This course explores methods for causal inference in educational research. The two primary goals of the course are (1) to provide students with the analytic tools and skills to assess the quality of research claims in education, human development, and other substantive fields; and (2) to provide students with the skills to conduct research that supports causal inference. The course will introduce students to four core methods of quasi-experimental design: difference-in-differences, regression discontinuity, instrumental variables and matching. The course will culminate in either a research proposal or completed paper using one of the methods from the course. Prior coursework minimally through multiple regression and familiarity with a statistical programming language (preferably R) is assumed.
Class meets on Wednesdays from 9:00am to 11:50am in HEDCO 146
David D. Liebowitz
Email: daviddl@uoregon.edu
Office: 102S Lokey Education Building
Office Hours: Thursdays, 9:00am - 10:30am or by appointment. On Zoom or in person. Sign up on Canvas from 9:00-10:00; drop-in last 30 minutes.
Zoom: link | Meeting ID: 633 606 7468
By the end of this term, we expect students will be able to:
Required Text:
Murnane, R.J. & Willett, J.B. (2011). Methods Matter: Improving Causal Inference in Educational and Social Science Research. New York: Oxford University Press.
Supplemental Optional Texts:
Angrist, J.D. & Pischke, J. (2009). Mostly Harmless Econometrics: An Empiricist’s Companion. Princeton, NJ: Princeton University Press.
Cunningham, S. (2021). Causal Inference: The Mixtape.
Huntington-Klein, N. (2021). The Effect: An Introduction to Research Design and Causality.
Angrist, J.D. & Pischke, J. (2014). Mastering ’Metrics: The Path from Cause to Effect. Princeton, NJ: Princeton University Press.
Methods Matter (MM) by Murnane and Willett should be ordered online or purchased from the Duckstore. Other readings are linked directly or downloadable from the course website.
In general, each analytic method will be covered over two weeks. Class in the first week will be divided into two parts. First, we will discuss thoroughly the methods and their applications following a series of questions for which you are required to prepare answers in advance. Then, we will preview a Data Analysis and Replication Exercise (DARE) and guide you through a series of analytic programming steps. Class in the second week will also be divided into two parts. The first part of this class will be devoted to analyzing an empirical paper and comparing its results with the analytic replication and extensions you will have conducted. In the second part of the class, we will provide a conceptual overview of the new methodological strategy that will be introduced in the following unit.
Each week, you will read papers from the scholarly literature as well as chapters from Murnane and Willett (2011) and prepare responses to a set of detailed questions that we will post on the course website no later than the prior Thursday for each class. During class, we will discuss the questions at which point we will “cold” call on students at least partially at random. We do this to formatively assess class members’ grasp of the course content, to promote more equitable distribution in class participation, and to promote a spirit of accountability for developing a thorough understanding of complex and technical readings. Class discussions will focus on, but not necessarily be limited to, your answers to these questions. Cold calls will be restricted to only those pre-published questions, but the discussion will extend to student questions and extension points.
Research is an inherently collaborative experience. Building skill in research collaboration during graduate studies can be an important complement to technical and analytic skills. We strongly encourage you to form study groups to engage in the work of the class. Group members may jointly prepare responses to the reading questions (these are not for submission). We also encourage you to form pairs (3 people maximum) for the data replication exercises. Both members of the pair should contribute equally to the submitted product and each member of the pair will receive the same score on the assignment. Please email us if you have attempted to form a group or pair without success.
This course will complement the understanding of methodologies for generating causal inference in social science research design with applied exercises in the analysis of data to carry out these techniques. The teaching team will be able to support your learning in two commonly used statistical languages: R and Stata. R and RStudio are available free of charge to download. We will not directly support other programming languages, but students are welcome to complete either their DAREs or research project using other software with which they are familiar as long as they produce correct results.
Students can download the latest version of R here. We strongly recommended that students also download the RStudio GUI, available here. Both softwares are free.
While we will teach you how to effectively use R and RStudio to conduct analyses, one of the key skills required to use R is the ability to find answers on your own. Many common questions or problems are either available on blogs or have been asked and answered in discussion forums already. Finding and deciphering those answers is an important skill you should seek to hone. You will never remember all of the programming commands!
Here are some sites where you can find the answers to many R questions and learn new tricks:
Artificial Intelligence (AI) chatbots and the large language models (LLMs) on which they rely have dramatically increased the speed and efficiency of many programmers. Members of the teaching team regularly use such tools in their analytic and drafting tasks. That said, they are not (as least currently) substitutes for skilled analysts and writers. Beyond AI chatbots’ known proclivity for “hallucinating” facts and reproducing social biases, their solutions to programming tasks often require adaptation and revision by a knowledgeable human. Further, because their ability to generate text relies on using billions of phrase chunks in the public domain to predict the next word, their language on technical topics can be imprecise when many other writers on these topics are also imprecise. Thus, while we encourage you to investigate how AI chatbots can help improve your programming and statistical analysis skills, we caution you to skeptically review all code and language produced to ensure its alignment to the course expectations. To be explicit: you may use AI chatbots for assistance with your assignments. If you use one to generate your responses, you must indicate so on your assignment. You do not need to do so if you have used these tools only to help with coding tasks and/or light editing of your written responses. You are likely already familiar with multi-purpose LLMs (e.g., OpenAI’s ChatGPT or Anthropic’s Claude. As you gain more experience as a programmer, you may want to use an AI tool that is designed specifically to help with coding such as GitHub Copilot, AskCodi or Codex.
This is a high-level sketch of our weekly schedule. For more details, see here.
| Week | Class Date | Topics | Required Readings | Assignments Due |
|---|---|---|---|---|
| 1 | Apr. 1 | A Common Vocabulary | MM Ch. 1-5, 7 Hoxby 2016 |
Course survey (Apr. 3) |
| 2 | Apr. 8 | Difference-in-differences (I) | MM Ch. 8 Dynarski 2003 |
- |
| 3 | Apr. 15 | Difference-in-differences (II) | Liebowitz et al. 2022 | DARE #1 (Apr. 14) |
| 4 | Apr. 22 | Regression discontinuity (I) | MM Ch. 9 Angrist & Lavy 1999 Dee & Penner 2017 |
Research project proposal (Apr. 21) |
| 5 | Apr. 29 | Regression discontinuity (II) | Holden 2016 | DARE #2 (Apr. 28) |
| 6 | May 6 | Instrumental variables (I) | MM Ch. 10 & 11 Dee 2004 Angrist et al. 2016 |
- |
| 7 | May 13 | Instrumental variables (II) | Kim et al. 2011 | DARE #3 (May 12) |
| 8 | May 20 | Matching (I) | MM Ch. 12 Diaz & Handa 2002 |
- |
| 9 | May 27 | Matching (II) | MM Ch. 13 & 14 Umansky & Dumont 2021 |
DARE #4 (May 26) |
| 10 | Jun. 3 | Presentations | None | Research project presentation (Jun. 3) |
| Finals | NA | - | - | Final research project (Jun. 11) |
MM = Methods Matter
Final grades will be based on the following elements:
Data Analysis and Replication Exercises (DAREs) (4 DAREs, worth 10 points each for a total of 40 points): For each of the four units, you will complete a Data Analysis and Replication Exercise (DARE) that will require you to integrate statistical programming, methodological understanding and scholarly writing skills. In the preceding class, you will be provided with guidance and a detailed assignment that will entail familiarizing yourself with a replication dataset for one of the papers we have read. You will need to familiarize yourself with this small data set, produce a set of tables and figures that replicate some of the results from the paper, and write up your results in a memo following scholarly reporting standards. Each DARE will have a detailed assignment sheet and you will receive model responses after your work is submitted You will need to upload your work to Canvas by 11:59pm of the Tuesday in the week in which the DARE is due.
Final Research Project (10 points for the presentation, 35 points for the written manuscript): The central culminating project of this class is to initiate and/or complete an original research project, present the project to the class, and produce a written product that might take on a variety of forms. Our notion of what constitutes a complete project is intentionally vague to permit a variety of projects at different stages and is subject to negotiation between student and instructor. Some examples of the types of projects you might complete:
We favor projects that will lead ultimately to a scholarly product that can appear on your CV. This includes, but is not limited to conference presentations, published papers, proposals, qualifying papers, a dissertation. In order to ensure that you make an early start and continue to make progress on your project, you must submit an initial one page overview of your proposed project to us by April 21 at 11:59pm. You will make a public presentation of your project to other class members on June 3. Your final written project is due June 11 at 5:00pm. You will have the opportunity to receive extended written comments on one (1) draft of your written project. You may submit your written project by June 5 at 5:00pm and will receive extended comments that will help you revise your work for your final paper. If you choose this option, you will receive brief, general comments on your final draft.
Final grades are based on the following scale: A+ 98-100, A 94-97, A- 90-93; B+ 87-89, B 83-86, B- 80-82; C+ 77-79, C 73-76, C- 70-72. Any work scoring less than a C- will be required to be revised and resubmitted. Students taking the course on a Pass/Not Pass basis must earn a minimum grade of 80% and complete all assignments for the class.
Graduate students are expected to perform work of high quality and quantity, typically with forty hours of student engagement for each student credit hour. For this course, the following table shows the number of hours a typical student would expect to spend in each of the following activities:
| Educational activity | Hours | Explanatory comments |
|---|---|---|
| Class attendance | 26 | 10 class meetings of 2:40 (includes 10 min. break) |
| Class reading and prep | 30 | Includes out-of-class preparation of responses to detailed questions on reading |
| Data Analytic & Replication Exercises (DAREs) | 24 | DAREs should take approx. 6 hours each (on avg.) |
| Research project | 40 | Includes data analysis, preparation of presentation, and drafting of written project |
| Total hours | 120 | These are approximations. Reading and especially analytic time will vary per individual |
The University of Oregon is located on Kalapuya Ilihi, the traditional indigenous homeland of the Kalapuya people. Today, descendants are citizens of the Confederated Tribes of the Grand Ronde Community of Oregon and the Confederated Tribes of the Siletz Indians of Oregon, and they continue to make important contributions in their communities, at UO, and across the land we now refer to as Oregon.
In-person attendance at all classes is expected and required. Students must contact the instructor in case of illness or emergencies that preclude attending class sessions. Messages can be left on the instructor’s e-mail at any time of the day or night, prior to class. Students will be expected to submit written responses to reflection questions if they miss class.
It is the policy of the University of Oregon to support and value equity and diversity and to provide inclusive learning environments for all students. To do so requires that we:
In this course, class discussions, projects/activities and assignments will challenge students to think critically about and be sensitive to the influence, and intersections, of race, ethnicity, nationality, documentation, language, religion, gender, socioeconomic background, physical and cognitive ability, sexual orientation, and other cultural identities and experiences. Students will be encouraged to develop or expand their respect and understanding of such differences.
Maintaining an inclusive classroom environment where all students feel able to talk about their cultural identities and experiences, ideas, beliefs, and values will not only be my responsibility, but the responsibility of each class member as well. Behavior that disregards or diminishes another student will not be permitted for any reason. This means that no racist, ableist, transphobic, xenophobic, chauvinistic or otherwise derogatory comments will be allowed. It also means that students must pay attention and listen respectfully to each other’s comments
As a parent of three young-ish children, I understand the difficulty in balancing academic, work, and family commitments. Here are my policies (with credit to Daniel Anderson) regarding children in class: