governors: Election results and candidate longevity
In PPBDS/primer.data: Data for Preceptor's Primer for Bayesian Data Science

governors

R Documentation

Election results and candidate longevity

Description

This data is from the paper "Longevity Returns to Political Office" by Barfort, Klemmensen & Larsen (2019). The purpose of the study (pdf) was to explore how winning an election influences the lifespan of a candidate. The data set purports to includes all the candidates in U.S. gubernatorial elections after 1850 who were deceased as of 2012, but there are clearly serious data quality issues, especially in the 1800s.

Usage

governors

Format

A tibble with 3,587 observations and 13 variables:

state: character variable indicating the state in which an election took place
year: integer variable indicating the year in which an election took place
first_name: character variable indicating the first name of a candidate
last_name: character variable indicating the last name of a candidate
party: character variable indicating a candidate's party
sex: character variable indicating a candidate's sex
died: date variable indicating the candidate's date of death
status: character variable indicating whether a candidate was the challenger or incumbent
win_margin: double variable indicating the percentage margin by which the election was won (positive values) or lost (negative values)
region: character variable indicating region of the country: 'Midwest', 'Northeast', 'South', or 'West'
election_age: years a candidate lived before the election took place
death_age: years a candidate lived
lived_after: years a candidate lived after the election took place

Details

First, for a given election, only the two candidates who received the highest number of votes were included. Second, candidates with unknown dates of death were excluded, resulting in fewer observations for elections in recent years, since most recent candidates are still alive. Third, in a few instances, only the year of birth or death could be determined; in these cases, the date was taken to be July 1 of that year.