md_bayes_draw_lambda: MD Bayes Draw Lambda

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

Description

Tunes and estimates MD Bayes algorithm (ei.MD.bayes). Returns a data frame of lambda posterior distribution draws. Similar to md_bayes_draw, but used primarily for assessing posterior distribution tests.

Usage

  md_bayes_draw_lambda(dat, race_vote_split, form, 
                ntunes=10, totaldraws = 100000, seed=12345,
                sample = 100000, thin = 100, burnin = 100000,
                ret.mcmc=TRUE)

Arguments

dat	data.frame() object of just raw candidate vote and raw population counts. Put vote results in first set of columns, put population counts next
race_vote_split	Numeric vector of length 2 indicating where vote column ends (e.g., 3), and population counts begin (e.g., 4): c(3,4)
form	Formula object, e.g.: cbind(V1, V2, novote) ~ cbind(VtdAVap_cor, VtdBVap_cor, VtdHVap_cor, VtdOVap_cor)
ntunes	Numeric; how much to tune tuneMD. Default = 10
totaldraws	Numeric; How many total draws from MD. Default = 100000
seed	Numeric. Default = 12345
sample	Numeric. Default = 100000
thin	Numeric. Default = 10
burnin	Numeric. Default = 100000
ret.mcmc	Logical. Default = TRUE

Value

Posterior distribution of lambdas. This is often used for assessing RPB in elections with a small number of precincts.

Author(s)

Loren Collingwood <[e-mail]>; Justin Gross <[e-mail]>

References

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

Examples

  
  # TOY DATA EXAMPLE
  canda <- c(10,8, 10, 4, 8)
  candb <- 20-canda
  white <- c(15, 12, 18, 6, 10)
  black <- 20 - white
  toy <- data.frame(canda, candb, white, black)
  
  # Generate formula for passage to ei.reg.bayes() function
  form <- formula(cbind(canda,candb) ~ cbind(black, white)) 
  # Then excute md_bayes_draw(); not run here due to time
  # md_bayes_draw_lambda(toy, c(2,3), form )
    

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