How to say you found an effect
Reject the null hypotheses
How to say you couldnt find an effect
Fail to reject the null hypothesis
The ___ means the probability that a difference is due to random sampling error
P value
(Ie.0.046 = 4.6% chance)
Probabillity of making a type 1 error
Alpha
Probability of making a Type II error
Beta (B)
Ways to avoid false negatives (type II error):
More power!
Probability of correctly rejecting a false H0
Power
Overlap area under curve that is in between the critical errors
Beta (probability)
To reduce Beta overlap and increase power, you want to make the sampling distribution ___
Skinnier
Way to increase your power to make rejection region bigger that is not appropriate to ever do ____, because it increases your probability of a Type I error
Increase alpha level/signifiant level
A non-directional test is a ___ tailed test
2 tailed
__ tailed tests typically have more power
One
You can ____ population variance to increase power
Decrease
Reducing variability in sampling distributions would ___ the width of the curves
Decrease
You can make sampling distributions skinnier (less overlap/Beta), by
Reducing population variance
2 ways you can reduce population variance:
1) use a less diverse population
2) use repeated measures
The best way to increase power in your experiment
Increase sample size
Removing ___ can decrease population variance
Outliers
Increasing the magnitude of true difference between null & alternative hypotheses (changing the means, mu1-mu2) can also increase power and is called the ___ ___.
Effect size
Standardized effect size =
Mu 1 - mu 2 / SD
The 3 measure for effect size for a t-test (d), Cohen’s d:
Small - 0.2
Med - .5
Large - .8
