BIOS 501, Statistical Methods II (Accelerated)
INSTRUCTORS: Paul Weiss (lectures) Lisa Elon (computer labs)
GCR 308 GCR 336
Phone: (404) 712-9641 Phone: (404) 727-7335
FAX: (404) 727-1370 FAX: (404) 727-1370
LECTURES: T TH 3:00 - 4:20 pm Location: TBD
PREREQUISITES: Some calculus and BIOS 500, or permission of instructor
LABS: Day/Time Instructor(s) Room
T 4:30 - 6:20 pm Elon GCR P13
W 8:30 - 10:20 pm Weiss GCR P13
W 12:00 - 1:50 pm Elon GCR P13
(STUDENTS MUST REGISTER FOR ONE OF THE ABOVE LABS)
TEACHING Rizwana Khan, Anna Jolly, Megan Price, and Sameera Wijayawardana will also
OFFICE Paul Weiss: Tuesdays and Thursdays, 10-11 am, or by appointment
HOURS: Please e-mail at least 24 hours in advance in order to schedule a time with me. If a face-to-face meeting is not essential, ask questions on the class web site discussion board in order to get a faster answer (often in £ 24 hours.)
Teaching Assistants: See Class web site for hours.
Required: Weiss, N.A. (1995). Introductory Statistics (6th ed.). Addison-Wesley Publishing
On Reserve: D.G. Kleinbaum, L.L. Kupper, and K.E. Muller, A Nizam. Applied Regression
Analysis and Multivariable Methods, 3rd edition, Duxbury Press, 1998.
(1995) Fundamentals of Biostatistics
OPTIONAL TEXTS: Paul D. Allison. Logistic Regression Using the SAS System. SAS Institute Inc., 1999
Scott Menard. Applied Logistic Regression, 2nd Ed. Sage Publications
This course is the follow-up to Biostatistical Methods I. Students will apply many of the concepts learned in the first semester in a broader field of statistical analysis: model construction. Topics covered include Linear Regression, Analysis of Variance, Logistic Regression and Survival Analysis. Students who successfully complete this course will have a deep understanding of many analytical methods used by public health researchers in daily life.
OBJECTIVES Upon completion of the course, students will be able to:
1. Choose appropriate methods, models, parameters and hypotheses for a
variety of problems related to simple and multiple linear regression, logistic regression and survival analysis.
2. Run SAS programs and interpret SAS output from these programs.
3. Check assumptions underlying the regression methods.
4. Use related statistical tables.
GRADING A: 95-100 A-: 90-94
SCALE B+: 85- 89 B: 80-84 B-: 76-79
C: 66- 75
F: Below 66
EVALUATION Labs: 20%
Exam I: 10%
Midterm Paper: 25%
Final Exam: 35%
SITE The web site will be our primary means of communication outside of class.
- Instructors will post announcements on the web site.
- You may post questions, and instructors will then post answers, on the discussion boards located on the class web pages.
- Class notes will NOT BE posted on the web site. Students who have to miss classes for any reason should arrange to get the notes from another student. The professor is NOT RESPONSIBLE for students missing lectures.
- The following will also be posted: homework assignments and solutions, practice exams and solutions; lab notes, homework and solutions.
- You are expected to access the web site daily, M-F, to look for newly posted documents and discussion board messages. It is suggested that you also check the site on weekends, for updates to posted documents and other important announcements.
- It is recommended that you subscribe to the bios501 listserv. Listserv subscribers will receive e-mail notification each time a new announcement or document is posted on the class web site.
The site is password protected. Use the same ID and password for first semester. If you do not have an ID and password, please see the instructor.
- Please attend all lectures and labs in order to avoid falling behind. It will be difficult to fully understand new material being presented in class and labs if you fall behind. We realize that many of you are working full or part-time, and that your jobs require you to travel out of town occasionally. Please arrange your schedule in such a way that class and lab absences are kept to a minimum. If you do miss a class or lab, it is your responsibility to catch up, and you should do so immediately.
- Each week, you will be expected to read the appropriate sections of the textbook (Weiss) in advance of class. Check pages 5-6 of this syllabus to determine the appropriate textbook pages to read. To get the most out of the readings, you should do them at least twice: once before the appropriate lecture, and again after. In most cases, the reading serves to reinforce ideas taught in class. Occasionally, however, the reading will complement the lab and lecture material; you will still be responsible for this material on homework assignments and exams.
You might find it
helpful to read Kleinbaum’s, Allison’s and Rosner’s explanations where
- Use the web-page discussion boards and specified office hours freely. As soon as a course-related question arises, post it on the appropriate web site discussion board. If the question is more private in nature, e-mail the instructor directly (firstname.lastname@example.org for class-related or lab-related questions), or get in-person help from the instructor or a TA during the next available office hour. Ask questions far in advance of quizzes and exams!
- The homework for this class will be helpful in preparing you for the midterm paper and exams. As was the case last semester, this homework will not be collected on a weekly basis. Also, as was the case last semester, solution sets will provide only the answers. It will be the responsibility of the students to attend office hours and review sessions to obtain more complete solutions.
- It is highly recommended that you attempt all of the homework provided in class. These homework sets will serve as guidelines in helping you prepare for the exams, and also help you in formulating analyses germane to the midterm paper. Many of the homework sets will feature open-ended questions, designed to get you to think analytically, and may not have a single correct answer.
- You are encouraged to work together on reading assignments and homework. Sharing ideas in this way may lead to a better understanding of the material.
- Lab homework will count for 20% of the course grade. Lab homework will be assigned and collected on a weekly basis. Each assignment will be graded in detail. If you do not have time to complete the assignment during the lab itself, you should complete it later and turn it in during your next lab, one week later. A grade of 0 will be recorded for late labs. After grading, the labs will be returned to you in lab. Answer keys will be posted on the BIOS 501 lab web site.
- This class is much more computationally intensive than was BIOS 500. To this end, in-class exams could be much more cumbersome to complete in the time allotted. I have divided the term into three sections: Simple Linear Regression, Multiple Linear Regression and Model Selection, and Logistic Regression and Survival Analysis. Upon completion of the first section, an in-class exam will test students on the theory and application of simple linear models. The second section will culminate in a paper where the student will present a methodology for using multiple linear regression to answer questions about a data set. The final exam will have a take-home component as well as an in-class component.
- In-class exams will be closed-notes/closed-book. The professor will provide a sheet with the necessary formulas, as well as copies of the necessary tables. For the final exam, you will need to bring the output from your take-home component to answer questions on the in-class component. This output should be titled and footnoted for ease of identification, but you are not allowed to bring in any outside note sources to the exam. Also, bring your calculator.
- Exams will be held from 3-6pm on the days specified on pages 5-6 on this syllabus (dates are tentative). The three-hour time period is allowed so that time pressure will not be a factor. If you have a course-related or unavoidable personal conflict with the exam time, please e-mail Paul ten days before the exam --another exam time for the same day will then be arranged.
- Please do not schedule out-of-town trips or other appointments on exam and quiz dates, since it is often difficult to schedule make-up exams. In case of emergency or unavoidable conflicts, we will, of course, do our best to accommodate you. Keep in mind that make-up exams are usually given the day before the actual exam date.
- Prior to exams, you will be required to conduct data analyses using SAS. The output that you produce may be needed to answer questions during the in-class exams.
- Failure to take any exam will result in a course grade of 'F'.
- Review sessions will be held before each exam.
- Students are NOT allowed to work together on exams, or on any work that is related to quizzes or exams. All quiz and exam-related work must be done completely independently and must not be discussed with anyone except Paul Weiss.
- The midterm paper will be a data analysis project, due after the mid-semester break. This paper will not exceed five (5) pages, including graphs, tables and supplementary material. More information on this project will be provided after the simple linear regression quiz.
BIOS 501, SPRING '03: TENTATIVE LECTURE OUTLINE
Date Topics Reading
1/20 Overview of course, introduction to covariance and correlation W2: 14.4
R1: 11.1, 11.7-11.8
1/25 Simple Linear regression: general concepts, method of least squares, W: 14.1 – 14.3, 15.2
hypothesis tests on parameter estimates R: 11.2 – 11.4
1/27 ANOVA table for SLR: F-test inferences on parameters and the
coefficient of determination R: 11.4
2/1 Confidence intervals for parameter estimates and predicted values W: 15.2 – 15.4
2/3 Checking the assumptions for regression (Regression Diagnostics ) W: 15.1, 15.5
2/8 More Regression Diagnostics W: 15.1, 15.5
2/10 Multiple regression: Introduction, general concepts, hypothesis tests; W: Module A
Post 2nd exam take-home portion on web (CD-ROM in book)
2/15 Catch-up and Review
2/17 Exam I: Simple Linear Regression. 3-6pm.
2/22 ANOVA for Multiple Regression: F-test inferences; the extra sum-of- R: 11.9
squares principle; interactions
2/24 Multiple and Partial F-tests for multiple regression R: 11.9
3/1 Comparing regression results for two or more groups: dummy
variables, coding schemes. R: 11.8
3/3 Selecting the best regression equation: stepwise regression K: 16.1 - 16.6
W: Module B; B5
3/8 Selecting the best regression model: All-possible-regressions K: 16.1-16.6
W: Module B; B6
3/10 Catch-up, Review, and Questions for Midterm Project
3/14 – 3/18 Spring Break (No classes)
1. R: Rosner
2. W: Weiss
BIOS 501, SPRING '03: TENTATIVE LECTURE OUTLINE (Continued)
Room Topics Reading
3/22 One-way ANOVA: Intro; general concepts. Inferences on main effects. K:17.1 - 17.5
3/24 One-Way ANOVA: An Example
Paper due at 3:00 pm. No late papers will be accepted.
3/29 One-way ANOVA: multiple comparisons K: 17.6 – 17.8
3/31 One-way ANOVA: multiple comparisons (Contd.) R: 12.4
K: 17.7 - 17.8
4/5 Special topics in linear modeling
4/7 Logistic Regression: Review of 2x2 tables; odds ratios; logistic K: 22, 23.1-23.4
regression model, maximum likelihood estimation. R: 13.7
4/12 Logistic regression: odds and odds ratio (OR) calculations K: 23.1 – 23.4
4/14 Logistic regression: analysis strategy, CIs for odds ratios, prediction and Same Reading as 3/28
4/19 Interactions; interpreting odds ratios in the presence of interactions
4/21 Logistic regression wrap-up. K: 17.1 - 17.5
W: Module C
4/26 Survival Analysis: introduction, Kaplan-Meier estimator R: 14.8-14.9
4/28 Catch-up and review
5/3 Exam III. Covers ANOVA, Logistic Regression, and Survival Analysis