Create the inputs to evaluate the quality of proportion estimations

create_prop generates a dataframe with the following elements: sum, degrees of freedom, sample size, standard error, and coefficient of variation. The function allows grouping in several domains.

Usage

create_prop(
  var,
  denominator = NULL,
  domains = NULL,
  subpop = NULL,
  design,
  ci = FALSE,
  deff = FALSE,
  ess = FALSE,
  ajuste_ene = FALSE,
  rel_error = FALSE,
  log_cv = FALSE,
  unweighted = FALSE,
  standard_eval = FALSE,
  eclac_input = FALSE,
  ci_logit = FALSE,
  scheme = c("eclac_2020", "eclac_2023")
)

Arguments

var: numeric variable within the dataframe, is the numerator of the ratio to be calculated.
denominator: numeric variable within the dataframe, is the denominator of the ratio to be calculated. If the var parameter is dummy, it can be NULL.
domains: domains to be estimated separated by the + character.
subpop: integer dummy variable to filter the dataframe.
design: complex design created by survey package.
ci: boolean indicating if the confidence intervals must be calculated.
deff: boolean design effect.
ess: boolean effective sample size.
ajuste_ene: boolean indicating if an adjustment for the sampling-frame transition period must be used.
rel_error: boolean relative error.
log_cv: boolean logarithmic coefficient of variation.
unweighted: boolean add non-weighted count if required.
standard_eval: boolean indicating if the function is wrapped inside another function, if TRUE avoid lazy eval errors.
eclac_input: boolean return eclac inputs
ci_logit: boolean indicating if interval confidence is logit, only available for proportions.
scheme: character variable indicating the evaluation protocol to use for CEPAL standard. Options are "eclac_2020" and "eclac_2023". The "eclac_2020" option does not support ratio estimation.

Value

dataframe that contains the inputs and all domains to be evaluated.

Examples

library(survey)
library(dplyr)

epf <- mutate(epf_personas, gasto_zona1 = if_else(zona == 1, gastot_hd, 0))
dc <- svydesign(ids = ~varunit, strata = ~varstrat, data = epf, weights = ~fe)
old_options <- options()
options(survey.lonely.psu = "certainty")

create_prop(var = "gasto_zona1", denominator = "gastot_hd", design = dc)
#>        stat         se  df     n         cv
#> 1 0.6138541 0.02485186 248 48308 0.04048496

enusc <- filter(enusc, Kish == 1)

dc <- svydesign(ids = ~Conglomerado, strata = ~VarStrat, data = enusc, weights = ~Fact_Pers)
options(survey.lonely.psu = "certainty")
create_prop(var = "VP_DC", denominator = "hom_insg_taxi", design = dc)
#>        stat         se  df     n         cv
#> 1 0.8937474 0.04366055 486 24465 0.04885111
options(old_options)