chapter 11: one sample confidence interval

Question 1

4 criticisms of Ho testing

Answer 1

1. it doesnt tell us if results are important or useful
2. Ho tells us p(D⎮Ho) but we want to know p(Ho⎮D)
3. it is a trivial exercise (john tukey) you can always find a difference to some decimal place if you have a large n
4. 𝛂= .05 is arbitrary. a continuum of uncertainty becomes a reject-do-not-reject decision
   - “surely god loves the .06 nearly as much as he loves the .05” (rosnow and rosenthal)

Question 2

what is a confidence interval?

Answer 2

a confidence interval is a segment or interval on a number line such that 𝛍 has a high probability of lying on the segment
*can be either 1 or 2 sided

(a type of interval estimate of a population parameter)

Question 3

what is confidence coefficient?

Answer 3

confidence coefficient is the probability that the interval contains 𝛍

Question 4

when is a one sided confidence interval constructed?

Answer 4

a one sided confidence interval is constructed when the researcher has made a directional prediction about the population mean; otherwise the researcher constructs a two sided interval

Question 5

purpose of calculating a confidence interval?

Answer 5

to provide a range of values that, with a certain measure of confidence, contains the population parameter of interest.

Question 6

the 3 assumptions

Answer 6

1. random sampling
2. population is normally distributed
3. population SD is unknown

Question 7

interval estimation vs. hypothesis testing

Answer 7

confidence interval provides more information that the null hypothesis test

it can also be used to test any null hypothesis for 𝛍 simply by looking at the interval

Question 8

effect magnitude statistics

Answer 8

a measure of the practical significance of results

1. measures of effect size
2. measures of strength of association

Question 9

what is hedges g

Answer 9

hedges g is an estimator for cohens effect size parameter, d

• the g statistic is interpreted the same as cohens d
• g can be computed from a one sample t statistic