Sign up  |  Log in

Doubt about 2- tailed t statistic test

When we are performing a two-tailed t-statistic test , at level of significance ( alpha) = 5 and p value = 2.7 ,

Do we compare p value with alpha or with alpha/2 i.e. will 2.7 be compared with 5 or 2.5? 

Personalized CFA® exam prep is here! Through this exclusive Analyst Forum offer, you can save 10% on a June 2020 Level II CFA PremiumPlus™ or Premium Study Package. Use promo code KS-AF10 at check out. Offer expires Oct 31, 2019.

Alpha

Alphas and p-values are bounded by 0 and 1. The numbers you gave do not meet this rule.

Amaira wrote:

When we are performing a two-tailed t-statistic test , at level of significance ( alpha) = 5 and p value = 2.7 ,

Do we compare p value with alpha or with alpha/2 i.e. will 2.7 be compared with 5 or 2.5? 

In a two-tailed test we compare alpha / 2 vs the p-value. I recommend you to use the common nomenclature for p-values and alphas which are 0.05, 0.01 , etc or use 5%, 2.7%, 2.5% at least.

Las almas de todos los hombres son inmortales, pero las almas de los justos son inmortales y divinas.
Sócrates

Harrogath wrote:

In a two-tailed test we compare alpha / 2 vs the p-value. I recommend you to use the common nomenclature for p-values and alphas which are 0.05, 0.01 , etc or use 5%, 2.7%, 2.5% at least.

In a two-tailed test, we compare the p-value (assuming a 2-tailed is generated by the software or looked up in a table) to alpha not alpha/2. Alpha and the p-value are only divided by two when doing a 1-tailed test since you want half the area under the curve (again, assuming everything was originally two-tailed and from a symmetric distribution).