Start a weighting specification

Opens a weighting recipe on a sample and its design base weights. The object it returns is inert: it holds the data, the name of the base-weight column and an empty list of steps, and computes nothing. Every step_*() function takes such an object and returns it with one more step appended; prep() estimates the result.

Usage

weighting_spec(data, base_weights = NULL, nonprob = FALSE)

Arguments

data

data.frame with the sample units (one row per case).

base_weights

unquoted name of the design base-weight column. For a non-probability sample with no design weights, leave it NULL and set nonprob = TRUE: every unit then starts with a base weight of 1.

nonprob

logical. Declare the sample as non-probability (an opt-in panel, a volunteer or river sample). Required when base_weights = NULL. The flag is recorded so the report states the sample is non-probability and adds the methodological caveat; a non-probability sample is usually adjusted with step_pseudoweight() (inverse participation propensity against a reference) and/or step_calibrate() / step_model_calibration() to a reference_sample(). A non-probability panel that already carries recruitment/base weights can pass them as base_weights together with nonprob = TRUE.

Value

an object of class "weighting_spec".

Examples

rec <- weighting_spec(sample_survey, base_weights = pw)
rec
#> 
#> == Weighting specification (weightflow) ==
#> Data    : 467 cases
#> Base wts: pw
#> Steps   : (none yet)
#> Status  : not estimated
#> 
# a non-probability sample: no design weights, base weight 1
np <- weighting_spec(sample_survey, base_weights = NULL, nonprob = TRUE)