Chapter 17

Question 1

dichotomous decision-making

Answer 1

null hypothesis significance tests can result in one of only two possible decision – either we reject H0 or we retain it -> confidence intervals provide an alternative to this

Question 2

2 forms of estimates

Answer 2

• Point estimates: single numerical value that estimates population parameter
• Interval estimates: a range of values that estimate a population parameter (more precise)

Question 3

basic logic behind confidence intervals

Answer 3

• In a normal distribution, 95% of sample means fall within 1.96 SD away from the population mean -> therefore, we can also say that for 95% of the sample means, the population is no further than 1.96 SD’s away -> basis of confidence intervals
• 95% fall within 1.96, 99% fall within 2.58

Question 4

confidence interval

Answer 4

• limits of the interval estimate (ie. Upper and lower limits)
• After constructing a 95% CI, you can say you’re 95% confident that the population mean falls within the upper and lower limits
• The interval varies from estimate to estimate, but the population mean does not vary
• Small samples will give a wide CI and large samples will give a small CI

Question 5

width of a CI

Answer 5

range covered by the interval (distance between upper and lower limit)

Question 6

advantages about CIs

Answer 6

• Final answer is a statement about parameter of interest
• CI shows influence of random sampling variation and sample size
• CI’s make it harder to confuse a statistically significant difference with an important one
• Emphasizes a range of values that might characterize parameter in question (rather than the “yes or no” of hypothesis testing)

Question 7

confidence coefficient (C)

Answer 7

degree of confidence (ie. 95%)

Question 8

ways to change width of CI

Answer 8

• If n is quadrupled, the standard error of the mean will be cut in half, the df changes, the t-value is smaller -> CI’s will be slightly less than half as wide
• If you cut the standard error of the mean in half, you’ll also cut the width/range of the CI in half
• When Ox is unknown and Sx is being used instead, the .95 CI requires a slightly higher multiplier (CI’s need to be winder) to account for error in Sx that we don’t have with Ox
• Increasing n reduces width of CI