**DEPARTMENT OF APPLIED ECONOMICS AND
STATISTICS
STAT 675 – LOGISTIC REGRESSION
**

**Go to John Pesek's Home website **

** **

**Go to SAS 9.3 Documentation Index **

** **

**Go to Base SAS (9.3) Documentation
**

** **

**Go to SAS/STAT (9.3) Documentation**

** **

**Go to SAS/GRAPH (9.3) Documentation**

** **

**Instructor:** John D. Pesek, Jr.

227
Townsend Hall

Phone: (302)831-1319

E-mail:
pesek@udel.edu

Web-page:
http://www.udel.edu/pesek/stat675/index.html

**Office
Hours: **By
appointment

**Texts: Required:**
*An Introduction to Categorical Data Analysis, *2^{nd} Edition by Alan Agresti, Wiley 2007.

**Optional:** *The Little SAS Book: A Primer*, 5** ^{th} Edition**, by Lora A Delwiche and Susan J. Slaughter,
SAS Institute Inc., 2012.

**Objectives:** The purpose of this course is to learn
how to analyze, interpret and assess the validity of logistic regression models
in various applied contexts such as medicine and marketing. The course will use
primarily procedures in the SAS system to do the data analysis.

**UNIVERSITY OF DELAWARE GENERAL EDUCATION GOALS:** The objectives of this course align with the following General
Education Goals (**bolded**):

**Attain effective skills in **(a) oral and (b) written communication**,
(c) quantitative reasoning, and (d) the use of information technology – **By doing
homework assignments, supportive reading and lab exercises. These involve
analysis and computation, both by hand and using SAS software.

**Learn to think critically
to solve problems – **By
doing homework assignments and class exercises. These involve analysis and
computation, both by hand and using SAS software.

**Be able to work and learn
both independently** **and
collaboratively – **By doing independent homework assignments and
class exercises in collaboration with a partner. These involve analysis and computation,
both by hand and using SAS software.

**Course
Requirements:**

1. In case class is cancelled because of
bad weather or other contingencies, make up classes may be necessary. It may
also be necessary to reschedule exams.

2. On
Tuesdays, March 5, 2013 and April 16, 2013 there will be inclass examinations.
There will also be a final at a date to be announced.

**Important!** The final and inclass exams **must**
be taken at the scheduled time unless prior arrangements are made. There **must**
be adequate cause in the **judgment** of
the instructor.

**Note:** Circumstances
such as weather may cause rescheduling of the exams.

**Note: **Plane
reservations leaving before the final date are **not **considered adequate cause.

**Note:** Attendance at a wedding is **not **adequate
cause for taking exams at other times. This is regardless of whether you are
getting married, you are a member of the wedding party, or you are just in
attendance.

**Note: ** Having more than one final on the same day is
also not adequate cause.

**Note:** There are many legitimate
reasons such as illness and family emergency.
If you have one of these, please contact the instructor.

**Note:** According to
university policy, students with either learning and/or physical disabilities
are entitled to special consideration. To qualify for this consideration, the
instructor must be given official notification of the disability from the
appropriate university unit.

Both
the exams and the final will be **cumulative.** Some questions on the exam will require the
demonstration of lessons learned in class and in the homework while some will
also require the resourceful use of that knowledge. When preparing for exams,
it is important to remember that in addition to using the skills acquired in
class, it is also necessary to be able to decide which skills are needed for a
particular problem.

I
often give out practice exams. These are usually the previous year's exams. You
should note the standard caution that the same topics may not be covered on the
exams as the practice exams.

3. There
are also a number of required assignments.
Each assignment is due on a specific date to be announced. Students are
responsible for providing a readable and understandable presentation of
results. Without an appropriate excuse an assignment turned in after the due
date will receive a grade of zero. In general assignments will be submitted
using Sakai.

4. We
will follow parts of the text fairly closely. Students are expected to keep up
on reading the text and will be asked questions about what they read in class.
There will be weekly study assignments as well.

5. Class
attendance is very important. Students
are held responsible for all work covered and for meeting course deadlines.

**Note: **Missing class in order to complete work
for another class or to study for an exam in another class is **not** a
proper excuse.

6. During
class students are expected to be respectful of each other and the
instructor. Questions are welcome if
they are germane to the current topic. Other questions are welcome when asked
in private. In this class data will come
from a variety of disciplines. It is
natural to be most interested in data from your own field of study. However, respect and courtesy toward data
from other disciplines is expected.

7. Academic
honesty is expected at all times (See the Student Guide to University Policies
for complete information on the Code of Student Conduct at http://www.udel.edu/stuguide/12-13/index.html
).

1.
In general students are expected to
know and follow the university's policy on responsible computer use.

**Grading Procedure:**

The final course grade will be based
upon the student’s performance on the assignments and the exams. The assignments will count 20-30 percent of
the grade, the exams 70-80 percent of the grade.

Important
information for listeners (official auditors). At the University of Delaware
you may take a course as a listener. Then you are not required to take exams or
turn in homework although you may do so. However, you ARE required to attend
class. If a student has listener status and has a large number of unexcused
absences, the instructor may give a grade of LW (for listener withdrawn)
instead of L (for listener). I have
decided to enforce this policy. While an LW will not affect your GPA or your
graduation, prospective employers are often concerned to see withdrawn courses
on your transcript and it is best to avoid them.

**Requests
for score changes:**

If you feel either an assignment or an
exam deserves a higher score, you may make a written request. The written request must be made on a **separate **sheet of paper and the
assignment or exam must be attached. If
you are still not satisfied, you may make an appointment with the instructor to
discuss the matter. For Sakai
assignments, written requests should be made through the Sakai message
system. In this case there will be no
need to include the assignment since I will have retained a copy.

**Consultations**:

Students are encouraged to visit the
instructor in his office. Students may
make appointments or drop by (In the last case the instructor may not always be
available). Students are also encouraged to communicate with the instructor
using E-mail or the Sakai message system.

**Announcements**:

Announcements about the course will be
made by e-mail. Students are expected to
pay close attention to e-mail messages from the instructor.

**Handouts**:

In
general handouts will be available on Sakai in “pdf”
format. Students are expected to download and print copies to have available in
class. Some handouts may still be
provided in class. To access the handouts, go to the URL

**Tentative list of topics (Some may not be covered):**

z Categorical
Response Data

z Nominal/Ordinal
Scales

z Review
of Binomial and Multinomial Distributions

z Inference
for a proportion

z Inference
for discrete data

z Contingency
Tables

k
Joint, Marginal and Continuous
Distributions

k
Independence

k
Comparing proportions in two by two
tables

z Odds
ratios

z Tests
of independence

z Tests
of independence for ordinal data

z Exact
inference for small samples

z Association
in three-way tables

z Generalized
linear models

k
In general

k
Binary data – logit and probit

k
Others Poisson and Negative binomial

z Statistical
Inference and model checking

z Logistic
regression

k
Interpretation

k
Odds ratios

k
Inference

k
Dummy variables

k
Model selection

k
Model checking

k
Sample size and power

z Multicategory
logit models

k
Multinomial

k
Ordinal(cumulative logits)

z Log
linear models

z Matched
pairs

z Correlated
and clustered responses

z Mixed
logistic regression models