Introduction to Hypothesis Testing Chapter 4 Introduction to Hypothesis Testing Part 2 Thurs. Sept. 5, 2013
One-Tailed and Two-Tailed Hypothesis Tests Directional hypotheses – can you make a prediction about the direction of the effect? One-tailed test – focus on either upper or lower tail of distribution Ex? Nondirectional hypotheses – cannot predict a direction of the effect Two-tailed test – check both tails of the distribution
Determining Cutoff Points With Two-Tailed Tests Divide up the significance between the two tails For 1 tailed-test, if using.05, all .05 is in the relevant tail For 2-tailed test, if using .05, split betw 2 tails, so .05/2, have .025 in each tail
1- vs. 2-tailed tests Given that a 1 tailed test has a larger rejection region, The 1-tailed, directional, test is preferable. If you have any idea of which direction your results may go, specify that in the Research Hyp. Leads to better theory testing. But…another viewpoint:
Notice the change in critical values: For .05 significance level, 1-tailed test has 5% of scores in one tail critical value is 1.64 (or -1.64 if lower tail) 2-tailed test has 2.5% of scores in each tail critical values are 1.96 and -1.96 How does this affect the rejection region? Two-tailed test example:
Hypothesis Tests in Research Articles Reported with regard to specific statistical procedures (r = .78, p < .05) Look for p < .05 or p < .01 or asterisks in tables (*) to indicate stat significance “Near significant trend” if p < .10 Not significant, noted by ‘ns’
Hypothesis Tests in Research Articles Shown as asterisks in a table of results *Indicates that these results fell in the ‘rejection region’ and they differ significantly from the null hyp. (If no, asterisk failed to reject the null)