In the book: "Essentials of Statistics for Business & Economics", the part about matched sample t-test, I see that it uses the two-tailed t-stat. However, instead of stating the null and alternate hypotheses as "equal 0" and "not equal 0", it sets: "H0 d <= 0" and "H1: d > 0", which seems like a one-tailed test. I'm wondering whether this is correct?
Asked
Active
Viewed 29 times
4
-
Why do you want to include < 0 as part of the null hypothesis? – Michael R. Chernick Sep 29 '19 at 17:11
-
4It sounds like you're correct and the book has an error. – Sal Mangiafico Sep 29 '19 at 18:04
-
2That's certainly a one-tailed test as the hypotheses are stated. – Glen_b Sep 30 '19 at 03:03
1 Answers
0
Then the book has a mistake. A two-sided t-test has $H_0: d = 0$ and $H_a: d \ne0$.
(I assume that $d = \mu_1 - \mu_2$ or $d = \mu_2 - \mu_1$.)
Dave
- 28,473
- 4
- 52
- 104