11 - Effect Size And Power

Question 1

Effect size

Answer 1

The extent to which two populations do not overlap

Question 2

Importance of effect size

Answer 2

• summarise a series of experiments with the same IV and DV
• is independent from sample sizes, different means and variances
• complement test of statistical significance
• rank several IV within the same experiment
• help us calculate the sample size needed for a study

Question 3

Level of effect size

Answer 3

Small d=0.2 (85% overlap)
Medium d= 0.5 (67% overlap)
Large d=0.8 (53% overlap)

Standardised mean difference.
d= (M1-M2)/SD

Question 4

The larger the difference between two population means

Answer 4

The greater the defect size

Question 5

The smaller the variance within two populations

Answer 5

The greater the effect size

Question 6

The greater the effect size

Answer 6

The greater the power

Question 7

Statistical power

Answer 7

The probability that the test will correctly reject a false null hypothesis

Question 8

Factors that influence power:

Answer 8

• the bigger the sample, the bigger the power
• changing the design of a study can increase the power to detect significant differences
• the smaller the SD the greater the power (more stable conditions, precise measures, less diverse pop)
• use less stringent level of significance (low power 1%, big power 10%)
• use one tailed test

Question 9

Low power means

Answer 9

That event is H1 is true, this study is not likely to give significant results in support of it.

Question 10

Effect size d formula

Answer 10

(M1-M2)/SD

Question 11

Within group SD

Answer 11

Tells us about the degree of error dur to individual differences

Question 12

Decrease from 0.05 to 0.01

Answer 12

We decrease the chance of type 1 error, rejecting the null hypothesis when it’s true but we increase the probability of type 2 error, accepting the null hypothesis when it’s false.