rake_survey: Rake survey
In pewresearch/pewmethods: Pew Research Center Methods Miscellaneous Functions

Description Usage Arguments Details Value Examples

Applies the raking algorithm to a survey using specified raking targets in order to obtain weights. Wrapper around the calibrate function from the survey package with argument calfun = "raking".

rake_survey(
  .data,
  pop_margins,
  base_weight = 1,
  scale_to_n = TRUE,
  epsilon = 5e-06,
  maxit = 100,
  ...
)

`.data`	A `data.frame` containing the survey data
`pop_margins`	A list of tibbles giving the population margins for raking variables. `create_raking_targets` outputs such a list in the correct format.
`base_weight`	The survey's base weight variable, if applicable. Can be ignored if the survey data doesn't come with a base weight.
`scale_to_n`	If `TRUE`, scales the resulting weights so so that they sum to the number of observations in the survey.
`epsilon`	A `numeric` variable indicating the tolerance for the raking algorithm. When the proportions of the raking variables are within tolerance, the raking algorithm will stop.
`maxit`	Maximum number of iterations before stopping if raking has not converged.
`...`	Other arguments passed to `survey::calibrate()`.

The variables in .data must exactly match the variables in pop_margins both in name in the factor values. Additionally, they should contain no missing data.

A vector of survey weights.

library(dplyr)
# Prepare variables from the survey data for raking
dec13_excerpt_raking <- dec13_excerpt %>%
  mutate(rk_sex = sex,
         rk_recage = dk_to_na(recage, pattern = "DK/Ref"),
         rk_receduc = dk_to_na(receduc, pattern = "DK/Ref")) %>%
  impute_vars(.) %>%
  mutate(rk_sex_receduc = interaction(rk_sex, rk_receduc, sep = ":"))
# Prepare population marginal distributions for raking
# Here we will use the acs_2017_excerpt dataset included wih the package
targets <- create_raking_targets(acs_2017_excerpt,
                                 vars = c("sex", "recage", "receduc", "sex:receduc"),
                                 wt = "weight")

# Now that we have raking variables and population targets, we can create a raking weight
fake_weight <- rake_survey(dec13_excerpt_raking, targets)