PPD 558Fall 2017Syllabus
Sol Price School of Public Policy
University of Southern California
PPD 558: Multivariate Statistical Analysis
Thursdays6:00-9:20pm
Instructor:Jung Hyun Choi
Office:VPD207
Phone:(213) 821-9732
Email:
Office Hours:Monday2:00-4:00 pmor by appointment
TA Office Hours:Wednesday 1:00-3:00 pm
Course description:
This course will provide you with the analytical and quantitative skills required to conduct applied statistical research and to think critically aboutmethodology and proper interpretation of results when reading and analyzing empirical research such as that found in academic journals, white papers,and policy papers from national and international organizations.
The foundation of this course is multivariate regression analysis.We willbegin with the Ordinary Least Squares (OLS) model and expandour coverage to topics including logistic models, instrumental variables, panel data, and experimental methods to evaluate the impacts of public policies.We will discuss common problems faced by these methods, techniques for diagnosing and addressing these problems, and selection of the appropriate econometric tools to answer any given question.
Prerequisite: Applied Social Science Statistics (Stat Lab, PPD502x, PPD525, or equivalent)
Learning objectives:
This course will focus on training students to be capable practitioners and sophisticated consumers of quantitative research methods for policy analysis.While we will be making use of econometric theory, it is viewed as a means to an end: this course has a strong applied (rather than theoretical) orientation, so our coverage of econometric theory will be limited to those elements that directly serve the primary goal of enabling students to be successful users of quantitative empirical analysis.
A major goal of this course is to train students to effectively use econometric methods to inform the solution of complex policy, management, and planning problems. This will be achieved through lectures, discussions, in-class student presentations and activities, problem sets, and a final paper. For the final paper, students will be required to make use of these methods to create an original empirical analysis of a real problem that will serve as their class project.
Software Requirements:
The Stata software package is required for in-class data analysis, some problem sets, and for the final paper. There are several different versions of Stata available; the minimum required version is Stata/IC (Small Stata will not suffice).You should bring a laptop with Stata installed to all class meetings, as we will make frequent use of Stata during lectures and for in-class exercises.
Grades:
Students will complete an in-class midterm exam and a final exam, along with 5 problem sets and an analysis project on a topic of their choosing.Students are expected to complete all assignments on time. Unless otherwise specified, hard copies of all written assignments are due at the beginning of class on the date indicated; late assignments will not be accepted.If for some reason you are unable to hand an assignment in on time in person, you should submit it via email by the deadline and turn in an identical hard copyas soon as possible.
Problem Sets25% (5% each)
In-Class Assignments and Participation 5%
Midterm Exam20%
Final Exam25%
Analysis Project25%
Readings and resources:
(Required) A.H.Studenmund, Using Econometrics: A Practical Guide (6th Edition, 2011). Pearson/Addison-Wesley.
(Recommended for PhD students, optional for others) Joshua D. Angrist and Jorn-Steffen Pischke, Mostly Harmless Econometrics (2009). Princeton University Press.
(Recommended for PhD students, optional for others) Khandker, Shahidur R.,Gayatri B.Koolwal, and Hussain A.Samad,Handbook on Impact Evaluation: Quantitative Methods and Practices (2010), World Bank.
Available CC BY 3.0 IGO)
This book (freely available as a pdf from the World Bank) provides more advanced coverage of post-midterm topics such as experimental methods and instrumental variables, and also provides a useful reference for students looking to learn about material beyond that covered in PPD 558, such as propensity score matching and regression discontinuity methods.
(Optional) Alan C. Acock, A Gentle Introduction to Stata (Revised 3rd Ed., 2012). Stata Press.
May be a useful resource for students new to Stata; covers Stata basics and some foundational topics from this course, but not the more advanced topics or anything beyond this course.
Required textbook readings are indicated in bold in the course schedule, and optional readings from other textbooks are italicized. Various journal articles and other selections are included as optional supplemental readings, and will be made available on Blackboard. All assignments, lecture notes, etc., will be available via USC’s Blackboard site,
There are also many free online resources for learning how to use Stata that you may find to be helpful; for example, UCLA's stats group provides tutorials, learning modules, etc., at
Analysis project:
Students, working in small groups, will use multivariate econometric analysis methods studied in this course to address a real policy issue. The project will be done in groups of 3 or 4 people, with the possible exception of doctoral students who wish to work alone and use the project as part of their dissertation. Each group must choose a policy issue, select the appropriate method from the techniques learned in the course, obtain suitable data, perform the analysis, and write up the results. You must clearly differentiate your analysis from previous work; it doesn’t need to be a topic that has never been studied, but your analysis must contribute something new, not just replicate what others have done.
Group formation:During the first class meeting, I will have students provide information on their policy interests (top three areas of interest; for example: 1. Health, 2. Education, 3. Labor). I will make this information available to the class through Blackboard, and you can form your own groups based on it (I retain the right to make adjustments if necessary). By the end of Week 3, each group should send a single email to me listing the names of all group members. If you are not part of a group by the end of Week 3, please let me know at that time so that I can help you find a group.
Project proposal/description:You are expected to work on this project throughout the semester (especially important to identify any data availability problems early, and possibly pick a different topic). By the start of class in Week 6, each group must submit (via Blackboard) a project description and arrange a meeting with me (to take place prior to the following class) to discuss your project. This should be a brief(1-2 pages) description of the problem you will be addressing (i.e., policy issue and research question), expected dependent variable, initial causal model, description of the data you plan to use, sources of those data, and a project timeline.
Data and methods report: By the start of class in Week 8, each group must submit (via Blackboard)a more detailed description of the dataset, variables, and model that you will estimate based on those data. Unless you obtain instructor approval to do otherwise, you should use individual- or firm-level datasets. Be certain to justify the variables, data, and model you have chosen.For some, data collection may prove to be time-consuming (and frustrating!), so I strongly suggest that you start early. The Internet is a good source, especially for data collected by government agencies (examples include fedstats.sites.usa.govand A Google search for datasets on specific topics is often a good way to identify data sources. Note: you are still free to further modify your dataset and alter your methodology after this, I just want to make sure that you have gotten past the initial data search stage and have something that will allow you to perform your analysis.
Progress report: By the start of class in Week 10, groups must submit (via Blackboard) a progress report that briefly describesyour progress and provides an updated timeline.
Presentations: Groups will present their projects in class in Week 12 and 13, weeks prior to the due date of the final paper (allowing additional time to revise your projects after receiving feedback from me and from your classmates). Each group must submit their presentation slides (via Blackboard) by 5pm on the day prior to your presentation(I assume you will make your slides in PowerPoint, but please submit the file inpdf format to avoid formatting problems when it is moved to the classroom computer). A significant portion of your participation score will be derived from your comments and questions during your classmates’ presentations.
Final report: The final paper is due(as a pdf document) via Blackboardby the beginning of class in Week 15(each group should submit a single assignment; there is no need for each group member to submit a copy). Blackboard will not accept late submissions, so be sure to allow plenty of time for the submission of your paper; the Blackboard submission link will be made available well in advance of the due date, so you will have ample opportunity to submit the paper and resolve any technical issues that might arise.
Format of written report: Your written report should be modeled after the research articles read throughout the semester (though this sort of format is not exclusive to academic journals, and is used by many organizations throughout the world; a paper from the World Bank is available on Blackboard as an example). It must include the following sections (in this order):
- A one-page executive summary. The summary should review the policy question motivating the analysis and the major findings.
- A background section that describes the motivation for the analysis, including information on your intended audience/client, the problem you are addressing, and the context.
- A brief literature review of the relevant research (at most 1 page).
- A detailed description of your data, their sources, how they were collected, and their reliability. Any limitations of the data should be described here.
- A description of your methods and why they are appropriate.
- A description of the results of your analysis and any limitations or concerns about its validity. The regression results must be presented in tabular form (as seen in the various research papers we will read throughout the course).
- A conclusion section that develops the implications of your analysis for the problem or policy you are addressing.
- A references section that identifies the sources of all material cited in your paper.
- Two technical appendices; the first should list all of the variables used in the analysis and their descriptive statistics, and the second should provide the Stata output showing your regression results.
The report must not exceed 15 pages (double-spaced, 12 point Times New Roman font, 1-inch margins), not including the executive summary, references, and technical appendices.
Supporting files: In addition to submitting your paper via Blackboard, you must also email your final dataset (in Stata format) and do-file to me (by the same deadline) in order to allow reproduction of your analysis.
Peer evaluations: Each student must complete a peer evaluation in which you will assess the contributions of your group members. These evaluations are completely confidential and will never be shown to your group members; I will be the only person to see them. Please respond as honestly as possible, and do not discuss these evaluations with anyone else (either before or after you complete them). These peer evaluations are important togive me a better sense of how groups worked, and to provide an opportunity for you to bring to my attention any issues that arose over the course of this project (you are also welcome to come speak with me throughout the course regarding any serious issues that arise and cannot be resolved internally). In particular, I will give serious consideration to comments indicating that a student did not contribute satisfactorily to the group (did not do a fair share of the work, was uncooperative, did not meet agreed-upon deadlines, etc.) when assigning individual grades for the project, as well as positive comments noting particularly outstanding contributions. The evaluation forms will be made available on Blackboard once the final papers are turned in, and must be submitted via Blackboard within 1 week of the project due date.
A note to anyone contemplating giving less than full effort and free-riding on the work of your teammates: you are not the first to think of this strategy, and your predecessors who have attempted it have found this to be an extremely unsuccessful strategy (both in terms of theirindividual gradesfor this project, and their prospects for group work for the remainder of their degree programs – a reputation for not contributing your fair share to group projects is easily acquired but not so easily removed, and will follow you well beyond this class).Low individual efforton the group project is rarely an issue in this class, but in case you happen to be one of the unusual individuals who considers it,you have been warned!
Grading of final project:Your project grade will be based primarily on the final written report, but will also take into accountthe quality of your in-class presentation, as well as timely and satisfactory submission of other deliverables such as the project description, data and methods report, and progress report.In the case of especially meritorious or poor individual contributions as reflected in the peer evaluations and my own interactions with each group, each individual’s final grade for the project may differ from the base grade assigned to the group.
Course Schedule (Summary)
Topics/Daily Activities / Readings from textbook / Deliverables(due by beginning of classunless otherwise specified)Week 1
8/24/17 / Course Introduction / Studenmund Chapter 17
Week 2
8/31/17 / Regression Analysis
(Introduction) / Studenmund Chapters 1-5
Week 3
9/7/17 / Regression Analysis
(Model Specification) / Studenmund Chapters 6-7
Week 4
9/14/17 / Multicollinearity and Autocorrelation / Studenmund Chapters 8-9 / Problem Set 1
Week 5
9/21/17 / Autocorrelation (continued) and Heteroskedasticity / Studenmund Chapters 9-10 / Problem Set 2
Week 6
9/28/17 / Regression Analysis in Practice / Studenmund Chapter 11 / Analysis Project Description
Week 7
10/5/17 / Midterm Examination,
6:00 – 8:00pm
Week 8
10/12/17 / Limited Dependent Variable Models / Studenmund Chapter 13 / Analysis Project Data and Methods
Week 9
10/19/17 / Instrumental Variables Estimation / Studenmund Chapter 14 / Problem Set 3
Week 10
10/26/17 / Experimental Methods / Studenmund Chapter 16 (through p. 525) / Analysis Project Progress Report
Week 11
11/2/17 / Panel Data and Fixed Effects / Studenmund Chapter 16 (remainder) / Problem Set 4
Week 12
11/9/17 / Project Presentations
(In Class Exercise) / Problem Set 5
Week 13
11/16/17 / Project Presentations
(In Class Exercise)
Week 14
11/23/17 / No class meeting
(Thanksgivings)
Week 15
11/30/17 / Final Thoughts, Course Review and Summary / Final Written Project
Wednesday,
12/6/17 / Final Examination,
9:00 – 11:00am
Course Schedule and Readings
Week 1: Course Introduction; Empirical Research
Before the next class, you may wish to review the statistics material that you are expected to know already: Studenmund Chapter 17, and slides on Blackboard
Week 2: Regression Analysis (Introduction)
Studenmund Chapters 1-5
Week 3: Regression Analysis (Model Specification)
Studenmund Chapters 6-7
Graddy, Elizabeth (2001), “Juries and Unpredictability in Products Liability Damage Awards,” Law & Policy, 23:1, 29-45.
Week 4: Multicollinearity and Autocorrelation
Studenmund Chapters 8-9
Reed, Dwayne, Daniel McGee, Katsuhiko Yano, and Jean Hankin (1985), “Diet, Blood Pressure, and Multicollinearity,” Hypertension, 7, 405-410.
Jun, Kyu-Nahm, and Christopher Weare (2010), “Institutional Motivations in the Adoption of Innovations: The Case of E-Government,” Journal of Public Administration Research and Theory, 21, 495-519.
Week 5: Autocorrelation (continued) and Heteroskedasticity
Studenmund Chapters 9-10
Bertelli, Anthony, and Peter John (2013), “Public Policy Investment: Risk and Return in British Politics,” British Journal of Political Science, 43, 741-773.
Sengupta, Nishan, Michael B. Nichol, Joanne Wu, and Denise Globe (2004), “Mapping the SF-12 to the HUI3 and VAS in a Managed Care Population,” Medical Care, 42:9, 927-937.
Week 6: Regression Analysis in Practice
Studenmund Chapter 11
Week 7: Midterm Exam, 6:00 – 8:00pm
Week 8: Limited Dependent Variable Models
Studenmund Chapter 13
Graddy, Elizabeth, and Ke Ye (2008), “When Do We ‘Just Say No’? Policy Termination Decisions in Local Hospital Services,” Policy Studies Journal,36:2, 219-242.
Aguila, Emma, and Julie Zissimopoulos (2013), “Retirement and health benefits for Mexican migrant workers returning from the United States,” International Social Security Review, 66:2, 101-125.
Week 9: Instrumental Variables Estimation
Studenmund Chapter 14; Angrist & Pischke Chapter 4; Khandker, Koolwal, and Samad Chapter 6
McCarthy, T.J. (2013), “The Vietnam War, College Deferments, and Men’s Marital Prospects,” Working Paper.
Cawley, John, and Chad Meyerhoefer (2012), “The Medical Care Costs of Obesity: An Instrumental Variables Approach,” Journal of Health Economics, 31, 219-230.
Week 10: Experimental Methods
Studenmund Chapter 16 (through p. 525);Angrist & Pischke Chapter 2; Khandker, Koolwal, and Samad Chapters 3 and 5
Aguila, Emma (2011), “Personal Retirement Accounts and Saving,” American Economic Journal: Economic Policy, 3, 1-24.
Boarnet, Marlon G. (2001), “Enterprise Zones and Job Creation: Linking Evaluation and Practice,” Economic Development Quarterly, 15, 242-254.
Joyce, Geoffrey F., Julie Zissimopoulos, and Dana P. Goldman (2013), “Digesting the Doughnut Hole,” Journal of Health Economics, 32, 1345-1355.
Week 11: Panel Data and Fixed Effects
Studenmund Chapter 16 (remainder);Angrist & Pischke Chapter 5;
Shen, Yu-Chu, and Vivian Y. Wu (2013), “Reductions in Medicare Payments and patient Outcomes: An Analysis of 5 Leading Medicare Conditions,” Medical Care, 51:11, 970-977.
King, Andrew A., and Michael J. Lenox (2001), “Does It Really Pay to be Green? An Empirical Study of Firm Environmental and Financial Performance,” Journal of Industrial Ecology, 5:1, 105-116.