Power Analysis L5 Flashcards
Learning objectives
Explain the concept of statistical power and its relationship with Type I and Type II error.
Describe the effect size and ways to calculate it.
Understand how to select the sample size.
How do we make a descision when the F ratio exeeds the F critical?
we reject the null hypothesis.
Independent variable is related to the dependent variable.
Result: Statistically significant effect.
How do we make a descision when the F ratio does not exeed the F critical?
we fail to reject the null hypothesis.
H0 : Means are equal; no evidence that IV is related to DV
Result: No statistically significant effect
What does “no statisticaly significan effect” mean?
we are not sufficiently confident (because of uncertainty about the population means) that the difference(s) we observe generalise to the populations we are interested in
Failure to find a significant effect does not necessarily mean the means are equal
What is statistical power?
The probability of detecting a true (population) effect given a particular sample
or
In other words, statistical power is the probability of not making a type-II error
Why doesnt “No statistically significant effect” not prove the null hypothesis?
It is possible there is a difference in the population, but our samples were too small and/or noisy to reflect this
What is power defined as?
1 - beta (where beta is the probability of making a type one error)
What are the three factors that determine the power of a study?
Alpha level (the threshold you set for uncertainty about an effect, e.g., a = 0.05)
* Sample size (n)
* Effect size (tells how meaningful is the difference between groups)
What does effect size indicate?
Effect size indicates the practical significance (or magnitude) of a research
outcome.
What are measures of association in effect size?
- Eta-squared (h2)
- R-squared (R2)
What are the measures of difference in effect size?
Cohen’s d