Lab 8 on Confidence Interval, Effect Size, and Statistical Power
1. Calculate the for the following fictional data regarding daily TV viewing 95% confidence interval
habits: = 4.7 hours; = 1.3 hours; sample of 78 people with a mean of 4.1 hours.µ 𝜎
a) What are the z cutoffs for the 95% confidence level? (1 point)
-1.96 & 1.96
b) What is the standard error? (1 point)
.147
c) What is the upper bound? (1 point)
4.388
d) What is the lower bound? (1 point)
3.812
e) State the confidence interval using brackets [] (1 point).
95% CI [3.812, 4.388]
2. Calculate the for the same fictional data regarding daily TV viewing habits: 99% confidence interval
µ = 4.7 hours; = 1.3 hours; sample of 78 people with a mean of 4.1 hours.𝜎
a) What are the z cutoffs for the 99% confidence level? (1 point)
-2.58 & 2.58
b) What is the standard error? (1 point)
1.3/ √78 = .147
c) What is the upper bound? (1 point)
4.479
d) What is the lower bound? (1 point)
3.721
e) State the confidence interval using brackets [] (1 point).
99% CI [3.721, 4.479]
3. Calculate the effect size for each of the following average SAT math scores. Remember, SAT math is
standardized such that = 500; = 100. And then, identify the size of effect using Cohen’s guidelines: µ 𝜎
small, medium, and large).
a) 61 people sampled with a mean of 480 (2pts)
-.2; small
b) 82 people sampled with a mean of 520 (2pts)
.2; small
c) 6 people sampled with a mean of 610 (2pts)
1.1; large
4. Define statistical power and state the minimum desired power for a research study. (2pts)
The minimum desired power is .8; statistical power is the probability of a null hypothesis being correctly
rejected.
5. If a study’s probability of making a type II error was .15 (15%), what was the study’s statistical power?
(Report as a proportion and a percentage) (2pts).
.85, 85%
5. List all five factors that affect statistical power and discuss why they would increase power (look at
book/PPT/notes) (5pts).
A) Increasing the effect size = greater distance and less overlap
B) Larger sample size = smaller standard error and larger test statistic.
C) One tailed test = increases the reject region
D) Decreasing the population SD = smaller denominator and larger test statistic
E) Increasing the alpha level = makes the test less conservative
6. Were you in class on Friday for Lab 8? (5pts)
No.
