Effect Size and Power 04/15/2014
The size of the difference between two samples or populations being tested
Effect size does not depend on sample size (does not take into account standard error)
Contextualize scores based on variability
What is the “overlap” of scores?
Overlap distributions (when proportion of overlap is less, it shows a bigger effect)
The degree to which the null hypothesis is false, e.g. not just that they are different but how much they differ
Example: Dowson (2000)
Investigation of time of day effects on learning: do children learn better in the morning or afternoon?
38 children ▯ half answered questions about a story in the morning and the other half in the afternoon
Found that afternoon was better: 15.2 questions answered in morning vs. 17.9 questions answered correctly
How significant is this difference?
“Contextualize” the scores based on the variability
What is the “overlap of scores?
Are all scores in one group larger …
Distribution of means vs. distribution of scores
Means can be close to each other and a lot of distribution overlaps OR means can be close to each other
and only a little of the distribution overlaps
When proportion of overlap is less, shows a bigger effect
Cohen effect size test HeightIQ correlation: r = 0.11 (p