Package WgtEff

library(WgtEff)

Weighting survey data in order to reduce sampling bias usually increases the data’s variance; that increase in variance diminishes the survey’s precision; and that loss in precision, a component of Total Survey Error (TSE), can be quantified by statistics which package WgtEff calculates for n = sample size and x = weight values:

• Design effect (DEFF) due to weighting => n * (sum(x^2) / sum(x)^2)

• Weighting loss => DEFF-1

• Root design effect (DEFT) => square root of DEFF

• Effective sample size (ESS) => sum(x)^2 / sum(x^2)

• Weighted margin of error (MOE) => unweighted MOE * DEFT

Package WgtEff contains functions for these statistics; it also contains a function FULL(x) that calculates and prints all of these statistics at once, along with an unweighted margin of error and sample size.

Package WgtEff works directly from the survey data’s weights column - no hand calculations required; just upload the data set, input the name of the weights column, and run the functions.