In: Statistics and Probability
what would the effect be of a one-tailed test and the probability of rejecting the null hypothesis?
This question is explained with the help of an example.
Suppose we are testing for a Mean with population standard deviation known, . So we would be using Normal Distribution and thus Z-Test for the following Hypothesis:
Case 1. Null Hypothesis, Ho: versus Alternative Hypothesis Ha:
Case 2. Null Hypothesis, Ho: versus Alternative Hypothesis Ha:
Tails of a test are dependent on the alternative hypothesis.
Case 1 corresponds to a two tailed test since it has not equals to symbol while Case 2 corresponds to a One-Tailed Test (Right Tailed) since it has a greater than symbol.
Now the effect of the type of Test - One Tailed or Two tailed comes into effect while taking the Decision and concluding for Rejection or Acceptance of Null Hypothesis.
For Case 1.
The Critical Value of Z would correspond to the Z-score of and the probability for Rejecting Ho will be given as: , Zobserved is the value of Test Statistics.
The Rejection Region is thus,
For Case 2.
The Critical Value of Z would correspond to the Z-score of and the probability for Rejecting Ho will be given as:
Thus, The Rejection Region is thus, .
Thus, For One Tailed and Two Tailed Tests the Rejection Regions are different and the corresponding probability of rejecting null hypothesis i.e. p-value varies.
If Rejection Regions and p-values are wrong basis the type of test, then we may wrongly Reject Ho when it is True and may wrongly accept Ho when it is false.