SC 704: Topics in Multivariate Statistics
Fall 2012
Monday/Wednesday 1:30 – 2:45 pm
O’Neill 245
Professor: Sara Moorman
Office: 404 McGuinn Hall
Office hours: Mondays 10:15-11:15; Wednesdays 3:00-4:00
E-mail: moormans@bc.edu
Phone: (617) 552 - 4209
About the Course
This applied course is designed for students in sociology, education, nursing, organizational
studies, political science, psychology, or social work with a prior background in statistics at the
level of SC703: Multivariate Statistics. It assumes a strong grounding in multivariate regression
analysis. The major topics of the course will include OLS regression diagnostics, binary,
ordered, and multinomial logistic regression, models for the analysis of count data (e.g., Poisson
and negative binomial regression), treatment of missing data, and the analysis of clustered and
stratified samples. All analyses in the course will be conducted using Stata, but no previous Stata
experience is necessary.
Readings
Required textbooks:
Enders, Craig K. 2010. Applied Missing Data Analysis. ISBN: 9781606236390
Long, J. Scott. 1997. Regression Models for Categorical and Limited Dependent Variables.
ISBN: 0803973748
Long, J. Scott and Jeremy Freese. 2006. Regression Models for Categorical Dependent
Variables Using Stata. 2nd ed. ISBN: 1597180114
Recommended textbook:
Acock, Alan C. 2010. A Gentle Introduction to Stata. 3rd ed. ISBN: 1597180750
Course reserves online:
Access “*” entries as .pdf files through the library website (http://www.bc.edu/libraries/) or
through the link on the course Blackboard page
(https://cms.bc.edu/webct/entryPageIns.dowebct).
SC 704 Topics in Multivariate Statistics
page 2 of 6
Software
This course requires the use of the program Stata. The most current version is available on the
computers in the Sociology graduate student lounge. For use on your own computer, you have
two options: (1) access the program through remote connection to apps.bc.edu, or (2) purchase
the program through BC’s Research Services. Ask your department administrator about
Campuswide GradPlan. Prices start at $32 (price for a six-month student license).
Assessment
Grading scale
A
93 – 100%
B
83 – 86%
F
0 – 59%
Task
Article critique
Project update I
Project update II
Peer review
Presentation
Final paper draft
AB-
90 – 92%
80 – 82%
Due date
Weekly, see schedule (N = 10)
October 15
November 14
November 26
December 3, 5, or 10
December 17
B+
C
87 – 89%
60 – 79%
Percentage of grade
10 at 3% each: 30%
5%
5%
5%
25%
30%
Article Critique
For each week, I’ve selected a recent publication from a major sociology journal that uses the
method we’re covering in class that week. You should read the article, and then complete two
tasks:
(1) Outline the analysis: What are the hypotheses or research questions, and how did the authors
go about testing them statistically?
(2) Critique the article: Make a list of what you think the authors did well and what they did
poorly. Specific questions to consider include: Does this analysis best answer the research
question given the data the researchers had available, or is there a discrepancy between the
research question and its empirical operationalization? Would you have chosen different
statistics instead of or in addition to the statistics employed? Were you left with any critiques of
the data or methods, or did the authors anticipate your concerns? If you had the data at hand,
would you be able to replicate the analysis? Were the results interpreted clearly and correctly?
Were the results presented effectively in tables and/or figures? Are the interpretations fair, or do
they seem to go beyond what the data can really support?
The purpose of this exercise is to prepare to discuss the article in class, so be sure to
outline and list rather than write an essay. You’ll submit your outline and list to receive credit. I
will simply note whether your work is complete or incomplete rather than judge the content of
your responses, so don’t worry if you don’t understand every last thing the authors did. Bring
your questions to class and we’ll work them out.
Research Project (More detail on each step to follow later in the semester)
I find that the best way to learn statistics is to practice them on real data that mean something to
you. Therefore, the major product of this course will be a journal-style research article (i.e., 20-
SC 704 Topics in Multivariate Statistics
page 3 of 6
35 pages in length, including the standard sections: title page, abstract, introduction, methods,
results, discussion, references, tables/figures). The article is required to include two or more of
the methods covered in class from September 19 onward. For example, you might run (a) a test
of mediation in a complex survey dataset, or (b) one outcome that requires binary logistic
regression and a second that requires Poisson regression, or (c) a multinomial logistic regression
on multiply-imputed data. Neither using Stata for your analyses nor testing a simple OLS
regression model “count” towards your two methods.
In mid-October and mid-November you will submit written “updates” that will be drafts
of sections of the paper. For instance, for one update you might submit the introduction and
methods sections, and for the next update you might submit the results and discussion sections,
or you might want me to look at revisions of your introduction and methods sections. Precisely
what you turn in will be up to you, although I’m happy to make recommendations on a case-bycase basis.
Updates have two purposes: (a) to ensure that you pace your work throughout the
semester, rather than try to write the whole paper the night before it is due, and (b) to provide
opportunities for my feedback on your work. As such, updates are required but not graded. If you
turn one in, you will receive full credit. It is all right if every last thing you have done in the
update is wrong, so please do not hide or avoid things you are uncertain about. The updates are
precisely the time for us to find those problems and fix them, not the time for me to assess your
work for a grade.
You will also exchange your second update with a classmate and complete peer reviews
for one another. I’ll match you up later in the semester based on the similarity of your topic, data,
or methods. Finally, you’ll give a conference-style presentation of your project in class, and
about a week later, submit your completed paper.
Although it’s certainly not a requirement, you should seriously consider using this project
as an opportunity to meet a degree requirement (e.g., area exams), prepare a conference
presentation, and/or develop a submission for publication. If you’re already working on a project,
I encourage you to use this course to develop it. If you’re starting from scratch, many datasets
are publicly available from universities and government agencies, and many more are available
to researchers through BC’s subscription to the Inter-University Consortium for Political and
Social Research (ICPSR) at the University of Michigan. Be aware that the deadline to submit a
paper for presentation at ASA 2013 in NYC will be in mid-January, and your course paper will
fit their submission criteria.
Submitting Your Work
x You should bring hard copy of your article critiques to class on their due dates. They are
due at 2:45 pm.
x You should e-mail me the updates and final draft of the research project as Word
documents. I’ll use the track changes function to give you feedback. They are due at
11:59 pm on their due dates.
x I’ll give you details later regarding what materials I need and how/when you should
submit them for the peer review and presentation assignments.
x I will not accept late work. If you have a conflict with a deadline, then make
arrangements with me well ahead of time.
SC 704 Topics in Multivariate Statistics
page 4 of 6
Academic Honesty
Your work must be your words and ideas. When writing papers, use quotation marks around
someone else’s exact words and identify whose words they are. If you come across a good idea,
by all means use it in your writing, but be sure to acknowledge whose idea it is. Do not allow
another student to copy your work. Failure to comply will result in (a) automatic failure of the
assignment, and (b) a report to the Dean and the Committee on Academic Integrity. For further
information, please review the College’s policies on academic integrity here:
http://www.bc.edu/offices/stserv/academic/resources/policy.html#integrity
Schedule
Date
Topic
September 5
Using Stata
September 10
Locating and using
data for secondary
research
Ordinary least
squares (OLS)
regression: Review
and diagnostics
Ordinary least
squares (OLS)
regression: Review
and diagnostics
Complex survey
data
September 12
September 17
September 19
September 24
September 26
Complex survey
data
Mediation
October 1
Mediation
October 3
Moderation
October 8
NO CLASS:
Columbus Day
Reading
Due
x Long chapter 2
x Long & Freese chapters 1-3
x Choi and David*
x
x
x
x
Johnson and Elliott*
Kreuter and Valliant*
Reynolds and Johnson*
Winship and Radbill*
Choi and
David
x
x
x
x
Baron and Kenny*
Flippen*
Hayes*
MacKinnon, Fairchild, and Fritz*
Reynolds and
Johnson
x Fairchild and McQuillin*
x Stets and Carter*
x Wu and Zumbo*
Flippen
SC 704 Topics in Multivariate Statistics
Date
Topic
October 10
Moderation
October 15
Missing data
October 17
Missing data
October 22
Multiple
imputation
October 24
October 29
Multiple
imputation
Binary outcomes
October 31
Binary outcomes
November 5
Ordinal outcomes
November 7
Ordinal outcomes
November 12
Nominal outcomes
November 14
NO CLASS
November 19
Nominal outcomes
November 21
November 26
NO CLASS:
Thanksgiving
Count data
November 28
Count data
page 5 of 6
Reading
Due
x Enders chapters 1, 2, and 10
x Hallerd*
Stets and
Carter;
Update 1
x Enders chapters 7, 8, and 9
x Silver, Silver, Siennick, and
Farkas*
Hallerd
x Kim and Pfaff*
x Long chapter 3
x Long & Freese chapter 4
Silver,
Silver,
Siennick, and
Farkas
x Cech, Rubineau, Silbey, and
Seron*
x Long chapter 5
x Long & Freese chapter 5
Kim and
Pfaff
x Long chapter 6
x Long & Freese chapters 6 and 7
x McCammon*
Cech,
Rubineau,
Silbey, and
Seron
Update 2
x Long chapter 8
x Long & Freese chapter 8
x Warner, Manning, Giordano, and
Longmore*
McCammon;
peer review
SC 704 Topics in Multivariate Statistics
Date
Topic
December 3
Class presentations
December 5
Class presentations
December 10
Class presentations
December 17
Reading
page 6 of 6
Due
Warner,
Manning,
Giordano,
and
Longmore
Final paper