The finite population correction is used when the sample size is a large fraction (approximately 5% or more) of the total population. You can apply a finite population correction as a design effect on a table.
The formula for the finite population correction is \(\sqrt{\frac{N-n}{N-1}}\), where \(N\) is the total population and \(n\) is the sample size. However, when calculating this correction for use as a design effect in Displayr, you should not use the square root.
Method
- Calculate the correction outside of Displayr. Remember to use \(\frac{N-n}{N-1}\) (i.e., without the square root).
- Select the table and from the object inspector, go to Properties >APPEARANCE > Significance > Advanced.
- Set Test Type > Proportion tests to z-test and uncheck Bessel's correction.
- Set Significance Levels > Weights and significance to Un-weighted sample size in tests.
- Put the finite population correction that you calculated previously in the Extra deff field.
Note that this procedure applies to the whole table, so if you have a table with multiple sub-populations, you should filter the table to only show the sub-population of interest.