Applied Quantitative Analysis

POL 346
Lectures: Monday and Wednesday 3:30-4:20pm (Jones 100)
Precepts: By assignment (Friend Center 005)

Instructional Team Information

Santiago Olivella (instructor)
Office: 039 Corwin Hall
eMail: olivella@princeton.edu
Office Hours: Wednesday 9:00am11:00am.
Brandon de la Cuesta (head preceptor)
Office:
eMail: bjmiller@princeton.edu
Office Hours:

Course Description
This class is designed to introduce you to some of the most commonly used statistical
models for studying observational data. It is meant to build on the skills and knowledge
developed in an introductory statistics course such as POL 345, and is geared towards
those who wish their senior thesis to include a quantitative component. In addition to
presenting the linear model in the light of the likelihood framework, it also covers models for discrete data including binary, multinomial, ordered and count outcomes. The
course will also rely entirely on the R programming environment, a powerful and versatile open-source statistics suite. The course is therefore ideal for those who wish to
become thorough consumers and apt producers of professional empirical research.
Learning Objectives
By the end of this course, you should be able to:
Understand the issues involved in identifying causal relationships using observational data, and design a study with these issues in mind.
Identify widely used probability distributions, and understand the contexts in which
they should be used as models of how data are generated.
Estimate and assess parametric models for continuous, binary, multinomial, ordered and count outcome variables, and interpret results of these estimations using
both graphical and text-based tools.

 Estimate models that involve a (known) form of dependence across observations

(e.g. panel data) using fixed-effects.
Have a minimal understanding of non-parametric and machine learning techniques
for prediction.
Course Prerequisites
Completion of POL 345 or equivalent course is required. A working knowledge of R
is assumed throughout this course. Basic training in differential calculus and matrix
algebra is ideal, but not required.
Class Structure
Our sessions will be divided into traditional lectures (with an emphasis on theory) and
applied precept sessions. In order to encourage learning from your peers, you will work
in pairs during the precept portions of the class.
Textbook
The course relies mainly on one textbook:
(F) Fox, John. Applied Regression Analysis and Generalized Linear Models.. 3rd Edition. Sage
Publications. ISBN-10: 0761930426
All other assigned readings will be made available electronically via Blackboard for
the duration of the course.
Software
The R programming environment (available for download at http://cran.rstudio.
com/) is the lingua franca of applied statisticians. Its main appeal is its open-source nature, which allows anyone to expand it and modify it however they want, and you to
use it for free. It is also cross-platform, making it highly portable. For all these reasons, we will learn how to apply all the concepts learned in class in R. In addition to
the R base, we will be using a user interface called RStudio (available for download at
http://www.rstudio.com/ide/download/).

Requirements and Evaluation

Grading in this class will be based on the components described below. Failure to meet
the requirements of the course will result in a failing grade.

Weekly Questions 10%

To ensure that you get the most of each session, you are required to either a) post at least
one question prompted by the study of the reading materials assigned for each week,
or b) answer at least one of the current questions posted by your classmates. Questions
and/or answers should be posted on Piazza no later than 5pm on the Sunday prior to
the first lecture of the week. These will help adjust the lectures according to your needs,
and will only be answered explicitly by the instructional team when they havent been
addressed in either the lectures or the precept sessions.
Weekly Problem Sets 30%
Seven problem sets will be distributed throughout the course. A PDF with your code and
written answers should be posted to the corresponding Blackboard assignments folder
by 3:30pm on the specified date using the following nomenclature: xxxPSetX.pdf, where
xxx is your netID and X is the problem set number (for example, if I were posting the 2nd
problem set, my file would be named olivellaPSet2.pdf). Each problem set will be graded
as X (unsatisfactory), X (satisfactory), or X+ (excellent), and counts equally towards
the final course grade.
Take-home exams 40% (20% each)
There will be two exams during the semester, one during the midterm week and another
during the final week of semester. Both exams will consist of at-home assignments to
be returned no later than 5:00pm on the Sunday following their distribution, and will
cover the material discussed in class up to that point. You must post your answers in
PDF format to the corresponding assignment folder in Blackboard, using the following
nomenclature: xxxExamX.pdf, where xxx stands for your NetID and X for 1 or 2, depending on the exam. Although you may use any resource you find useful, you are not
to share or discuss your work, completely or partially, with other students, as discussed in the
Academic Integrity section below.
Final Project 20%
As a final exercise, you will conduct an original data analysis using R. There will be two
deliverables for this project. A month and a half before Deans Date, you will submit
the main goal of your analysis (e.g. Establish whether policy X has proved effective
in achieving goal Y) and the details of your research design, including a plan for data
collection (with an explicit data source). On Deans Date, you must post a PDF document
with all code involved in the analysis as well as a short (max. 3 pages) discussion of your
results to the corresponding folder on Blackboard, using the following nomenclature:
xxxReplication.pdf, where xxx stands for your NetID.

Score

Grade

Score

Grade

Score

Grade

Score

Grade

94
90
87

A
AB+

83
80
77

B
BC+

73
70
60

C
CD

< 60

Grading Scale (in percentages)

Class Policies
Late work and Incompletes
For all assignments, late submissions will not be accepted without prior permission. For
the final project, extensions can only be granted with explicit authorization from the students residential college dean or director of studies.
Attendance
You will not be graded directly on your attendance to either the lectures or the precept
sessions. However, I strongly suggest students expecting to receive an A in this course
attend all sessions.
Technology in the classroom
You will frequently make use of computers in this course, particularly during precepts.
Please be respectful to your instructor and your peers by using your computers only for
class-related purposes. Put your phone away before class starts and dont bring it out.
Students with disabilities
Students with disabilities enrolled in this course who may need disability-related classroom accommodations are encouraged to make an appointment to see me before the end
of the second week of the semester. All conversations will remain confidential. Please
also arrange to have the required documentation sent to me for any accommodations at
least two weeks prior to the rst exam.
Religious observances
Some students may wish to take part in religious observances that occur during this
semester. If you have a religious observance that conflicts with your participation in the
course, please meet with me before the end of the second week of the semester to discuss
appropriate accommodations.
Academic Integrity
Work on each take-home exam is strictly individual, and you must refrain from discussing any portion of your progress or proposed solution with your peers. While I

encourage you to work with others when completing your problem sets, I expect each
student to write and submit their own solutions for those as well. In both instances, you
are expected to write your own code. In general, well follow the guidelines used for
POL 345, which inherit from those used in computer science courses at Princeton.
Programming necessitates that you reach your own understanding of the problem
and discover a path to its solution. When completing problem sets, discussions with
other people (whether via the Internet or in person) are permitted and encouraged. However, when the time comes to write code that solves the problem set, and when giving
answers to problems in an exam, such discussions (except with course staff members)
are no longer appropriate: the code must be your own work. Do not, under any circumstances, copy another persons code, regardless of whether you are working on a
problem set or an exam. Incorporating someone elses code into your program in any
form is a violation of academic regulations. Abetting plagiarism or unauthorized collaboration by sharing your code is also prohibited. Sharing code in digital form is an
especially egregious violation: do not e-mail your code to anyone. Novices often have
the misconception that copying and mechanically transforming a program (by rearranging independent code, renaming variables, or similar operations) makes it something
different. Actually, identifying plagiarized source code is easier than you might think.
For example, there exists computer software that can detect plagiarism.
This policy supplements the Universitys academic regulations, making explicit what
constitutes a violation for this course. Princeton Rights, Rules, Responsibilities handbook
asserts:
The only adequate defense for a student accused of an academic violation is
that the work in question does not, in fact, constitute a violation. Neither the
defense that the student was ignorant of the regulations concerning academic
violations nor the defense that the student was under pressure at the time the
violation was committed is considered an adequate defense.
I encourage you to read the Universitys Academic Integrity policy, available here. If you
have any questions about these matters, please consult a course staff member.

Calendar with topics, required readings and assignments

Date
1-Feb
3-Feb
8-Feb

Topic
Intro
A framework for model design
Design & Identification

Assignments

10-Feb
15-Feb
17-Feb
22-Feb

Design & Identification

The linear model (LM) part 1
LM part 2
LM interpretation

PS1 out

24-Feb

LM diagnostics

PS3 out / PS2 due

Readings

Rosenbaum 2010, Ch. 4 and

Ch. 5 (5.1-5.2.6)
Fox, Ch. 5
PS2 out / PS1 due
Fox, Ch. 6 (6.1 - 6.2), Ch. 7
(7.1, 7.2, 7.3.1, 7.3.4)
Fox, Ch. 11 (11.1-11.4), Ch.
12 (12.2.1 & 12.3) , Ch. 16
(16.5)
Continues on next page...

Date
29-Feb
2-Mar
7-Mar
9-Mar
21-Mar
23-Mar
28-Mar
30-Mar
4-Apr
6-Apr
11-Apr
13-Apr
18-Apr
20-Apr
25-Apr
27-Apr

Calendar continued from previous page

Topic
Assignments
Panel data part 1
Panel data part 2
PS 3 due
Binary outcomes part 1
Midterm out (due 11-Mar)
Binary outcomes part 2
Diagnostics for non-linear
models part 1
Diagnostics for non-linear
PS4 out
models part 2
Interpretation for non-linear
models part 1
Interpretation for non-linear
PS5 out / PS4 due
models part 2
Ordered outcomes part 1
Ordered outcomes part 2
PS6 out / PS5 due
Multinomial outcomes part 1
Multinomial outcomes part 2
PS7 out / PS6 due
Count outcomes part 1
Count outcomes part 2
PS7 due
Non-parametric prediction
Final Out (Due 29-Apr)
models part 1
Non-parametric prediction
models part 2

Readings

Fox, Ch. 14 (14.1)

Fox, Ch. 15 (15.4.1-15.4.2)

Hanmer and Kalkan (2013)

Fox, Ch. 14 (14.2.3)

Fox, Ch. 14 (14.2.1 - 14.2.2)
Fox, Ch. 15 (15.2)
James et al. 2014 Ch. 8