Designing Epidemiological Studies (3)

Question 1

When do you use odds ratio VS relative risk?

Answer 1

Odds ratio -

Relative risk -

Question 2

What is a Type I and Type II error?

Answer 2

Type I - false positive, inaccurately reject null

Type II - false negative, inaccurately accept null

Question 3

What is statistical power?

Answer 3

Statistical power = ability to detect an effect

Question 4

How does sample size affect statistical power?

Answer 4

The bigger the sample size, the more statistical power you have

Question 5

Statistical significance VS experiment significance?

Answer 5

Statistical significance - relationship found in results statistically is significant

Clinical significance - clinically statistical relevance does not matter much

Question 6

Specify the size of your ‘effect of interest’ – that is the difference between the “incidence” in Group 1 and Group 2

Answer 6

x

Question 7

How does changing your power affect the sample size?

Answer 7

x

Question 8

How does changing the size of the ‘effect of interest’ affect the sample size?

Answer 8

high incidence = small the sample size required

Question 9

If 20% of your sample drops out over the study period (are lost to follow-up), how could you mitigate this risk?

Answer 9

Increase the sample size

Question 10

What are the trade-offs and how do you balance the risks arising of Type I and Type II error?

Answer 10

x