Hypothesis Testing and Effect Size in Social Research

Part A1: Hypothesis examples for attitudes towards gun control and tolerance in adolescents

For each of the following research questions, create one null hypothesis, one non-directional research hypothesis, and one directional hypothesis.

1. What are the effects of working remotely on productivity?
   
   
2. What is the relationship between social isolation and depression?
   
   
3. What is the best way to treat an eating disorder?
   
   
4. How do early intervention programs impact adolescent drug use?

Using the data from Chapter 11 Data Set 2, test the null hypothesis that urban and rural residents have the same attitudes toward gun control. Use the T.TEST function to test the hypothesis. Assume that the variance between each group is equal.

1. State the null hypothesis (4 pts.)
   
   
2. State a directional research hypothesis (4 pts.)
   
   
3. Perform the T.TEST function in Excel (4 pts.)
   
   
4. Write 2-3 sentences reporting your findings. (6 pts.)
   
   
5. Rerun the analysis assuming the variance is not equal and compare your findings to the original T.TEST (4 pts.)
   
   
6. Do the two analyses return similar or different results? Write 1-2 sentences reporting your findings. (6 pts.)

Use the data from Chapter 12, Data Set 3 to see whether counseling adolescents at the beginning of the school year had a positive impact on their tolerance for other adolescents who are ethnically different from them. Assessments are made right before the treatment and then 6 months later. Did the program work? The outcome variable is score on the test of attitudes toward others, with possible scores ranging from 0 to 50; the higher the score, the more tolerance. Did the program work?

1. Manually compute the t-value and write a 2-3 sentence conclusion about the effectiveness of the program. (6 pts.)
   
   
2. Compute the t-test in Excel. (5 pts.)
   
   
3. You do not need to rewrite your conclusion based on the Excel analysis (unless you believe your original computation was done incorrectly); however, did you compute the same value as Excel? If not, describe the difference and explain why you think there is a discrepancy. (5 pts.)

Reading the Literature – Effect Size (d)

Article available .

The final sample consisted of 74 foster parents….The foster children of parents in sample were mostly boys…..

Qualified participants were randomly assigned to either an intervention or a control group…The intervention period….continue for two weeks. Parents in the intervention group received the Anger Outbursts DVD and were instructed to view the training materials at least once a week. Parents in the control group received no materials during this time.

The Parent Knowledge questionnaire included 20 multiple-choice and true-false items based on content from all segments in the Anger Outbursts program…. Scores were reported as the percentage of items correct out of 20. High scores on this measure indicated greater knowledge about anger issues in children.

The Parent Perception questionnaire included [a measure of] parents’ confidence in having the relevant parenting skills (three items, for example, “I can recognize the different types of anger outbursts in children”). Parents rated on a four-point scale (1 = not at all; 2 = a little; 3 = mostly; 4 = very much)…. Scores were reported as the average rating over all items.

Posttest means are presented in Table 1.

Means Performance on Parent Knowledge and Confidence Scales at Posttest for the National Sample of Foster Parents

   Intervention                    Control

Measure    M           SD       n           M      SD      n

Parent Knowledge^a    71.32         12.39    34         61.38     9.41      40

Parent Confidence^b   2.99           0.42      34           2.84      0.49     40

^a Standardized mean difference statistic (d) = 0.91

^b d = 0.33

Based on the information provided in the excerpt, answer the following:

1. The average intervention group parent answered what percentage of the knowledge questions correctly on the posttest (i.e., what is the mean percentage)?
   
   
2. The average control group parent answered what percentage of the knowledge questions correctly on the posttest (i.e., what is the mean percentage)?
   
   
3. What is the value of dfor the difference between your answers to Questions 1 and 2?
   
   
4. How would you describe the size of the effect for your answer to Question 3?
   
   
5. The mean “Parent confidence” score for the intervention group exceeds the mean “Parent confidence” score for the control group by how many points?
   
   
6. What is the value of d for the difference in your answer to Question 5?
   
   
7. How would you describe the size of the effect for your answer to Question 6?
   
   
8. The difference between the mean on “Parent confidence” for the intervention group (m = 2.99) and the mean for the control group (m = 2.84) is small in absolute terms (considerably less than one point). Yet the effect size is moderate. Speculate on why this is the case.