Chapter 6 Flashcards
(12 cards)
Type 1 Error
A study observes an APPARENT associoation between exposure and outcome when in truth there is NO association
Type 2 Error
A study observes NO association between the exposure and outcome when, in truth, there IS an association
3 main explanations for type 1 or 2 errors
- Chance due to RANDOM sampling error
- BIAS in sampling or measurement
- Confounding factors
Best way to reduce random sampling error
Increase the size of the sample as much as practical
What is a Null Hypothesis
Ho is a statement that there is no association. It’s the DEFAULT hypothesis that is either accpeted or rejected
What is the ALTERNATIVE hypothesis
Saying there is an association. Happens when you reject the NULL and accept the alternative
What does the p-value tell us
Tells us how probable it is that the observed association did not happen by chance.
What does p-value reflect
The strenght of the association and the sample size
The larger the sample size the _____ p-value will be
Smaller. Because your more certain that the data reflects the true population
What is a Confidene interval & what does it tell us?
CI is an admission that the observed effect estimate is probably not exactly right and the real answer is likely somehwere within the CI range of 95%
If we repeat the study 95% sure the data falls back with in the 95% CI range
IF the CI crosses the line of no effect (1) then
can’t say the data has effect. eg. CI of (0.9-1.9) crosses line so can’t say its effective
