Chapter 12: Power and Power calculations Flashcards
Alpha limits the probability of making a ___ error which means?
- type 1
- rejecting Ho and when it is true
Fill in the state of reality table with the correct formulas:
State of reality
Decision: Ho true | Ho false
Retain Ho| Correct D | Type 2 error
Reject Ho| Type 1 error | Correct D
A) 1- alpha
B) Beta (B)
C) alpha value
D) 1- Beta (B)
What does Beta limit the probability of ? What does that mean?
- limits the probability of making a type 2 error, retaining Ho when it is false.
What is power?(3)
- sensitivity of an experiment to detect a real effect of the IV if there is one.
- Probability that results of an experiment will allow the rejection of Ho if IV has a real effect
- Probability that results of exp will allow rejection of Ho if Ho is false.
What is powers range?
- 0.00 to 1.00
Is a power of 0.80 common in behavioural sciences?
What about 0.40-0.60 range?
- NO
- common
What are two things one must consider before an experiment?
- power analysis
- based on previous research
So in laymen terms what does power do?
- calc the prob of detecting a real effect/difference
Power varies directly with ___ and ___ but inversely with ____. (explain the blanks as well)
- Sample Size: more N= more power
- Large effect= more power
- alpha level- more stringent alpha = less power
Small effect size vs large effect size?
- small: distance between is small
- = distance between is large
Small sample size vs large sample size?
- small: increased variability, less precision(hard to tell the difference)
- large: decreased variability, more precision (easy to tell a difference)
What is the formula for power?
1- B (beta)
What are the two steps involved in computing power?
hint: sample outcomes, sample mean *
- Determine possible sample outcomes that would allow Ho to be rejected
- Determine probability of sample mean in critical region if hypothesized real effect of IV is true
- Resulting probability = power
In power questions you will be given what info? In step one calc.
What must you calc from it?
- N (sample size)
- Mean of null-hypothesis population ( Unull)
- Standard deviation of null population (you will have to calc the standard error)
- Alpha
- -> calc standard error and the Xcrit (rearrange z score formula to solve for it)
- b/c you will know Zcrit from the apha
Conclusion from calculating power???
If N= X, reject Ho if Xobt> Xcrit