ei_rxc: EI Bayesian simultaneous estimation for multiple races and candidates

Description

EI Bayesian simultaneous estimation for multiple races and candidates

Usage

ei_rxc(
  data,
  cand_cols,
  race_cols,
  totals_col,
  name = "",
  ntunes = 10,
  totaldraws = 10000,
  samples = 1e+05,
  thin = 1,
  burnin = 10000,
  ci_size = 0.95,
  seed = NULL,
  eiCompare_class = TRUE,
  ret_mcmc = FALSE,
  verbose = FALSE,
  diagnostic = FALSE,
  n_chains = 3,
  plot_path = NULL,
  par_compute = FALSE,
  n_cores = NULL,
  ...
)

Value

If ret_mcmc == TRUE, a list is returned containing results and a data frame of the full chains from the MCMC. If ret_mcmc == FALSE, results are returned in a dataframe

A dataframe of ei results

Arguments

data: A data.frame() object containing precinct-level turnout data by race and candidate
cand_cols: A character vector listing the column names for turnout for each candidate
race_cols: A character vector listing the column names for turnout by race
totals_col: The name of the column containing total votes cast in each precinct
name: A unique identifier for the outputted eiCompare object.
ntunes: Integer number of pre-MCMC tuning runs, defaulted to 10
totaldraws: Integer number of iterations per run in pre-MCMC tuning runs, defaulted to 10000
samples: Integer number of draws saved and used to compute estimates. Total chain length is sample*thin + burnin
thin: Integer specifying the thinning interval for posterior draws. Eg. if thin = 2, every second draw gets added to the sample
burnin: Integery specifying the number of initial iterations to be discarded, defaulted to 10000
ci_size: Numeric desired probability within the upper and lower credible-interval bounds, defaulted to 0.95
seed: A numeric seed value for replicating estimate results across runs. If NULL, a random seed is chosen. Defaulted to NULL.
eiCompare_class: default = TRUE
ret_mcmc: Boolean. If true, the full sample chains are returned
verbose: A boolean indicating whether to print out status messages.
diagnostic: Boolean. If true, run diagnostic test to assess viability of MCMC parameters (will return all chain results)
n_chains: Number of chains for diagnostic test. Default is set to 3.
plot_path: A string to specify plot save location. If NULL, plot is not saved.
par_compute: Boolean. If true, diagnostic test will be run in parallel.
n_cores: The number of cores to use in parallel computation. Defaulted to NULL, in which case parallel::detectCores() - 1 is used
...: Additional parameters passed to eiPack::tuneMD()

Author

Loren Collingwood <loren.collingwood@ucr.edu>, <loren.collingwood@gmail.com>

Hikari Murayama <hikari_murayama@berkeley.edu>

Ari Decter-Frain <agd75@cornell.edu>

References

eiPack, King et al., (http://gking.harvard.edu/eiR)