mrdwab
Conducting surveys is an expensive proposition, particularly when the populations you are working with are large.
Wouldn't it be nice if you could just survey a sample of your target population and still be able to say with certainty that your results are reliable?
Well, you can! You just need to know what the right sample size should be.
But first...
Magic!
Actually, no. That's not true. There's a pair of formulae that you need that looks like this:
\[ss = \frac{-Z^2\times p\times(1-p)}{c^2}\] \[pss = \frac{ss}{1+\frac{ss-1}{pop}}\]
Basically, you're going to have to provide:
(See this link for details on the formula.)
A basic function would look like this:
sample.size <- function(population, c.lev = 95, c.int = 5, distribution = 50) {
z = qnorm(0.5 + c.lev/200)
ss = (z^2 * (distribution/100) * (1 - (distribution/100)))/((c.int/100)^2)
ss/(1 + ((ss - 1)/population))
}
You could then use it like this:
sample.size(300)
## [1] 168.6957
Not too bad, but not too pretty either....
Use "devtools" and run it locally:
library(devtools)
library(shiny)
runGitHub("mrdwab/ddp_shiny_sample")
Two easy steps!
But wait! Do you want to see how the sample size changes when fiddling with the other input variables?
Use the sliders to get a live update of the resulting sample size!