
Confidence limits from a standard error, a coefficient of variation or a margin of error
Source:R/indigenous.R
limfrom.RdThe comparisons of this family take an estimate and its published confidence limits. Agencies that publish a standard error, a relative standard error (coefficient of variation) or a margin of error instead give the same information; this turns it into limits at the level wanted.
Details
With an estimate \(e\) and \(z = \Phi^{-1}(1 - \alpha/2)\) for
the level wanted, a standard error \(s\) gives \(e \pm z s\); a
coefficient of variation \(c\) (in percent) gives \(s = e c / 100\); a
margin of error \(m\) at level moelev (the American Community Survey
publishes 90% margins) gives \(s = m / z_{moe}\). The limits are symmetric
on the estimate's scale.
Examples
limfrom(c(24.1, 31.5), cv = c(8, 12))
#> ╭── Intervals from published precision ───────────────────────────────╮
#> │ Estimates 2 │
#> │ Interval normal, estimate +/- z x se (default 95%) │
#> │ │
#> │ estimate se interval │
#> │ -------- ----- -------------- │
#> │ 24.10 1.928 [20.32, 27.88] │
#> │ 31.50 3.780 [24.09, 38.91] │
#> │ │
#> │ Each interval is rebuilt from the published standard error, CV or │
#> │ margin of error, assuming a normal sampling distribution. │
#> ╰─────────────────────────────────────────────────────────────────────╯
limfrom(1520, moe = 210) # a 90% margin of error
#> ╭── Intervals from published precision ───────────────────────────────╮
#> │ Estimates 1 │
#> │ Interval normal, estimate +/- z x se (default 95%) │
#> │ │
#> │ estimate se interval │
#> │ -------- ----- ------------ │
#> │ 1,520 127.7 [1270, 1770] │
#> │ │
#> │ Each interval is rebuilt from the published standard error, CV or │
#> │ margin of error, assuming a normal sampling distribution. │
#> ╰─────────────────────────────────────────────────────────────────────╯