Power & Sample Size

Question 1

type 1 error

Answer 1

rejecting H0 when it is true (false positive)

Question 2

type 2 error

Answer 2

Failing to reject H0 when it is in fact false (false negative)

Question 4

what is α?

Answer 4

probability of a Type I error

Question 5

what is β?

Answer 5

probability of a Type II error

Question 6

what is power?

Answer 6

1 – β(type 2 error) = Prob (reject H0 when H0 false)
.i.e the probability of being correct

Question 7

why do we perform post-hoc power calculations?

Answer 7

to distinguish between studies which are ‘negative’ because there are no clinically important differences and those which are negative because the sample size was too small

Question 8

difficulties with post-hoc power calculations

Answer 8

• several outcome variables may be described and the primary outcome variable is not clearly specified
  
  • summary statistics are not supplied with the study results
  • the variability is seldom described in the form of a common standard deviation or may be given as standard errors

Question 9

What is the link between a type 1 and type 2 error?

Answer 9

when decreasing the risk of making a Type I error, the risk of making a Type II error is increased

Question 10

what does a low power indicate?

Answer 10

study is clearly underpowered
to detect the desired difference
> not a large enough sample to detect a true effect