mk_stat: Estimate a statistic from a simple random sample design

In danjdrennan/surveyr: Summary Statistics and Visualizations for Surveys

Description

Function to get point estimates and statistics associated with them. Given a point estimate (mean, total, prop), mk_stat will generate the

• sample size, n, at calling time

• point estimate

• estimator variance

• standard error

• coefficient of variation assuming a simple random sample of data.

Usage

mk_stat(y, stat = "mean", N = NULL, fpc = TRUE, weights = NULL)

Arguments

y	A numeric vector with at least two elements
stat	A string, either "mean", "total", or "prop" (proportion)
N	The population size. This must be supplied and must be larger than length(y). If unknown, recommend passing it as some N > n with fpc=FALSE.
fpc	TRUE/FALSE. Whether or not to compute a finite population correction (for variance).
weights	Optional. If passed, it must be a vector of weights at least as long as the input y.

Value

tibble with

• n (sample size)

• point estimate (stat: mean, total, proportion)

• estimator variance (not the same as population variance)

• estimator standard error

• coefficient of variation

References

Lohr, Sharon L. Sampling: Design and Analysis. Chapman and Hall/CRC, 2019.

Examples

N <- 10
n <- 5
y <- c(0, 0, 0, 1, 1)
weights <- rep(N/n, n)
mk_stat(y, stat="mean", N, fpc=TRUE, weights=weights)
# A tibble: 1 x 5
#      n point   var    se    cv
#  <int> <dbl> <dbl> <dbl> <dbl>
#1     5   0.4  0.15 0.387 0.968

N <- 10
n <- 5
y <- c(0, 0, 0, 1, 1)
weights <- rep(N/n, n)
mk_stat(y, stat="total", N, fpc=TRUE, weights=weights)
# A tibble: 1 x 5
#      n point   var    se    cv
#  <int> <dbl> <dbl> <dbl> <dbl>
#1     5     4    15  3.87 0.968

#' N <- 10
n <- 5
y <- c(0, 0, 0, 1, 1)
weights <- rep(N/n, n)
mk_stat(y, stat="prop", N, fpc=TRUE, weights=weights)
# A tibble: 1 x 5
#      n point   var    se    cv
#  <int> <dbl> <dbl> <dbl> <dbl>
#1     5   0.4  0.03 0.173 0.433

danjdrennan/surveyr documentation built on Dec. 19, 2021, 8:08 p.m.