Sample size + power

Question 1

study that’s too small

Answer 1

o Scientifically - Cannot detect clinically important effects
o Economically - wastes resources without capability of producing meaningful results
o Ethically - Exposes subjects to risks (potentially harmful treatments, inconvenience, loss of privacy and confidentiality) without advancing knowledge

Question 2

Study that is too large

Answer 2

o Scientifically - demonstrates scientifically (clinically) irrelevant, but statistically significant effects
o Economically - wastes resources by using more than necessary
o Ethically: Exposes unnecessarily large number of subjects to risks and/or denies some subjects potentially benefits

Question 3

power is probability of

Answer 3

rejeting null when there is some effect

Question 4

power is affected by

Answer 4

variation in outcome (less variation greater power)
significance level (greater significance greater power)
difference (effect) to be detected (greater effect greater power)
one tailed vs two tailed (power is greater in one tailed tests

Question 5

what should power be

Answer 5

0.8 or 0.9 by convention

Question 6

power can be called ____

Answer 6

true positive