Home > Statistics Every Writer Should Know > The Stats Board > Discusssion
one vs two tailed tests
Message posted by Kate (via 195.107.47.211) on June 21, 2001 at 11:52 AM (ET)
Ok, I've done a huge correlation matrix, 2 tailed, and I'm wondering if I can just make my results onetailed by dividing my r by two. I'm foolishly doing my hypotheses retrospectively and some I want to say are positively related and other findings I just want to say I unexpectedly found, should I report those as two tailed? Confused, hope haven't confused you.....
READERS RESPOND:
(In chronological order. Most recent at the bottom.)
Re: one vs two tailed tests
Message posted by JG (via 128.8.23.72) on June 21, 2001 at 4:10 PM (ET)
Your comments are extremely confusing and seem to be meaningless. Dividing correlation by 2 can't possible have any meaning and has nothing to do with one-tailed or two-tailed tests.
Re: one vs two tailed tests
Message posted by Peter (via 131.193.141.96) on June 26, 2001 at 5:11 PM (ET)
You cannot make a two tailed test one tailed by dividing your r by 2. Two tailed test means you are looking for a significant difference in either a positive or negative direction of your distribution. A two-tailed test is appropriate when you are not concerned with the direction of the signifianct result, just that you are looking for a significant result. A one-tailed test is appropriate when you want to specify the direction of your significant result. given your situation, reporting a one or two-tailed test does not enhance the significance of a result that you unexpectedly found. I would simply report the results and indicate which ones were significant at the two-tailed test level, since it seems you do not have any apriori hypotheses indicating which direction you hoped the correlation would be. a data example would help clarify your problem further.
Your $5 contribution helps cover part the $500 annual cost of keeping this site online.
