fast_jackknife_variance: Calculates Jackknife variance without reweighting

In NonProbEst: Estimation in Nonprobability Sampling

Description

Calculates the variance of a given estimator by Leave-One-Out Jackknife (Quenouille, 1956) with the original adjusted weights.

Usage

fast_jackknife_variance(sample, weights, estimated_vars, N = NULL)

Arguments

sample	A data frame containing the sample.
weights	A vector containing the pre-calculated weights.
estimated_vars	A string vector specifying the variables for which the estimators' variance are to be estimated.
N	Integer indicating the population size. Optional.

Details

The variance estimation is performed by eliminating an individual at each iteration with its corresponding weight and estimating the mean of the corresponding subsample, which is further used in the Jackknife formula as the usual procedure. The calculation of variance estimates through this procedure might take less computation time but also might not take into account the variance of the weighting method.

Value

A vector containing the resulting variance for each variable.

References

Quenouille, M. H. (1956). Notes on bias in estimation. Biometrika, 43(3/4), 353-360.

Examples

covariates = c("education_primaria", "education_secundaria")
data_propensities = propensities(sampleNP, sampleP, covariates)
psa_weights = sc_weights(data_propensities$convenience)
fast_jackknife_variance(sampleNP, psa_weights, c("vote_pens"), 50000)

Example output

Loading required package: lattice
Loading required package: ggplot2
   vote_pens 
0.0002292849 

NonProbEst documentation built on July 1, 2020, 6:08 p.m.