SSC Quickhelp

bullet1 Corrections

bullet2 Extreme proportions

If the proportion you expect is either very high (above .8) either very low (below .2) then you need a correction factor for extreme proportions to get a better estimate of the sample size needed for single proportion estimates.

You can safely leave this option checked. SSC will use it only when necessary.


» See also: Proportion

bullet2 Small sample sizes

As argued by Kupper and Hafner (1989), estimating a single value suffers from underestimation of the sample size needed if the sample size itself is small. SSC uses an approximation of the Kupper/Hafner findings to correct for small sample sizes.

You can leave this option checked. If sample size becomes bigger, the correction decreases.

bullet2 Small population

Most statistical tests used are developed with the assumption that the population is sampled with replacement. In practice, this is never the case: you never research the same individual twice in the same sample.

With large populations, the probability that you would investigate the same subject twice in the same sample becomes neglectible. However, if the population is quite small, it is likely you would have the same element more than once, if you'd use replacement.

Therefore, this correction factor applies. You can safely leave this option checked, unless you want to know what the underestimated effect of a small population is.


» See also: Population size

bullet2 Non-response

If not all sample elements will deliver measurements, you need to have a proportionately higher sample size to get the actual data volume you need.

Check this option only if response is below 100%.


» See also: Response %

bullet2 One-tailed test

SSC defaults to a two-tailed test for its calculations. If you happen to know the direction of the effect beforehand, you can check this option to conduct one-tailed tests.

For example, if you had reasons beforehand to believe Londoners earn more than Dubliners, you could check this option and reach a smaller sample size. On the other hand, if you do not know that Dubliners earn less, you would leave the option unchecked.


next References
» See also: Alpha