surveytable (version 0.10.0)

svyciprop_adjusted: Korn and Graubard confidence intervals for proportions, adjusted for degrees of freedom

Description

A version of survey::svyciprop( method = "beta" ) that adjusts for the degrees of freedom.

Usage

svyciprop_adjusted(
  formula,
  design,
  level = 0.95,
  adj = "none",
  aa_vr = NULL,
  aa_pop = NULL,
  ...
)

Value

The point estimate of the proportion, with the confidence interval as an attribute.

Arguments

formula

see survey::svyciprop().

design

see survey::svyciprop().

level

see survey::svyciprop().

adj

adjustment to the Korn and Graubard confidence intervals: "none" (default), "NCHS", or "NHIS".

aa_vr

used to produce age-adjusted confidence intervals only. The name of a categorical age variable located in design.

aa_pop

used to produce age-adjusted confidence intervals only. A data.frame with columns named Level and Population. Level must exactly match the levels of aa_vr. Population is the population count or proportion/weight for that level of aa_vr.

...

see survey::svyciprop().

Details

adj specifies the adjustment to the Korn and Graubard confidence intervals.

• "none": No adjustment is performed. Produces standard Korn and Graubard confidence intervals, same as survey::svyciprop( method = "beta" ).

• "NCHS": Adjustment that might be required by some (though not all) NCHS data systems. With this adjustment, the degrees of freedom is set to degf(design). Consult the documentation for the data system that you are analyzing to determine if this is the appropriate adjustment.

• "NHIS": Adjustment that might be required by NHIS. With this adjustment, the degrees of freedom is set to nrow(design) - 1. Consult the documentation for the data system that you are analyzing to determine if this is the appropriate adjustment.

To use these adjustments in surveytable tabulations, call set_survey() or set_opts() with the appropriate mode or adj argument. Age-adjustment can be turned on with set_survey(). But if adj = "none", no age-adjustment is performed.

For age-adjustment, aa_pop$Population can contain either population counts or proportions/weights for each level. Values are normalized internally, so counts and proportions produce the same confidence intervals when they describe the same standard population distribution.

Originally written by Makram Talih (2019). Age-adjusted calculation based on Natalie Young (2026).

Examples

set_survey(namcs2019sv)
set_opts(adj = "NCHS")
tab("AGER")
set_opts(adj = "none")