w8margin_matched: Check if w8margin Matches Observed Data

In svyweight: Quick and Flexible Survey Weighting

Check if w8margin Matches Observed Data

Description

Checks whether specified w8margin object and variable in observed data are compatible, and are expected to produce valid call to rake. Returns a logical true/false, and generates warning messages to specify likely issues. Intended to help quickly diagnose incompatibilities between w8margins and observed data.

Usage

w8margin_matched(w8margin, observed, refactor = FALSE, 
  na.targets.allow = FALSE, zero.targets.allow = FALSE)

Arguments

w8margin	w8margin object, or other object type that can be coerced to w8margin with a temporary variable name.
observed	factor vector (or, if refactor = TRUE, a vector that can be coerced to factor).
refactor	logical, specifying whether to factor observed variable before checking match.
na.targets.allow	logical, indicating whether NA values in target should produce error (FALSE, the default) or be allowed. NA values are never allowed in observed data.
zero.targets.allow	logical, indicating whether zero values in target should produce error (FALSE, the default) or be allowed.

Details

With default parameters (na.targets.allow = FALSE, zero.targets.allow = FALSE, and refactor = FALSE), the function checks whether a w8margin object is in the strict format required by rake; this format will also be accepted by rakesvy() and rakew8(). Changing the default parameters relaxes some checks. With the parameters altered, the function will only assess whether w8margin objects are usable by rakesvy() and rakew8(), which accept a more flexible range of target formats.

It should not generally be necessary to call w8margin_matched() manually when using rakesvy() and rakew8() to compute weights. However, may be useful to call directly, when manually calling underlying weighting functions from the survey package, or for diagnostic purposes.

Value

A logical, indicating whether w8margin is compatible with observed.

Examples

gender_w8margin <- as.w8margin(
    c(Male = .49, Female = .51),
    varname = "gender", 
    samplesize = 2179)
# Returns TRUE
w8margin_matched(gender_w8margin, gles17$gender)

gender_w8margin_alt <- as.w8margin(
    c(man = .49, woman = .51), 
    varname = "gender", 
    samplesize = 2179)
# Returns FALSE - level names in gles17$gender do not match level names in gender_w8margin_alt
w8margin_matched(gender_w8margin_alt, gles17$gender)

agecat_50plus_w8margin <- as.w8margin(
    c("50-59" = .35, "60-69" = .27, ">=70" = .38),
    varname = "educ",
    samplesize = 2179
)
gles17_50plus <- gles17[gles17$agecat %in% c("50-59", "60-69", ">=70"),]
# Returns FALSE - gles17$agecat has empty factor levels for <=29, 30-39, 40-49
w8margin_matched(agecat_50plus_w8margin, gles17_50plus$agecat)
# Returns TRUE - gles17$agecat is refactored to drop empty levels
w8margin_matched(agecat_50plus_w8margin, gles17_50plus$agecat, refactor = TRUE)

svyweight documentation built on May 3, 2022, 5:07 p.m.