B3.044 Prework Type 1 & 2 Errors and Sample Size Flashcards
what is a type 1 error
when you reject the null hypothesis, but you shouldn’t have (say groups are different but they’re not)
reason why study results are replicated by others
alpha
probability of making a type 1 error
what is a type 2 error
when you don’t reject the null hypothesis, but you should have (say they’re the same, but they’re different)
often occurs due to inadequate study power
beta
probability of making a type 2 error
how do you best avoid error?
minimize methodological error by careful study design and protocol adherence
what does an alpha = 0.05 signify
investigator is willing to accept a 5% risk of falsely concluding the groups differ
related to concept of statistical significance and p<0.05
what does a beta = 0.2 signify
investigator accepts a 1 in 5 chance of missing a true difference between groups
sample size
how many people are needed to study in order to detect a difference between treatments if it exists
power
probability a study can detect a true difference between groups
(1-beta)
if B=0.2, 80% chance of detecting specified difference if it exists
what are the 4 steps in calculating sample size
- determine study design, measures of frequency, measure of effect, statistical analysis method
- set acceptable levels of risk of a type 1 (alpha level) and type 2 (B level) error
- determine magnitude of difference In outcomes between 2 groups that the study would be designed to detect
- calculate sample size requirement
larger differences require _______ sample sizes in order to be detected
smaller
smaller differences require _______ sample sizes in order to be detected
larger
reduced risk of type 1 error results in a ______ sample size requirement
larger
reduced power results in _______ sample size requirement
smaller
