Name:______

Assignment #5 Due: Wednesday 12/6

1. A history teacher is interested in the efficacy of a tutoring strategy that encourages students to reenact historical events. She randomly assigns students to one of two groups: reenactment and no-reenactment (control). After tutoring each student individually, she gives them a test of the information taught in both conditions. The results are provided in the table below.

Reenact / Control
19 / 26
20 / 24
20 / 25
21 / 23
18 / 29
30

a) Using the hypothesis testing format provided in class, α = .05, a two-tailed test, and the data in the above table determine whether there is a statistically significant difference between the two groups on test scores? Check your hand calculations using SPSS and attach your output.

b) Calculate Cohen’s d for the difference between the two conditions.

c) According to Cohen how large is this difference?

d) When compared to the control group what percentile is the mean of the reenact group?

e) Based on the above analysis the researcher concludes that the new strategy is less effective than the standard strategy. What aspect of the study improves the teacher’s ability to make this causal claim?

2. A researcher is interested in the effectiveness of a new counseling strategy that purports to alleviate negative self image. Participants were assigned to one of two groups based upon a clinic worker’s perception of the patient’s need. After a period of time, participants were measured using a self-report of their current self image (A scale of 1 to 20, with 1 = negative self image and 20 = positive self image). Below are relevant descriptive statistics.

a)  Using the hypothesis testing format provided in class, α = .05 and a one-tailed test, perform the appropriate statistical test by hand, to determine if there is a statistically significant difference between groups.

b)  What key assumption of the independent samples t-test is being violated? How do you know?

c)  Based on the above analysis the researcher concludes that the new strategy is more effective than the standard strategy. What error is the researcher making?


3. A researcher is interested in testing the efficacy of a reading strategy that emphasizes the memorization of sight words. Previous research on the topic, using a 12 item reading comprehension test, found a difference between control (c= 6 words) and memorization (e = 8 words). The mean difference was 2 words with a pooled standard deviation of 4 items. Assuming a one-tailed test calculate the following by hand

a) According to Cohen how large is this difference?

b) When compared to the control group what percentile is the mean of the experimental group?

c) Using Piface (see tutorial for how to use Piface), and the effect size from above (Remember d = (e - c )/ sp ). Calculate the power for studies of the following sizes (N = n1+n2).

N / Power (1 – β)
24
44
54
64
74
84
94
104

d) Comment on what happens to the statistical power of the study as the sample size increases.


e) Using Piface, and d or Δ = .25. Calculate the power for studies of the following sizes (N = n1+n2).

N / Power (1 – β)
24
44
54
64
74
84
94
104

f) Comment on the power of this set of studies relative to the previous set. Why has the pattern changed?

4. The following questions relate to the article Some Practical Guidelines for Effective Sample Size Determination by Russell Lenth.

a) Lenth argues that sample size estimation is important for economical reasons. Briefly discuss the reasons he gives for this position.

b) What sources does the author suggest for finding variance estimates?

c) What are the shortcomings of using “canned” effect sizes?