ei_est_gen: Iterative EI Estimation

In eiCompare: Compares Different Ecological Inference Methods

Iterative EI Estimation

Description

Iteratively fits EI models for candidates and racial/ethnic groups

Usage

ei_est_gen(
  cand_vector,
  race_group,
  total,
  rho = 10,
  data,
  table_names,
  sample = 1000,
  tomog = FALSE,
  density_plot = FALSE,
  plot_path = NULL,
  beta_yes = FALSE,
  seed = NULL,
  ...
)

Arguments

cand_vector	Character vector of candidate names, taken from the dataset
race_group	Character vector of formula, e.g., "~ pct_latino"
total	Character vector (e.g., "totvote") of total variable name from data, variable in data is numeric
rho	Rho parameter for ei() estimate, defaults to 10, numeric
data	data.frame() object containing the data
table_names	Character vector of table names with same length as race_group. Used for formatting output. If only one racial group, must provide "Pct. Other" as second element of vector
sample	Number of samples used for EI calculation, default = 1000
tomog	Logical to display tomography plot. If true will will save pdf plot to working directory. Default is FALSE
density_plot	Logical to display density plot of betab and betaw. If true will save pdf plot to working directory. Default is FALSE
plot_path	Path to save. If NULL, plot is not saved.
beta_yes	Logical to export betas (b, w) in list object in addition to table of results. Default is FALSE
seed	An integer seed value for replicating estimate results across runs. If NULL, a random seed is chosen. Defaulted to NULL.
...	Arguments passed onto ei() function

Value

Data frame/table object containing EI individually estimated results. If beta_yes=TRUE, two list items, first the data frame table of results, second dataframe of betas themselves.

Note

If this results in an error, "Error in .subset2(x, i, exact = exact) : invalid subscript type 'list'", just rerun the algorithm again.

Author(s)

Loren Collingwood <[e-mail]>

References

eiPack. Gary King (1997). A Solution to the Ecological Inference Problem. Princeton: Princeton University Press.

Examples

# TOY DATA EXAMPLE

canda <- c(.1, .09, .85, .9, .92)
candb <- 1 - canda
white <- c(.8, .9, .10, .08, .11)
black <- 1 - white
total <- c(30, 80, 70, 20, 29)
toy <- data.frame(canda, candb, white, black, total)

# CREATE VECTORS
cands <- c("canda")
race_group <- c("~ black") # only use one group for example
table_names <- c("EI: PCT Black", "EI: PCT White")

# RUN ei_est_gen()
# KEEP DATA TO JUST ONE ROW FOR EXAMPLE (time) ONLY!
ei_est_gen(cands, race_group, "total",
  data = toy[c(1, 3, 5), ], table_names = table_names, sample = 100
)

# WARNING -- May take a little while to execute
# Load Package Data
data(corona)
# Create Character Vectors
cands <- c("pct_husted", "pct_spiegel", "pct_ruth", "pct_button", "pct_montanez", "pct_fox")
race_group3 <- c("~ pct_hisp", "~ pct_asian", "~ pct_white")
table_names <- c("EI: Pct Hisp", "EI: Pct Asian", "EI: Pct White")

# Run ei_est_gen function
results <- ei_est_gen(
  cand_vector = cands, race_group = race_group3,
  total = "totvote", data = corona, table_names = table_names
)

results
# Run ei_est_gen function; Exporting betas into data frame
results_w_betas <- ei_est_gen(
  cand_vector = cands, race_group = race_group3,
  total = "totvote", data = corona, table_names = table_names, beta_yes = TRUE
)

res1 <- results_w_betas[[1]] # table of mean estimates
res1
res2 <- results_w_betas[[2]] # betas of estimates for each precinct

eiCompare documentation built on Aug. 31, 2023, 5:16 p.m.