ei_rc_good_table: Create EI Comparison Table

In lorenc5/eiCompare: Compares EI, Goodman, RxC Estimates

Description

Takes output from EI model, EI RxC model, Goodman regression, and puts them into a data frame table for useful analysis and comparison.

Usage

ei_rc_good_table(ei, rc, good, groups, include_good = F)

Arguments

ei	Table/data frame object result from ei_est_gen. This assumes beta_yes=F in ei_est_gen(). See example below for beta_yes=T in ei_est_gen().
rc	Table/data frame from EI:RxC process from bayes_table_make()
good	Table/data frame from Goodman regression, from goodman_generalize(). Default is nothing
groups	Character vector of voting blocks (e.g., c("Latino", "White"))
include_good	Logical, default is FALSE (F), Set to TRUE (T) if including a Goodman table/data object

Value

Object of class ei_compare containing a 1. data.frame() slot of comparisons across the three models; 2. Character vector of group names used for later plotting

Note

Most of the time the user will not include the Goodman table, as they are interested in the EI vs. EI:RxC comparison

Author(s)

Loren Collingwood <[e-mail]>

References

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

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!
results <- ei_est_gen(cands, race_group, "total",
                      data = toy[c(1,3,5),], table_names = table_names, sample=100)

# Generate formula for passage to ei.reg.bayes() function
form <- formula(cbind(canda,candb) ~ cbind(black, white)) 
# Run Bayesian model
suppressWarnings (
  ei_bayes <- ei.reg.bayes(form, data=toy, sample=100, truncate=TRUE)
)

table_names <- c("RxC: PCT Black", "RxC: PCT White")
cands <- c("canda", "candb")
ei_bayes_res <- bayes_table_make(ei_bayes, cand_vector= cands, table_names = table_names)
ei_bayes_res <- ei_bayes_res[c(1,2,5),]
# Combine Results, results in object of class ei_compare 
ei_rc_combine <- ei_rc_good_table(results, ei_bayes_res, 
                                  groups= c("Black", "White")
)
# Produces data and character vector, which can be sent to plot()
ei_rc_combine



# Warning: Takes a while to run
# Load corona data
data(corona)
# Generate 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 Lat", "EI: Pct Asian", "EI: Pct White")
# Run EI iterative Fitting
results <- ei_est_gen(cand_vector=cands, race_group = race_group3,
			total = "totvote", data = corona, table_names = table_names)

# EI: RxC model 
# Generate formula
form <- formula(cbind(pct_husted,pct_spiegel,pct_ruth,pct_button,pct_montanez,pct_fox) 
~ cbind(pct_hisp, pct_asian, pct_white)) 
suppressWarnings (
ei_bayes <- ei.reg.bayes(form, data=corona, sample=10000, truncate=TRUE)
)
# RxC table names
table_names <- c("RxC: Pct Hisp", "RxC: Pct Asian", "RxC: Pct White")
# Table Creation, using function bayes_table_make in ei_est_generalize.R file
ei_bayes_res <- bayes_table_make(ei_bayes, cand_vector= cands, table_names = table_names)


# Combine Results, results in object of class ei_compare 
ei_rc_combine <- ei_rc_good_table(results, ei_bayes_res, 
				groups= c("Latino", "Asian", "White")
				)
ei_rc_combine

# If set beta_yes=T in ei_est_gen():
#ei_rc_combine2 <- ei_rc_good_table(results[[1]], ei_bayes_res, groups= c("Black", "White"))

lorenc5/eiCompare documentation built on June 5, 2019, 5:18 p.m.