Estimate the generalized entropy index, a measure of inequality
svygei(formula, design, ...)
# S3 method for survey.design
svygei(
formula,
design,
epsilon = 1,
na.rm = FALSE,
deff = FALSE,
linearized = FALSE,
influence = FALSE,
...
)
# S3 method for svyrep.design
svygei(
formula,
design,
epsilon = 1,
na.rm = FALSE,
deff = FALSE,
linearized = FALSE,
return.replicates = FALSE,
...
)
# S3 method for DBIsvydesign
svygei(formula, design, ...)a formula specifying the income variable
a design object of class survey.design or class svyrep.design from the survey library.
future expansion
a parameter that determines the sensivity towards inequality in the top of the distribution. Defaults to epsilon = 1.
Should cases with missing values be dropped?
Return the design effect (see survey::svymean)
Should a matrix of linearized variables be returned
Should a matrix of (weighted) influence functions be returned? (for compatibility with svyby)
Return the replicate estimates?
Object of class "cvystat", which are vectors with a "var" attribute giving the variance and a "statistic" attribute giving the name of the statistic.
you must run the convey_prep function on your survey design object immediately after creating it with the svydesign or svrepdesign function.
This measure only allows for strictly positive variables.
Matti Langel (2012). Measuring inequality in finite population sampling. PhD thesis: Universite de Neuchatel, URL https://doc.rero.ch/record/29204/files/00002252.pdf.
Martin Biewen and Stephen Jenkins (2002). Estimation of Generalized Entropy and Atkinson Inequality Indices from Complex Survey Data. DIW Discussion Papers, No.345, URL https://www.diw.de/documents/publikationen/73/diw_01.c.40394.de/dp345.pdf.
library(survey)
library(laeken)
data(eusilc) ; names( eusilc ) <- tolower( names( eusilc ) )
# linearized design
des_eusilc <- svydesign( ids = ~rb030 , strata = ~db040 , weights = ~rb050 , data = eusilc )
des_eusilc <- convey_prep(des_eusilc)
# replicate-weighted design
des_eusilc_rep <- as.svrepdesign( des_eusilc , type = "bootstrap" )
des_eusilc_rep <- convey_prep(des_eusilc_rep)
# linearized design
svygei( ~eqincome , subset(des_eusilc, eqincome > 0), epsilon = 0 )
#> gei SE
#> eqincome 0.13137 0.0024
svygei( ~eqincome , subset(des_eusilc, eqincome > 0), epsilon = .5 )
#> gei SE
#> eqincome 0.12161 0.002
svygei( ~eqincome , subset(des_eusilc, eqincome > 0), epsilon = 1 )
#> gei SE
#> eqincome 0.12053 0.0021
svygei( ~eqincome , subset(des_eusilc, eqincome > 0), epsilon = 2 )
#> gei SE
#> eqincome 0.13675 0.0035
# replicate-weighted design
svygei( ~eqincome , subset(des_eusilc_rep, eqincome > 0), epsilon = 0 )
#> gei SE
#> eqincome 0.13137 0.0024
svygei( ~eqincome , subset(des_eusilc_rep, eqincome > 0), epsilon = .5 )
#> gei SE
#> eqincome 0.12161 0.0019
svygei( ~eqincome , subset(des_eusilc_rep, eqincome > 0), epsilon = 1 )
#> gei SE
#> eqincome 0.12053 0.0019
svygei( ~eqincome , subset(des_eusilc_rep, eqincome > 0), epsilon = 2 )
#> gei SE
#> eqincome 0.13675 0.0032
if (FALSE) {
# linearized design using a variable with missings
svygei( ~py010n , subset(des_eusilc, py010n > 0 | is.na(py010n) ), epsilon = 0 )
svygei( ~py010n , subset(des_eusilc, py010n > 0 | is.na(py010n) ), epsilon = 0, na.rm = TRUE )
svygei( ~py010n , subset(des_eusilc, py010n > 0 | is.na(py010n) ), epsilon = .5 )
svygei( ~py010n , subset(des_eusilc, py010n > 0 | is.na(py010n) ), epsilon = .5, na.rm = TRUE )
svygei( ~py010n , subset(des_eusilc, py010n > 0 | is.na(py010n) ), epsilon = 1 )
svygei( ~py010n , subset(des_eusilc, py010n > 0 | is.na(py010n) ), epsilon = 1, na.rm = TRUE )
svygei( ~py010n , subset(des_eusilc, py010n > 0 | is.na(py010n) ), epsilon = 2 )
svygei( ~py010n , subset(des_eusilc, py010n > 0 | is.na(py010n) ), epsilon = 2, na.rm = TRUE )
# replicate-weighted design using a variable with missings
svygei( ~py010n , subset(des_eusilc_rep, py010n > 0 | is.na(py010n) ), epsilon = 0 )
svygei( ~py010n , subset(des_eusilc_rep, py010n > 0 | is.na(py010n) ), epsilon = 0, na.rm = TRUE )
svygei( ~py010n , subset(des_eusilc_rep, py010n > 0 | is.na(py010n) ), epsilon = .5 )
svygei( ~py010n , subset(des_eusilc_rep, py010n > 0 | is.na(py010n) ), epsilon = .5, na.rm = TRUE )
svygei( ~py010n , subset(des_eusilc_rep, py010n > 0 | is.na(py010n) ), epsilon = 1 )
svygei( ~py010n , subset(des_eusilc_rep, py010n > 0 | is.na(py010n) ), epsilon = 1, na.rm = TRUE )
svygei( ~py010n , subset(des_eusilc_rep, py010n > 0 | is.na(py010n) ), epsilon = 2 )
svygei( ~py010n , subset(des_eusilc_rep, py010n > 0 | is.na(py010n) ), epsilon = 2, na.rm = TRUE )
# database-backed design
library(RSQLite)
library(DBI)
dbfile <- tempfile()
conn <- dbConnect( RSQLite::SQLite() , dbfile )
dbWriteTable( conn , 'eusilc' , eusilc )
dbd_eusilc <-
svydesign(
ids = ~rb030 ,
strata = ~db040 ,
weights = ~rb050 ,
data="eusilc",
dbname=dbfile,
dbtype="SQLite"
)
dbd_eusilc <- convey_prep( dbd_eusilc )
# database-backed linearized design
svygei( ~eqincome , subset(dbd_eusilc, eqincome > 0), epsilon = 0 )
svygei( ~eqincome , dbd_eusilc, epsilon = .5 )
svygei( ~eqincome , subset(dbd_eusilc, eqincome > 0), epsilon = 1 )
svygei( ~eqincome , dbd_eusilc, epsilon = 2 )
# database-backed linearized design using a variable with missings
svygei( ~py010n , subset(dbd_eusilc, py010n > 0 | is.na(py010n) ), epsilon = 0 )
svygei( ~py010n , subset(dbd_eusilc, py010n > 0 | is.na(py010n) ), epsilon = 0, na.rm = TRUE )
svygei( ~py010n , dbd_eusilc, epsilon = .5 )
svygei( ~py010n , dbd_eusilc, epsilon = .5, na.rm = TRUE )
svygei( ~py010n , subset(dbd_eusilc, py010n > 0 | is.na(py010n) ), epsilon = 1 )
svygei( ~py010n , subset(dbd_eusilc, py010n > 0 | is.na(py010n) ), epsilon = 1, na.rm = TRUE )
svygei( ~py010n , dbd_eusilc, epsilon = 2 )
svygei( ~py010n , dbd_eusilc, epsilon = 2, na.rm = TRUE )
dbRemoveTable( conn , 'eusilc' )
dbDisconnect( conn , shutdown = TRUE )
}