2

I am curious if tests such as Rayleigh's test, Kuiper's test, and Watson's test are valid for bidirectional data (i.e. 180 degrees) as well as unidirectional data.
If not, what are the appropriate corresponding tests?

Nick Cox
  • 56,404
  • 8
  • 127
  • 185
user13449
  • 21
  • By "bidirectional data" do you mean data drawn from a bimodal distribution? – Mimshot Dec 07 '12 at 13:37
  • If there's only those two possible angles then there's no need for circular statistics at all. – John Dec 07 '12 at 13:53
  • Let me clarify - I'm testing for a bias in object orientation - objects have no top or bottom so 1 degree is exactly the same as 181 degrees. – user13449 Dec 07 '12 at 14:05
  • 4
    You should be able to double the values of all your angles and use standard circular tests. – Mimshot Dec 07 '12 at 14:20
  • Thanks mimshot - so does that mean that if I were using data that were 0-90 degrees I would quadruple it? – user13449 Dec 07 '12 at 14:39
  • 1
    Only if you considered 91 degrees to be the same as 1 degree (for example). – whuber Dec 07 '12 at 14:46

0 Answers0