# How to write a null hypothesis for a two tailed test

Some have argued that a one-tailed test is justified whenever the researcher predicts the direction of an effect. In this instance, Stata presents results for all three alternatives. Bond is better than chance at determining whether a martini is shaken or stirred, we would use a one-tailed probability.

Imagine you have developed a new drug that you believe is an improvement over an existing drug. In the previous example, only a sample mean much lower than the population mean would have led to the rejection of the null hypothesis.

The one-tailed hypothesis is rejected only if the sample proportion is much greater than 0. Compute the test statistic. And using that we saw that the result, the sample mean that we got, the 1. In order to reject the null hypothesis, the test statistic must be either smaller than —1.

If you look at the one-tailed test-- this area over here-- we saw last time that both of these areas combined are 0. In testing this drug, you are only interested in testing if it less effective than the existing drug. When using a two-tailed test, regardless of the direction of the relationship you hypothesize, you are testing for the possibility of the relationship in both directions. We now use the five-step procedure to test the research hypothesis that the mean weight in men in is more than pounds.

Besides, if we expect a specific tail under the alternative hypothesis, it is possible to choose a one-tailed hypothesis. Two-Tailed Versus One-Tailed Test When a hypothesis test is set up to show that the sample mean would be higher or lower than the population mean, this is referred to as a one-tailed test.

Rated 8/10 based on 31 review