power, effect size and sample size lec 8 Flashcards
why do we want more power
to detect an effect to correctly reject null hympothesis
what is power
the probability of avoiding a false negative (type 2 error)hat
what is the minimum power aimed for
0.8
why is low power bad
bad science (likely to produce false negative) and unethical ( wasting time and money / writing off)
what variables increase power
larger sample size
looking for larger effects (bigger the effect, more likely to find it)
how do you work out effect size
cohens d = (mgroup1 - mgroup2)/ sd pooled
what did choens d say the effect sizes are
0.2 = small
0.5 = medium
0.8 = largew
what does effect size tell us
how many SD the means are apart
how to work out how many participants needed
prospective power analysisw
what is prospective power analysis
calculates the minimum sample size
what is needed to know for the prospective power analysis
anticipated effect size and desired power level
what is the anticipated effect size
do we think it is going to be small med large?
what can work out this
gpower
what factors effect power
sample size
effect size
experimental design
a level
what design has the most power
repeated measure due to no of participants per conditon
