1
$\begingroup$

One of the assumptions of Pearson's chi-squared test (for independence) is that expected frequencies are all >=5 in a 2x2 table OR 80% of the expected frequencies are >=5 in larger tables.

If this is not the case for a 2x2 table, Yates' correction can be applied.

When presenting data, it is good practice to state the assumptions. If Yates' correction has been applied, is it correct to:

a) show that the expected frequency assumption is NOT met (hence the use of Yates' correction)

OR

b) not state the expected frequency assumption at all i.e. does it cease to be an assumption of the test?

$\endgroup$
3
  • 1
    $\begingroup$ The idea is to tell people what they need to know to duplicate what you have done, so the answer is a). $\endgroup$ – Carl Mar 31 at 12:22
  • $\begingroup$ Thanks for replying. Why did you not post that as an answer? $\endgroup$ – dtw Mar 31 at 15:49
  • 1
    $\begingroup$ These criteria are not "assumptions" of the test: they are rules of thumb to indicate when you should be cautious about interpreting the p-value as computed from a chi-squared distribution. The 80% rule of thumb, in particular, is too crude for really large tables. The communications issues, then, are that when you are aware the computed p-value might be misleading, (a) how should you determine the p-value and (b) how should you explain your determination to your readers? $\endgroup$ – whuber Apr 3 at 13:11
0
$\begingroup$

The idea is to tell people what they need to know to duplicate what you have done, so the answer is a) show that the expected frequency assumption is NOT met (hence the use of Yates' correction). In general, when doing something that is more correct, for example, correcting standard deviation for small number bias, using Student's t rather than a normal distribution when it is correct to do so, it is incumbent upon the author present with clarity such that the ability of the reader to duplicate independently whatever was done is maintained. However, one does not have to belabor the point, e.g., "with Yate's correction" is enough.

$\endgroup$

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.