0
$\begingroup$

I was told by a professor of a stats course last year (who I can't now contact), that it is unlikely to find statistically significant interaction terms using more than three factors.

Is this true?

If so, is there a reference to a journal/conference paper that I can cite?

Does the answer change depending on the type of data that is being analysed? (People vs chemical reaction vs algorithm execution times, etc) I have already looked through scholar.google.com, without much luck.

Improve this question
$\endgroup$
2
  • $\begingroup$ I would argue that it is not necessarily hard to find significant 3-way interaction terms, I think the biggest reason they are not used more is because the interpretation is difficult. See analytics.ncsu.edu/sesug/2008/ST-139.pdf . $\endgroup$
    – k6adams
    Commented May 1, 2015 at 4:58
  • $\begingroup$ @k6adams An interesting article, thanks. I'm still holding out for a "significant" citation. Also, its above three factors that I'm primarily interested in. $\endgroup$
    – Zack Newsham
    Commented May 1, 2015 at 15:52

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.