I use a statistical program that does not seem to offer the option of conducting t-tests, ANOVA, etc as one-sided tests. Only a 2-sided p-value is returned. Is it equally valid to arrive at a one-sided test result by a) dividing the resulting p-value by 2, or b) keeping p-value as is, but compare it to alpha x 2. So in the latter case, rather than considering <= 0.05 as significant, consider <= 0.1 as significant. Are these approaches equally valid? Is one more commonly used compared to the other? Thanks in advance!
Asked
Active
Viewed 11 times
0
-
Neither is correct in general. This is discussed in many posts on site. See here. For another example see the comment thread here: https://stats.stackexchange.com/questions/359239/getting-1-tailed-p-value-from-a-2-tailed-p-value-in-spss-v23 ... the answer there is probably also useful. – Glen_b Dec 14 '23 at 00:39