Experimental Power

Question 1

TYPE I ERROR

False-positve

Answer 1

Occurs if an investigator rejects a null hypothesis that is actually true in the
population
i.e. Null hypothesis is correct

• Critical value for our statistical test was 13.5 cm
• We rejected the null hypothesis that a new bird
  with a beak length of 15 cm came from the same
  population because 15 cm > the critical value and falls in the rejection region
• However, the probability α that we were wrong rejecting the null hypothesis was <0.05

Question 2

TYPE II ERROR

false-negative

Answer 2

Occurs if the investigator fails to reject a null
hypothesis that is actually false in the population
i.e. Null hypothesis is incorrect

• Usually the situation we like is to reject the null
  hypothesis and for the null hypothesis to be
  false (there really is a difference in the things we
  measured)
• The probability of getting it right and rejecting the null hypothesis when it is truly false is called the power of a statistical test, and it equals 1- β

Question 3

What is the power?

Answer 3

1. The variability of the data - the greater the variability the less precise our mean, and so the harder it is to detect differences between samples
2. Sample size (n) – the larger our samples the more precise our mean
3. The magnitude of the difference between sample means (the effect size) – we need a smaller sample size to detect a large difference (50% or 50cm) than a small difference (10% or 10 cm)
4. The Type I error probability (α), which is related to β.