(speeches)
data(votes)
data(roll_calls)

Example: party composition by gender

library(dplyr)

legislators %>%
  filter(assembly == 22) %>%
  count(party, gender) %>%
  filter(n >= 3)

Example: wealth distribution

library(ggplot2)

ggplot(wealth, aes(x = net_worth / 1e6)) +
  geom_histogram(bins = 50, fill = "steelblue") +
  labs(x = "Net worth (billion KRW)", y = "Count",
       title = "Distribution of legislator wealth")

Example: bill outcomes

bills %>%
  count(result, sort = TRUE) %>%
  head(5)
#>         result     n
#> 1 임기만료폐기 30679
#> 2 대안반영폐기 13692
#> 3         <NA> 12171
#> 4     수정가결  2213
#> 5     원안가결  1046

Korean font setup

ggplot2 plots with Korean text (axis labels, titles) may show broken characters. Run this once per session to fix it:

set_ko_font()
#> Korean font set: Apple SD Gothic Neo

This auto-detects a Korean font on your system (macOS, Windows, Linux). The interactive tutorials (run_tutorial()) handle this automatically.

Downloadable datasets

Larger datasets are available via download functions (requires the arrow package):

# Bill propose-reason texts (64,900 texts, ~26 MB download)
texts <- get_bill_texts()

# Co-sponsorship records (825,283 rows, ~3.8 MB download)
proposers <- get_proposers()

# Morpheme tokens for speeches (663,582 rows, ~1.3 MB download)
# Pre-tokenized with the Kiwi morphological analyzer - no Korean NLP
# setup needed on the student side
tokens <- get_speech_tokens()

Tutorials (한국어 수업 자료)

The package includes nine Korean-language tutorials designed for classroom use in political science methods courses:

Each tutorial is available in two formats:

Option A: Interactive browser (recommended for self-study)

# Launch an interactive tutorial with exercises in the browser
run_tutorial(1)  # or run_tutorial("01-tidyverse-basics")

Requires the learnr package (install.packages("learnr")). Students can type and run code directly in the browser with hints and solutions.

Option B: Plain R Markdown (for editing in RStudio)

# List available tutorials
list_tutorials()

# Copy an Rmd file to your working directory
open_tutorial(1)  # or open_tutorial("01-tidyverse-basics")

Students can edit and knit the Rmd file in RStudio at their own pace.

Joining datasets

All datasets share the member_id and/or assembly columns for easy joining:

# Merge legislators with wealth data
leg_wealth <- legislators %>%
  inner_join(wealth, by = "member_id", relationship = "many-to-many")

CSV access

CSV versions of the smaller datasets are available for teaching file I/O:

# Find the file path
path_to_file("legislators.csv")

# Read directly
legislators_csv <- read.csv(path_to_file("legislators.csv"), fileEncoding = "UTF-8")

Data sources

All data in this package are derived from publicly available sources.

Inspired by: palmerpenguins (Horst, Hill & Gorman, 2020), which demonstrates how domain-specific datasets can make methods teaching more engaging.

Disclaimer

This package is intended for educational use only (teaching quantitative methods in political science courses). The datasets have been processed, filtered, and in some cases sampled for classroom convenience. They should not be treated as authoritative records.

For research or policy analysis, please consult the original data sources listed above and verify the data independently.

Citation

citation("assemblykor")

Yang, Kyusik (2026). assemblykor: Korean National Assembly Data for
Political Science Education. R package version 0.1.4.
https://CRAN.R-project.org/package=assemblykor

License

MIT (package code). Data licenses as noted above.