Export weightflow weights to a survey design — as

as_svydesign() builds a linearization (ultimate-cluster) design from a prepped recipe, treating the final weights as fixed; as_svrepdesign() builds a replicate-weights design from a bootstrap (weightflow_boot) or jackknife (weightflow_jack) object. Both require the 'survey' package. With the replicate-weights design you can estimate any statistic for any domain (svytotal, svymean, svyratio, svyby, ...).

Usage

as_svydesign(object, ids, strata = NULL, weight_name = ".weight", ...)

as_svrepdesign(object, ...)

Arguments

object: for as_svydesign, a prepped recipe or a data frame with the weight and design columns; for as_svrepdesign, a weightflow_boot or weightflow_jack object.
ids, strata: column names of the PSU and the stratum.
weight_name: name of the weight column.
...: passed to the survey constructor.

Value

A survey.design / svyrep.design object.

Details

Only as_svrepdesign() propagates the variability of the weighting adjustments (nonresponse, calibration, ...), because each replicate re-runs the whole recipe. as_svydesign() is design-based linearization on the fixed final weights: its standard errors reflect the sampling design but treat the adjustments as known without error, so they are usually smaller. Use as_svrepdesign() (with bootstrap_weights() / jackknife_weights()) when the adjustment variability should be included.