Power Analysis Flashcards
What is statistical power?
Statistical power is the ability to correctly reject the null hypothesis.
high power = high avoidence of type _ errors.
Type _ error = B
Power = ?
Type 2 = B
Power = 1-B
What did Cohen (?) set the power limit at?
power be at least 0.8 as a lower value results in too greater risk of a type 2 error (Cohen, 1992)
Power is determined by what 3 things?
Sample size, effect size, significance criteria
What can post-hoc power reveal
It can reveal whether the reason a null hypothesis was not rejected because the study had inadequate power
Post-hoc power can lead to the belief that the results portray a false negative when in fact…
Levine & Enson (2001): it was a true negative
The fallacy of classical inference?
Friston (2012) - If a study is overpowered with too many PS, even the smallest effect size will appear significant which is not really meaningful from the null.
Optimum number of PS [for neuroimaging] is 16-32
- if one finds a significant effect with a small sample size, it is likely to have been caused by a large effect size.
observed power is determined by observed p-value, therefore….
a non-sig p value will always give rise to low post-hoc power.
Hoenig & Helsey (2001):
for interpretation of observed results, the concept of power would be ill-used
Goodman & Berlin (1994):
