Sunday, February 27, 2011

Notes from Basic Statistical Practice 2: Sample Size

Notes from Basic Statistical Practice 2: Sample Size

Power of a test

-          Ability to find a difference in distribution when there really is a difference.

-          Increase power of a test

o   Increase size of alpha

§  If using 0.05, increase to say 0.01

§  Reduces risk of type 2 error, increases chance of type 1 error

-          Reduce beta without increasing alpha

o   Be more specific in prediction

o   One tailed test more powerful than 2 tailed test

-          Reduce overlap between 2 distributions

o   Beta is reduced

o   Difference between the means

o   Size of std deviation.

 

-          Look for strong, rather than subtle effects

-          Improve sensitivity of the measure of dependent variable

 

-          Planning sample sizes with power approaches – what sample size is required to achieve given power

 

-          Different formula for factorial sample size

 

-           

No comments: