- Last Time: Using the
t-distribution for estimation and hypothesis testing
- Major issue: sample standard
deviation (s), rather than population value (sigma).
- Major twist: degrees of freedom is now required.
- Assure that the use of t is appropriate by checking conditions for
normality (e.g. sample size, histogram)
- Notes:
- Recall that we have a test coming up next Thursday.
- Chapter 12 homework (over Section 12.1 and 12.2) will also be due
on Tuesday, but chapter 12 will not be on the test.
- Today:
- Return Homework for Chapter 11
- The hypotheses are not about action: they are about reality! Either the person on trial is guilty or innocent: we don't consider other alternatives. Either the PhD could happen or it couldn't.
- In problem 49, you need to understand what the numbers tell you mean: is the safety training working, or not? That translates into 0 (no change) versus a change to fewer (mu < 0).
- Review #12.32.
We dealt with this one last time, but let's have another look:
- test of hypothesis, using Minitab: Stat -> Basic Statistics ->
1-sample t
- Section 12.4: Proportions p and associated C.I.s
- recall from 9.3 (The sampling distribution
of a proportion) the test statistic for p
- Confidence interval estimator of p (when p is
unknown!)
- We can try to use p hat (the estimate of p): #12.84, p. 380
- Conservative standard error: use p hat=.5: #12.84
- Nielson
example (recode, using Data->code->numeric to numeric: 1 -> 1, 2:5 -> 0)
- #12.91
- Sample size calculations for p:
- Use p hat again!
- Conservative standard error: use p hat=.5: #12.84
- Next time: Review for Test 2; Problems for
chapter 12, section 1 and 2 due next time.
Website maintained by Andy Long.
Comments appreciated.