Logic of Frequentist Statistics

Question 1

Null and alternative

Answer 1

-Comparison, what we are testing
-Null- no effect in the pop
-Alternative- there is an effect in the pop

Question 2

Alpha

Answer 2

-0.05
-Sets numeric standard to compare to p-value
-Probability of rejecting H0 when its true

Question 3

P-value

Answer 3

-Probability of the observed result, plus more extreme results, if H0 were true
-Decides if we fail to reject or reject H0

Question 4

P-value interpretations

Answer 4

-If p-value < 0.05, effect is significant and reject H0
-If p-value > 0.05, effect is not significant and fail to reject H0

Question 5

Sample size and p-value

Answer 5

Larger sample sizes give smaller p-values, so with more data is it more likely to find a significant effect

Question 6

Effect size

Answer 6

Magnitude of an effect in a standardized unit

Question 7

Cohen’s d

Answer 7

-Effect size
-d=0.8, mean of 5 is 0.8 SD away from mean of 10

Question 8

Confidence intervals

Answer 8

-Plausible range of values based on theoretical samples that would have intervals that contain the parameter
-Wide intervals = more uncertainty
-Tight intervals = less uncertainty

Question 9

Limitations of frequentist statistics

Answer 9

-Null is always quasi false
-Non-intuitive
-False confidence

Question 10

Probability in frequentist statistics

Answer 10

Long run relative frequency of an event occurring beyond chance level