One or two tailed test for t test

[archived_user]

How do you determine to use a one tailed or two tailed level of signifigance? What in the question do you base this off of?

 

[archived_user]

If the null hypothesis is μ = k, it’s a 2-tail test.
If the null hypothesis is μ ≤ k, or the null hypothesis is μ ≥ k,it’s a 2-tail test.

 

[archived_user]

S2000magician wrote:
If the null hypothesis is μ = k, it’s a 2-tail test.
If the null hypothesis is μ ≤ k, or the null hypothesis is μ ≥ k,it’s a 2-tail test.

If the null hypothesis is μ ≤ k, or the null hypothesis is μ ≥ k,it’s a 1-tail test.
Also you would use a t-stat when the # of observation is ≤ 30. Correct ?