Bayesian "Now-Cast" Estimation of Event Probabilities in Multi-Party Democracies.
coalitions 
Overview
The coalitions package implements a Bayesian framework for the estimation of event probabilities in multi-party electoral systems (Bauer and others, 2019). To support estimation the package also implements scrappers that obtain data for German federal and general elections as well as Austrian general election. The implementation can be extended to support other elections.
To get started, see our workflow vignette
Check out our interactive shiny app on German (state and federal) elections/surveys
Updates are available from our KOALA_LMU twitter account!
Installation
# To install from CRAN use:
install.packages("coalitions")
# To install the most current version from GitHub use:
devtools::install_github("adibender/coalitions")
Usage
Detailed workflow is outlined in the workflow vignette.
A short overview is presented below.
Scrape surveys
The wrapper get_surveys() which takes no arguments, downloads all surveys currently available at wahlrecht and stores them in a nested tibble:
library(coalitions)
library(dplyr)
library(tidyr)
surveys <- get_surveys()
surveys
## # A tibble: 7 x 2
## pollster surveys
## <chr> <list>
## 1 allensbach <tibble [42 × 5]>
## 2 emnid <tibble [226 × 5]>
## 3 forsa <tibble [236 × 5]>
## 4 fgw <tibble [84 × 5]>
## 5 gms <tibble [96 × 5]>
## 6 infratest <tibble [110 × 5]>
## 7 insa <tibble [305 × 5]>
Each row represents a polling agency and each row in the surveys column again contains a nested tibble with survey results from different time-points:
surveys %>%
filter(pollster == "allensbach") %>%
unnest()
## # A tibble: 42 x 6
## pollster date start end respondents survey
## <chr> <date> <date> <date> <dbl> <list>
## 1 allensbach 2018-02-23 2018-02-02 2018-02-15 1271 <tibble [7 × 3…
## 2 allensbach 2018-01-25 2018-01-05 2018-01-18 1221 <tibble [7 × 3…
## 3 allensbach 2017-12-21 2017-12-01 2017-12-14 1443 <tibble [7 × 3…
## 4 allensbach 2017-11-30 2017-11-22 2017-11-27 1299 <tibble [7 × 3…
## 5 allensbach 2017-10-25 2017-10-07 2017-10-19 1454 <tibble [7 × 3…
## 6 allensbach 2017-09-22 2017-09-13 2017-09-20 1074 <tibble [7 × 3…
## 7 allensbach 2017-09-19 2017-09-06 2017-09-14 1083 <tibble [7 × 3…
## 8 allensbach 2017-09-06 2017-08-22 2017-08-31 1043 <tibble [7 × 3…
## 9 allensbach 2017-08-22 2017-08-04 2017-08-17 1421 <tibble [7 × 3…
## 10 allensbach 2017-07-18 2017-07-01 2017-07-12 1403 <tibble [7 × 3…
## # ... with 32 more rows
survey <- surveys %>% unnest() %>% slice(1)
survey %>% unnest()
## # A tibble: 7 x 8
## pollster date start end respondents party percent
## <chr> <date> <date> <date> <dbl> <chr> <dbl>
## 1 allensbach 2018-02-23 2018-02-02 2018-02-15 1271 cdu 32.0
## 2 allensbach 2018-02-23 2018-02-02 2018-02-15 1271 spd 17.5
## 3 allensbach 2018-02-23 2018-02-02 2018-02-15 1271 greens 12.0
## 4 allensbach 2018-02-23 2018-02-02 2018-02-15 1271 fdp 11.0
## 5 allensbach 2018-02-23 2018-02-02 2018-02-15 1271 left 9.50
## 6 allensbach 2018-02-23 2018-02-02 2018-02-15 1271 afd 13.0
## 7 allensbach 2018-02-23 2018-02-02 2018-02-15 1271 others 5.00
## # ... with 1 more variable: votes <dbl>
Calculate coalition probabilities
For each survey (row) we can calculate the coalition probabilities
survey %>% get_probabilities(nsim=1e4) %>% unnest()
## # A tibble: 6 x 4
## pollster date coalition probability
## <chr> <date> <chr> <dbl>
## 1 allensbach 2018-02-23 cdu 0
## 2 allensbach 2018-02-23 cdu_fdp 0.0500
## 3 allensbach 2018-02-23 cdu_fdp_greens 100.0
## 4 allensbach 2018-02-23 spd 0
## 5 allensbach 2018-02-23 left_spd 0
## 6 allensbach 2018-02-23 greens_left_spd 0
References
Bauer, Alexander, Andreas Bender, André Klima, and Helmut Küchenhoff. 2019. “KOALA: A New Paradigm for Election Coverage.” AStA Advances in Statistical Analysis, June. https://doi.org/10.1007/s10182-019-00352-6.
Bender, Andreas, and Alexander Bauer. 2018. “Coalitions: Coalition Probabilities in Multi-Party Democracies,” March. https://doi.org/10.21105/joss.00606.