I have a data frame with prices/dates, is there a method that checks if the variance is homogeneous during all the series? I know there are many test like fligner.test, bartlett.test but I need to pass to those methods few groups....I don't have groups I have only one series and I need to know if the variance is constant.

Thank you

  • $\begingroup$ Do you have several prices for a given date? $\endgroup$ – ocram Nov 7 '11 at 13:24
  • $\begingroup$ Is this a time series, in which you wish to check if the variance remains constant over time? $\endgroup$ – onestop Nov 7 '11 at 13:31
  • $\begingroup$ @ocram, no ONE price and ONE date fro about 700 days. $\endgroup$ – Dail Nov 7 '11 at 13:58
  • $\begingroup$ @onestop, Exactly, That's what I am looking for. $\endgroup$ – Dail Nov 7 '11 at 13:58

What you're looking for is a test for homoskedasticity. I haven't need to use this myself yet, so I can't offer much in-depth advice, but the wikipedia page on homoskedasticity lists a couple of tests, including the Breusch–Pagan test, which assumes normality in the data, the Koenker–Basset test which generalises the Breusch–Pagan test for non-normal data.

the bptest function in the lmtest R package implements the both tests (set studentize=true for the Koenker–Basset test). Looks like it's only for linear models though, so it might not work for your data.

There's a few interesting questions about these test on this site. Check out charlie's answer on Why does the Breusch-Pagan test fail? in particular.


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.