The Student Room Group

t-tests - converting two-tailed results to one-tailed

Hi,

Does anyone know how to convert a two-tailed result into a one-tailed? I'm doing my dissertation and as my hypotheses are one-tailed, when im reporting statistically significant results, they have to be one-tailed too.

I can't get help from my supervisor etc as its my dissertation but im sure i read somewhere (can't remember where) something about halving a two-tailed significance result to get a one tailed sig result (though it may have been double it....:confused:)

I'm mighty confused and stressin myself out over this... please help!!

Thanks

Clare :smile:
Reply 1
why don't you just do a one tailed result in the first place? with the critical region at 1.64
Reply 2
1 tailed: Specifies a direction, and a result.
2 tailed: No direct direction, just an effect.

Eg 1 Tailed. Words with meaning will be remembered more than names.

2 tailed: There wiill be a difference in the recall of words with meaning and names.

HOPE THIS HeLPs!

S
Reply 3
SNAZZ42
Eg 1 Tailed. Words with meaning will be remembered more than names.

2 tailed: There wiill be a difference in the recall of words with meaning and names.
S


or more concisely put:
Ho : x= 0
H1 : x > 0 (one tailed) or H1: x does not equal 0 (two tailed)

You need to alter your alternate hypothesis and critical value (and therefore region). A two tailed test of 5% los will have one same crit value as a one tailed test with 2.5% los, and a one tailed test with 5% los is half equivalent as a two tailed test with 10% los. Therefore if your los remains the same but your doing a one tailed test, your critical region is going to increase there will be increase liklihood that you will obtain a significant result (providing the anticipated effect is in the right direction). Hope this makes sense...
Tom
Reply 4
This is correct, u just half the value you got from spss to convert it into a one tailed result.
Hi, if you half the values, do we still round to 3dp for the p value? e.g. .003/2 = .0015, would this be rounded to .002?