A classic finding in psychological research is that people cannot tell when others are lying to them, yet they are very confident of their lie detection abilities. Kathryn Morris (2003) created an exercise for her social psychology students that was designed to enhance their understanding of the relation between confidence and accuracy in lie detection. To evaluate the effectiveness of the exercise, Morris recruited 32 students. She randomly assigned 17 of these students to participate in the exercise, and the remaining 15 students did not participate in the exercise. All 32 students completed a test on confidence and accuracy in lie detection. The tests were graded as either correct or incorrect. Morris’ results could be displayed in a 2 × 2 table where test result (correct or not correct) is cross tabulated by whether or not the student participated in the exercise.

1. Which statistical test seems most appropriate for the statistical analyses?

a. t test for one sample

b. t test for two independent samples

c. t test for two related samples

d. t test for a correlation coefficient

e. One-factor F-test

f. Two-factor F-test

g. Repeated-measures F test

h. One-variable χ² test

i. Two-variable χ² test

A paragraph from Morris’ results section is shown below. Use the information presented in this paragraph to answer questions 2 to 7.

After grading answers on a correct-incorrect dichotomy, a chi-square analysis revealed that significantly more students who participated in the exercise (71%) correctly understood the (non)relation between confidence and accuracy in lie detection ability than did students who had not participated in the exercise (13%), χ²(1, N = 32) = 10.61, p < .01. (p. 113)

2. The degrees of freedom for the analysis were equal to ___.

3. Obtained χ² was equal to ___.

4. With a significance level of .01, critical χ² is equal to ___.

5. What null hypothesis was Morris testing?

6. What was the alternative hypothesis?

7. Based on her results, what statistical decision would Morris have made? Reject or retain the null hypothesis?