Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
11 Cards in this Set
- Front
- Back
Effect size
|
a standardized difference between the means of two distributions
|
|
D
|
effect size is represented by what symbol?
|
|
Power
|
How likely are we to do what we want to do? (ie, correctly reject a false null)
|
|
Power
|
How much overlap is there between distributions for the null and alternative hypotheses?
|
|
A use of power
|
Given that we have a fixed alpha, what technique can we use to determine the required sample size to reject a specific alternative hypothesis? (ie, How big would N have to be so that if the alternative hypothesis were true, the null would be rejected 95% of the time?)
|
|
alpha
|
the probability of rejecting a true null
|
|
1- Beta,
power |
the probability of rejecting a false H0
|
|
What are the three factors that affect power?
|
effect size, sample size, significance level (one vs. two tailed)
|
|
specify effect size, know H1
|
We can only compute 1-B if we specify/know...
|
|
1. reduce the population variance (by using a matched samples or dependent samples design)
2. Increase alpha (your definition of unlikely). 3. increase the effect size 4. use a larger sample size |
What are ways to increase power?
|
|
A value for the alternative hypothesis must be specified. (in indep samples t test the difference betw means is just > or <).
|
What is the unique thing about power computations and the alternative hypothesis?
|