Stat 322 - Final Review

Office Hours: Monday, 11-12 in my office, 1-3 in 02-206, also by email

Final Exam Time: Tuesday, 7-10pm, in the studio (also have seats Wed 1-4pm)

Format of Final Exam: Similar to the midterms, the exam will be open book and notes. You will have access to Minitab and should bring a calculator. Remember, “open book” means you still need to organize your notes and highlight key ideas for quick reference. The exam will be cumulative. I recommend studying from homeworks, quizzes, class notes, and your text. For earlier material, I would recommend focusing on understanding the big ideas (sampling distributions, parameter vs. statistic, p-values, type I and type II error, identifying appropriate inferential procedure, effect of sample size, how to verify technical conditions, meaning of “confidence”).

For Chapter 14, you should be able to

· Decide when chi-square procedures are appropriate to use

1. multinomial experiment – goodness of fit (e.g., Reeses colors)

2. homogeneity of several population proportions or distributions

3. association between two categorical variables (two way tables)

· Set up a two-way table

· Describe conditional distributions for cases 2 and 3, including segmented bar graph. What does this graph look like if the two variables are not related?

· State appropriate null and alternative hypotheses

· Find expected counts

· Check technical conditions

· Calculate chi-square value by hand and in Minitab

· Determine degrees of freedom for chi-square distribution

· Make decision (e.g., reject or fail to reject) using the p-value

· Carry out and discuss “follow-up” analysis examining components in chi-square sum

From Chapter 16 (16.1-16.4) you should be able to

· Understand the reasoning behind the “3-s” rule

· Know how to compute control limits (by hand and with Minitab)

· Know how to interpret control charts (for mean, standard deviation, range, proportion, count)

· Be able to decide which chart to use based on variable measured

Some Big Picture Ideas

· Simulation of sampling distribution of test statistic, empirical p-value

· Scope of conclusions (population generalize to, when draw cause and effect)

· Comparisons vs. Association

· Examine visual displays before applying inference procedures

· Understand goals/limitations of confidence intervals and tests of significance

· Effects of various components (sample size, sample variability, confidence level, effect size)

· How to measure, explain, reduce variability

· Be comfortable with Minitab output

Recommended problems

1) p. 641, problem 5

2) p. 662, problem 32

Include numerical and graphical summaries, state your conclusions in context, including consideration of issues of generalizability and causation

3) p. 666, problem 47

4) p. 708, problem 5

5) p. 708, problem 7

6) p. 708, problem 12

7) p. 717, problem 19

8) p. 717, problem 23

Some Notes to Keep in Mind

· Don’t claim you have an “SRS” if you don’t. You might still want to convince us your sample is representative, but that’s not the same as having actually collected a simple random sample.

· With a paired t-test, the appropriate graphical summary is a graph of the differences, not the graphs of the individual groups

· Don’t forget to put hats and bars when appropriate.

· Make sure your interpretation of a confidence interval for a mean doesn’t sound like a prediction interval.

· To be able to use the standard deviation formulas we know and love (e.g., ), we really need our population size to be more than 20 times the size of the sample.

· Make sure with Goodness of Fit hypotheses that you are setting the proportions equal to specific numerical value(s). For example, in the equally likely case, they aren’t just equal, they are all equal to 1/number of categories.

· The less conservative technical conditions for the chi-square tests for two-way tables larger than 2 × 2 are: average expected count > 5 or at most 20% of the expected counts < 5.