Lecture 15

Question 1

Lecture 15:

What is the purpose of a Post Hoc test?

Answer 1

Runs comparisons between pairs of treatment levels

Question 2

Lecture 15:

What is a Tukey’s HSD Post Hoc Test?

Answer 2

Allows for only pairwise comparisons
- smaller mean differences are statistically significance
- simpler to calculate & applies to most research designs

Question 3

Lecture 15:

When should you perform a Post Hoc test?

Answer 3

Only perform when F is significant

Question 4

Lecture 15:

What is the first step of a Tukey’s HSD?

Answer 4

Conduct t-tests for each combination of levels of IV - two sample assuming equal variance

Question 5

Lecture 15:

What form of “correction” is used for Tukey’s test?

Answer 5

Bonferroni Adjustment - correction to prevent increasing Type I error rate

Question 6

Lecture 15:

What is the purpose of the Bonferroni Adjustment?

Answer 6

It is a correction to prevent increasing Type I error rate
- interpret significance based on adjusted P level
- eg; P = 0.0167 but based on alpha 0.05… < or >

Question 7

Lecture 15:

How do we identify pairwise differences?

Answer 7

Compare pairwise differences by comparing means

Question 8

Lecture 15:

What kind of Statistical test would you run…
“Investigating new drug for treating heart failure. Patients randomly assigned to 2 groups. One group gets a placebo & other receives new drug. Measure their left ventricle fraction after 6months”

Answer 8

This would be an independent T-Test because…
- two groups (placebo & drug)
- measured once (after 6months)
- control & non-control group

Question 9

Lecture 15:

What kind of Statistical test would you run…
“Want to now the age kids gain ability to catch a ball. Select kids from grades 2,3, & 4 and have them complete a skill assessment (success rate at catching a ball with 10 attempts)”

Answer 9

This would be a One-Way ANOVA because…
- 3 groups are measured (grades 2,3,4)
- only testing one thing (ability to catch a ball)
- testing on a ratio scale (scale up to 10)

Question 10

Lecture 15:

What kind of Statistical test would you run…
“Want to know if sprinting performance contributes to pole vault performance. Measure sprint speed & height achieved in pole vault in 12 varsity athletes.”

Answer 10

This would be a correlation because…
- looking if one relates to the other
- haven’t split into any groups, just looking if there’s a relationship

Question 11

Lecture 15:

What kind of Statistical test would you run…
“Want to see whether a new nutritional approach will improve marathon times. In a group of accomplished marathoners, you administer 12wks dietary intervention. Compare their race times before & after dietary intervention.”

Answer 11

This would be a Repeated Measures T-test because…
- measured twice 9 (pre & post test for same thing)
- same group of people measured both times

Question 12

Lecture 15:

What kind of Statistical test would you run…
“Want to find out which contributes to diabetes. Group of 1000 nurses; assess body weight, height, BMI, dietary habits, exercise, smoking, alcohol composition, & socioeconomic status. You also assess their level of Diabetes.

Answer 12

This would be a Regression test because…
- you’re looking at the possible predicting factors for something