Power

Question 1

What is power?

Answer 1

→ measure of sensitivity that detects a real effect
→ probability of correctly rejecting false null
→ prob. Of not making type 2 error

Question 2

What is alpha?

Answer 2

Probability of making type I error

Question 3

What is 1-alpha?

Answer 3

Probability of retaining null when it is true.

Question 4

What are the characteristics of null hypothesis?

Answer 4

① kurtotic
2 normally distributed
③ defined by u0 and sigma

Question 5

What are the assumptions of random sampling mod el of hypothesis testing?

Answer 5

①.SD is the same for Ho and HI distribution
② each distribution is unimodal and symmetrical
③ if no treatment effect→ alt dis = null dis (100% overlap)
→ only type I error is possible
④ treatment effect →alt dis not equal to null dis.
→ mean1 is diff from mean0

Question 6

When to test for power? (A priori option)

Answer 6

① are the features sufficient to detect effect(IV. On DV )
② expected effect size.
③ p(of detecting effect if it actually exists)
⑥ helps calculating N

Question 7

When to test for power → A posteriori

Answer 7

① retained Ho but wanted to reject
②rejected Ho and wanted to s practical sig and N
③ retained Ho and wanted to →power can be used to show if there was an effect it could be detected.

Question 8

What are 3 major factors affecting power?

Answer 8

① effect size → treatment effect and variability
②sample size

Question 9

What is effect size?

Answer 9

→ distance b/w means of population ① &②
→ degree to which exp manipulation separates H0& H1
→ larger effect size = smaller overlap
→ not impacted by sample size and type of measure.

Question 10

What 2 factors impact overlap?

Answer 10

→treatment effect
→ variability

Question 11

How to increase treatment effect?

Answer 11

→ not by increasing samplexize but by increasing for e.g. The dosage being given.
→ if the effect size is big there is no/smalles type 2 error

Question 12

What is the influence of variability?

Answer 12

By reducing v.de increase effect size

Question 13

How is variability reflected in SS?

Answer 13

When we deviate from. Sample mean and square it that gives us SS which is basically error (variability)

Question 14

What happens to effect size when ne reduce variability?

Answer 14

’ Increases

Question 15

How can we reduce variability in exp?( increase internal validity how?)

Answer 15

① monitoring experimental procedures
→ treating au participants equally
→ increases criterion reliability(DV)
→ reducing experimenter bias
② choice of research design
→ related samples D is a better choice b/c leas variability
→ more powerful.

Question 16

What is the rel b/w sample size & power?

Answer 16

→ sample size directly impacts standard errorwhich impacts test ratio
→ I hard to reject to null when n is small
→ but if n is too big then hard to believe credibility
But…
→ a big n might have a small effect size - not necessarily valid

Question 17

What is delta?

Answer 17

→ includes info from effect size & sample size → varies as a function of effect size and sample size