differences between one-tailed and two-tailed P-value

[lgarcia]

I am doing hypothesis testing using t-test and p-value. I calculate my p-value based on the degrees of freedom and the t-test value. The p-value is a two-tailed calculation. But I am not very sure what the difference would be if I do it with one-tailed p-value. What is the meaning of that? How should I interpret that?
Right now I have around 18 to 25 degrees of freedom, depending on the test, and I may get values of p around 0.001-0.065 for a two-tailed calculation. I reject the test if I get more than 0.05 for the two-tailed calculation. Is that reasonable?
Thanks!

[fed1]

I reject the test if I get more than 0.05 for the two-tailed calculation. Is that reasonable?
Thanks!

I think it is reject if you get less than 0.05?

I would definitely use two sided, pretty much always.

[lgarcia]

fed1, thanks for your response. I'm following most literature I have found about p-value interpretation. Here is one from Wikipedia. From there that I am rejecting my null hypothesis when more than 0.05.
What is not very clear to me is the one or two-sided p-value and when to use one or the other.

[Joinedup]

Depends whether the hypothesis you're testing is one or two tailed.
The one tailed p table is appropriate for testing one tailed hypotheses and the two tailed p table is for two tailed hypotheses. You may need to skip back to the hypothesis section of your texts to sort it out.

Not that I'm any sort of expert.

[gianmarco]

Hi!!
As for two-tailed vs one-tailed test:
if we think to t-test as a means to ascertain the statistical significance of the difference between two samples' mean, the two-tailed version is appropriate when we have no expectations about the direction of that difference.

The opposite holds true for the one-tailed version.

Regards,
gm

[Ksharp]

I think it is reject null hypothesis if p value is less than 0.05 (Alpha=.05)

[lgarcia]

if we think to t-test as a means to ascertain the statistical significance of the difference between two samples' mean, the two-tailed version is appropriate when we have no expectations about the direction of that difference.

"the direction of the difference" means whether it is postive or negative?

[gianmarco]

Hi!

Yes, you got it.


Gm

[gianmarco]

see also:

http://stanford.edu/~maureenh/quals/...ts/node54.html

[lgarcia]

thanks! yeah, I had found this other one that I think is more explicit:

differences between one-tailed and two-tailed tests

It is from a very good source too. So, highly recommended for those interested