mp_bootstrap {manifestoR}

Compute bootstrap distributions for scaling functions

Description

Bootstrapping of distributions of scaling functions as described by Benoit, Mikhaylov, and Laver (2009). Given a dataset with percentages of CMP categories, for each case the distribution of categories is resampled from a multinomial distribution and the scaling function computed for the resampled values. Arbitrary statistics of the resulting bootstrap distribution can be returned, such as standard deviation, quantiles, etc.

Usage

mp_bootstrap(
  data,
  fun = rile,
  col_filter = "^per(\\d{3}|\\d{4}|uncod)$",
  statistics = list(sd),
  N = 1000,
  ignore_na = TRUE,
  rescale = TRUE,
  ...
)

Arguments

data

A data.frame with cases to be scaled and bootstrapped

fun

function of a data row the bootstraped distribution of which is of interest

col_filter

Regular expression matching the column names that should be permuted for the resampling (usually and by default the handbook v4_categories (plus cee_categories) per variables)

statistics

A list (!) of statistics to be computed from the bootstrap distribution; defaults to standard deviation (sd). Must be functions or numbers, where numbers are interpreted as quantiles.

N

number of resamples to use for bootstrap distribution

ignore_na

if TRUE (default), for each observation drop silently the columns that have an NA value for the permutation

rescale

if TRUE (default), rescale the permuted values after the permutation to the sum of the values of the col_filter columns instead of 100

...

more arguments passed on to fun

References

Benoit, K., Laver, M., & Mikhaylov, S. (2009). Treating Words as Data with Error: Uncertainty in Text Statements of Policy Positions. American Journal of Political Science, 53(2), 495-513. http://doi.org/10.1111/j.1540-5907.2009.00383.x