Power

Question 1

Power

Answer 1

-the ability of a study to detect a difference between two groups if such difference truly exists

Question 2

Beta and power

Answer 2

-beta (type 2 error) is set to 0.2 and power is set to 0.8 (80%)

Question 3

What affects power?

Answer 3

-depends on various aspects especially sample size (n), mean effect difference (effect size), variability of observations (lower the variability the higher the power) and acceptable level of statistical significance (p)

Question 4

Power calculation

Answer 4

-misnomer, actually sample size calculation using set values of alpha ( p value, generally 0.05), power (1-B; generally 80%), absolute effect size which the researcher considers as clinically relevant and an estimated measure of variance

Question 5

Standard difference

Answer 5

-expression of effect size given by (target difference in means/ SD of observations)

Question 6

Methods to increase power

Answer 6

• use a larger significance level (higher p value) but this may increase type 1 error
• use a larger sample size (more expensive but will reduce variance)
• use larger effect sizes (consider only larger deviations from null hypothesis to be significant)
• reduce variability (match subjects or make more precise measurements)
• one sided test is more powerful than two sided test