winswgt: Weighted winsorized mean and trimmed total
In martinSter/robsurvey: Robust Survey Statistics Estimation

Description Usage Arguments Details Value Utility functions Note See Also Examples

Weighted winsorized estimators of the mean and total are available in two forms:

bare-bone functions: weighted_mean_winsorized and weighted_total_winsorized,
estimation methods: svymean_winsorized and svytotal_winsorized (incl. variance estimation based on the functionality of the survey package).

weighted_mean_winsorized(x, w, LB = 0.05, UB = 1 - LB, na.rm = FALSE)

weighted_total_winsorized(x, w, LB = 0.05, UB = 1 - LB,
  na.rm = FALSE)

svymean_winsorized(x, design, LB = 0.05, UB = 1 - LB, ...)

svytotal_winsorized(x, design, LB = 0.05, UB = 1 - LB, ...)

`x`	numeric vector (`weighted_mean_winsorized` or `weighted_total_winsorized`); a formula object or variable name (`svymean_winsorized` or `svytotal_winsorized`)
`w`	numeric vector of weights
`LB`	lower bound of winsorizing, such that 0 ≤q LB < UB ≤q 1
`UB`	upper bound of winsorizing, such that 0 ≤q LB < UB ≤q 1
`na.rm`	a logical value indicating whether `NA` values should be stripped before the computation proceeds.
`design`	a `survey.design` object (see svydesign in survey)
`...`	additional arguments (not used)

Overview

Robust winsorized Horvitz–Thompson total or Hajek mean

bare-bone functions: return the estimate (no variance estimation)
estimation methods on the basis of survey (incl. variance estimation)

Variance

Variance estimates of the mean or total estimator are computed as first-order linearization using the design-based-estimation capabilities available in package survey.

Domain estimation

Estimates for domains can be obtained using the svyby wrapper in the survey package (see examples).

Estimate (scalar) or object of class svystat.rob

For the methods svymean_winsorized and svytotal_winsorized, the following utility functions can be used

summary gives a summary of the estimation properties
robweights retrieves the robustness weights
coef, vcov, residuals, and fitted retrieve, respectively, the estimate, variance, residuals and fitted values

winswgt is a generic name for the functions documented.

svymean_huber, svytotal_huber, svymean_trimmed, svytotal_trimmed, weighted_mean_huber, weighted_total_huber, weighted_mean_trimmed, weighted_total_trimmed

library(survey)
data(api)
dstrat <- svydesign(id=~1, strata=~stype, weights=~pw, data=apistrat, fpc=~fpc)
svymean_winsorized(~api00, dstrat, LB = 0.05)
# Domain estimates
svyby(~api00, by = ~stype, design = dstrat, svymean_winsorized, LB = 0.1)