This class provides an introduction to categorical data analysis, including inference for proportions, analysis of contingency tables, trend tests, logistic regression, matched     pairs analysis, polytomous response, ordinal data and classification trees.  Applications in biological, biomedical, epidemiology and social sciences will be presented.  Students will use R for statistical analyses.

Lecture: Tuesday and Thursday, 1:00 – 2:15 pm , Location: Gates Hall 114                Prerequisites: BTRY 3020 or 6020 or equivalent.  BTRY 3080 highly recommended.

Instructor:                      Melissa Smith, Lecturer, Department of Statistics and Data Science

295 Ives Hall              [email protected]

Office Hours: TBA, See “Course Essentials” module in Canvas. I am also available at other times, by appointment.

Attendance:

Attendance at lectures is strongly encouraged, but not required. You are responsible for being aware of the announcements and content.

Website:

The course website is available through Canvas. We will use Ed Discussions also. Please register with the site and check it regularly. This is a good venue for getting answers to questions that you have about the homework assignments.

Text:

An Introduction to Categorical Data Analysis (3rd Edition) (2019) 3rd  by Alan Agresti. The 3rd edition is available through the Cornell bookstore ebooks. This is an excellent book  with many good examples. The course matches the book fairly closely. Homework          assignments may contain problems from the textbook.

Learning Outcomes:

After completing this course, undergraduate and graduate students should be able to: 1. Perform inference for proportions, including the binomial test, the Wald test and     Fishers exact test. Also, calculate confidence intervals for a proportion.

2. Summarize patterns of association for two categorical variables, using an odds ratio, relative risk, chi-squared tests of association.

3. Summarize patterns of association among binary and multi-category response data using logistic regression models and multinomial response models.

4. Be able to calculate odds and predicted probabilities from logistic regression models and multinomial response models.

5. Build and interpret models for matched pairs analysis, choice modelling and CART.

6. Be able to calculate measures of agreement for dependent categorical variables and test hypotheses about marginal homogeneity.

7. Be able to clean and analyze a dataset with multiple predictor variables and a    categorical outcome. Be able to do one-way and two-way analyses, as well as fit a logistic regression model

Additional Learning Outcomes for Graduate students only:

1.   Analyze, interpret and give an oral presention about a published paper that uses applied categorical data analysis.

2.   Design, implement and lead an in-class active learning activity.

Statistical Software:

We will be using R for statistical analysis in this class.

Class Conduct:

Computer use is strongly discouraged during class. Tablet use is ok. Likewise, students should not their cell phones during class, as it is distracting to everyone.

Prelim 1:

The first prelim is an in-class exam covering topics from the beginning of the course up   to, but not including logistic regression. The exam questions will be similar to homework questions. Students may use a non-graphing calculator. A formula sheet with important formulas and statistical tables will be provided.

Prelim 2: The second prelim will be a take-home exam in which you are asked to analyze a data set and write a report on your findings. You will work in groups of 3 – 4 students.  Standards and evaluation criteria for graduate students projects will be higher.

Final Exam: This exam will cover the material from the last two-thirds of the course: from logistic regression modeling to classification trees. It will be similar in format to Prelim 1, with calculations and short answers.

Special Projects (Graduate students only): Graduate students need to choose 2 of the following 3 projects, to be completed by the end of the semester. The first chosen         project should be completed in the first half of the semester, and the second project     should be completed in the second half of the semester. The choice of topic and timing need to be discussed and cleared beforehand with the professor.

1.   Analyze, interpret and give an oral presention of 20-minutes about a published paper whose main topic includes an analysis of categorical data. You are            required to attend the oral presentations of at least 3 other presentors.

2.    Design and implement (and possibly lead) an in-class active learning activity

Grade Calculation:

Grades for undergraduate students will be based on the homework, Prelim 1, Prelim 2    and the Final Exam. Grades for graduate students will be based on the above four items, plus the two special projects.

Homework assignments are equally weighted. The lowest score on the homework assignments will be replaced by the next lowest homework score.

Grade scale:

Satisfactory/Unsatisfactory Option: This class may not be taken with the S/U option.

Course Performance Assessments: The first prelim be an in-class exam. The second        prelim will be a take-home exam in which you will analyze a dataset. This will be done in groups of 2-4 students. The semi-final exam will be during semi-finals week. The              University has not yet set dates for the semi-final exams.

Exam dates     Prelim 1:  Thursday Sept 29, during class time. Different location.

Prelim 2 (Take Home exam):  TBA [9 days, Tues to Thurs )

Final Exam:  TBA

Homework Assignments: Homework assignments will be posted on the course website and will be due about one week later. (Homework may not be assigned every week.)      There will be 6 or 7 homework assignments.                                                                               Students may discuss homework problems with one another, but only at the level of a   “corridor conversation” with no notes taken. Homework that is late receives 10% off for within 24 hours late and 20% for 24-48 hours late. A zero after that. If you have a good  reason why you cannot meet a deadline (such as sickness or a family situation), please   check with me, no later 6 pm on the day before the deadline. In these cases some           arrangement can usually be found.

Regrade Requests: If there is a dispute about grading (a homework set or an exam), you may submit a regrade requests via Gradesope. All of the work, and not just the disputed question, will be regraded.

Academic Conduct:

Each student in this course is expected to abide by the Cornell University Code of           Academic Integrity. Any work submitted by a student in this course for academic credit should be the student’s own work.

All homework assignments are to be completed by students working on their own. You may discuss the homework problems with others if you wish, but only at the level of a  discussion in a corridor. No notes should be taken away from such discussions.

You may not work through the solutions with other students, and you cannot share     computer files. You may not discuss the homework with past students who may have  knowledge of the details of the homework set.  You are also not allowed to derive        advantage in any way from the existence of solutions prepared in prior years, whether the solutions were former students' work or copies of solutions that had been made    available by the instructors.

You are prohibited from buying and selling course materials that I have written. Such behavior constitutes academic misconduct.

I treat violations of Academic Integrity seriously.  Prior violations:

1)   Student copied homework solutions on 3 homework assignments and claimed that they were her own

Penalty: 0 for all assignments

Report filed with the College’s Committee on Academic Integrity End result: Student failed the course

2)   Student had homeworks and handouts in sight during final exam Penalty: 0 for the final exam

Report filed with the College’s Committee on Academic Integrity                      End result: D in the course (no credit received) and student did not graduate with his/her class. Had to take an additional course to graduate.

3)   Student checked their cell phone during an exam. Penalty: 0 for the exam (instead of 82)

Report filed with the College’s Committee on Academic Integrity End result: D in the course (would have gotten a B-)

Violations will be handled in accordance with the Code of Academic Integrity available  athttp://www.theuniversityfaculty.cornell.edu/AcadInteg/code.htmland you can learn more athttp://www.theuniversityfaculty.cornell.edu/AcadInteg/ .

If you have any questions about this policy, please ask me.