Time and Location: 1:30-3:30pm on Wednesday, December 11, in Eck133
- Bring a calculator
- Bring two letter-size formula sheets, two-sided
- a normal table, a t-table and a chi-square table will be provided in the exam
Coverage: Section 5.1, 5.2, 6.1, 6.2, 6.3, 6.4, 7.1, 7.2, 8.1, 8.2, 9.1, 9.2, 10.1 (see the study guide below for details)
Study Guide for the Final Exam
Section 5.1-5.2
- CLT
- Binomial Formula
- When is Binomial Formula applied?
- CLT for Binomial distribution (np >=10, n(1-p) >=10)
- Skip "Continuity Correction" on p.327
- Skip "Weibull Distribution" on p.330-331
Section 6.1
- How to make a confidence interval
- Interpretation of confidence intervals
- Sample size calculation
- When can one NOT use a confidence interval (e.g., not SRS). Read the caution on p.354
- Skip "Bootstrap" on p.355-356
Section 6.2
- Hypothesis testing: hypothesis (null, alternative, one-side, two-side), test statistics, p-values, significance level, conclusion
- Interpretation of P-values
- Relationship of a two-sided test and confidence intervals (p.373-375)
Section 6.3
- Significance is not importance
- Hypothesis testing cannot tell us if the data is properly collected (e.g., if the experiment or the survey is done properly)
- Beware of searching of significance
Section 6.4
- Definition of Type I error, type II error, and the power of a test
- How to increase the power of a test?
- Computation of the power of a test
Section 7.1
- t-distribution
- One sample t-test, t-interval (check for skewness and outlier before using t-procedures)
- matched pairs t-test
- Skip p.419-425 (power of the t test, non-Normal populations, ?sign test)
- Skip Exercise 7.45-7.53 on p. 431-432
Section 7.2
- Two sample t-test (unequal population SDs)
- pooled two-sample t-test (equal population SDs)
- Check for skewness and outliers before using t-procedures
- Skip the software approximation for the degrees of freedom on p.445
Skip Section 7.3
Section 8.1
- large-sample C.I. for a single proportion (np >= 15, np(1-p) >= 15)
- Wilson's Plus-Four C.I.for a single proportion (np >= 10, np(1-p) >= 10)
- Test for a single proportion (note the SE is different from the SE for a C.I.)
- Choose a sample size to achieve a specific margin of error
Section 8.2
- large-sample C.I. for the difference of two proportions
- Wilson's Plus-Four C.I.for the difference of two proportions
- Test for the difference of two proportions (note we used the pooled SE here)
- Skip Relative Risk on p. 500
Section 9.1-9.2
- Chi-square test for two-way tables
- Expected cell counts
- a chi-square test for a 2x2 table is equivalent to a two-sided test for the difference of two proportions
- Skip Meta Analysis on p. 520
Skip Section 9.3 (and also skip Exercise 9.49-9.55 on p.543-544)
Section 10.1
- subpopulation (comparison of several populations)
- Assumption of simple linear regression model
- the population regression line v.s.. the sample regression line
- the LS estimate for the intercept and slope are unbiased estimates for the population intercept and slope
- SE for the intercept and slope
- How to reduce the SE of the slope?
- estimate for sigma?
- Use R summary output to find C.I.s and perform t-tests for the intercept and slope
- Prediction intervals v.s. confidence intervals