1
$\begingroup$

I've got a few frequencies, and I'm trying to work out if each of the frequencies is significantly different from zero. Most of the frequencies I can eyeball, but I wanted to have some for of statistic for some of the other frequencies.

Does anybody know of an appropriate statistical test for this? I couldn't seem to get SPSS or SAS to do this with Chi-square.

I've seen similar 'significantly different from zero' tests for continuous or ordered-categorical data but haven't found anything for frequencies.

Any ideas?

Simon.

$\endgroup$
3
$\begingroup$

Note that if a population proportion is 0, you can't observe any counts in that category.

As a result you can always use this rejection rule:

If a sample count is not 0, reject the hypothesis that the population frequency is 0.

This test has type I error rate of 0 (there are no false rejections, since you can never reject unless the population frequency is non-zero). Power depends on sample size and the underlying frequency.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.