Welcome to Gov 2003!
Fall 2021
Matthew Blackwell
Gov 2003 (Harvard)
1/6
Why causal inference?
• Social science theories are almost always causal in nature.
2/6
Why causal inference?
• Social science theories are almost always causal in nature.
• We should understand when our methods can have a causal
interpretation.
2/6
Why causal inference?
• Social science theories are almost always causal in nature.
• We should understand when our methods can have a causal
interpretation.
• Manksi: “data + assumptions = conclusions”
2/6
Why causal inference?
• Social science theories are almost always causal in nature.
• We should understand when our methods can have a causal
interpretation.
• Manksi: “data + assumptions = conclusions”
• Causal inference is about making assumptions and conclusions more
transparent.
2/6
Why causal inference?
• Social science theories are almost always causal in nature.
• We should understand when our methods can have a causal
interpretation.
• Manksi: “data + assumptions = conclusions”
• Causal inference is about making assumptions and conclusions more
transparent.
• The old way was kitchen sink regression + causal weasel words:
2/6
Why causal inference?
• Social science theories are almost always causal in nature.
• We should understand when our methods can have a causal
interpretation.
• Manksi: “data + assumptions = conclusions”
• Causal inference is about making assumptions and conclusions more
transparent.
• The old way was kitchen sink regression + causal weasel words:
• “associated with”, “leads to”, “the [causal?] effect of”, “[in|decreases]”,
“more likely”, “encourages”, “is linked to”, “correlated”, “predicts”
2/6
Why causal inference?
• Social science theories are almost always causal in nature.
• We should understand when our methods can have a causal
interpretation.
• Manksi: “data + assumptions = conclusions”
• Causal inference is about making assumptions and conclusions more
transparent.
• The old way was kitchen sink regression + causal weasel words:
• “associated with”, “leads to”, “the [causal?] effect of”, “[in|decreases]”,
“more likely”, “encourages”, “is linked to”, “correlated”, “predicts”
• Credibility/causal revolution: pick a causal estimand and a research
design to identify it.
2/6
Staff
• Instructor: Matt Blackwell
3/6
Staff
• Instructor: Matt Blackwell
• Teaching Fellow: Sooahn Shin
3/6
Teaching resources
• Lecture: general theoretical and practical issues.
4/6
Teaching resources
• Lecture: general theoretical and practical issues.
• Goal is to be recorded and uploaded fairly quickly.
4/6
Teaching resources
• Lecture: general theoretical and practical issues.
• Goal is to be recorded and uploaded fairly quickly.
• Section: more targeted examples with an eye toward assignments.
4/6
Teaching resources
• Lecture: general theoretical and practical issues.
• Goal is to be recorded and uploaded fairly quickly.
• Section: more targeted examples with an eye toward assignments.
• Canvas site: place for all the course materials
4/6
Teaching resources
• Lecture: general theoretical and practical issues.
• Goal is to be recorded and uploaded fairly quickly.
• Section: more targeted examples with an eye toward assignments.
• Canvas site: place for all the course materials
• Lecture notes, section materials, etc.
4/6
Teaching resources
• Lecture: general theoretical and practical issues.
• Goal is to be recorded and uploaded fairly quickly.
• Section: more targeted examples with an eye toward assignments.
• Canvas site: place for all the course materials
• Lecture notes, section materials, etc.
• Gradescope: submitting assignments.
4/6
Teaching resources
• Lecture: general theoretical and practical issues.
• Goal is to be recorded and uploaded fairly quickly.
• Section: more targeted examples with an eye toward assignments.
• Canvas site: place for all the course materials
• Lecture notes, section materials, etc.
• Gradescope: submitting assignments.
• Ed Board: discussion of course material.
4/6
Teaching resources
• Lecture: general theoretical and practical issues.
• Goal is to be recorded and uploaded fairly quickly.
• Section: more targeted examples with an eye toward assignments.
• Canvas site: place for all the course materials
• Lecture notes, section materials, etc.
• Gradescope: submitting assignments.
• Ed Board: discussion of course material.
• Slack: logistical and social discussion, DMs for help/study groups
4/6
Teaching resources
• Lecture: general theoretical and practical issues.
• Goal is to be recorded and uploaded fairly quickly.
• Section: more targeted examples with an eye toward assignments.
• Canvas site: place for all the course materials
• Lecture notes, section materials, etc.
• Gradescope: submitting assignments.
• Ed Board: discussion of course material.
• Slack: logistical and social discussion, DMs for help/study groups
• Office Hours: ask even more questions.
4/6
Textbooks
• Responsibility = material covered in lectures.
5/6
Textbooks
• Responsibility = material covered in lectures.
• Good books that I’ll draw upon:
5/6
Textbooks
• Responsibility = material covered in lectures.
• Good books that I’ll draw upon:
• Imbens & Rubin: fairly technical, but covers basics well.
5/6
Textbooks
• Responsibility = material covered in lectures.
• Good books that I’ll draw upon:
• Imbens & Rubin: fairly technical, but covers basics well.
• Hernan & Robins: slightly less technical, more biostat influence.
5/6
Textbooks
• Responsibility = material covered in lectures.
• Good books that I’ll draw upon:
• Imbens & Rubin: fairly technical, but covers basics well.
• Hernan & Robins: slightly less technical, more biostat influence.
• Angrist & Pischke: most readable, opinionated, a bit narrow.
5/6
Textbooks
• Responsibility = material covered in lectures.
• Good books that I’ll draw upon:
• Imbens & Rubin: fairly technical, but covers basics well.
• Hernan & Robins: slightly less technical, more biostat influence.
• Angrist & Pischke: most readable, opinionated, a bit narrow.
• Morgan & Winship: good combo of potential outcomes and graphs.
5/6
Work
• Problem sets (50%)
6/6
Work
• Problem sets (50%)
• There will be 9 problem sets, you are only responsible for 8 (we’ll drop
the lowest score).
6/6
Work
• Problem sets (50%)
• There will be 9 problem sets, you are only responsible for 8 (we’ll drop
the lowest score).
• Mix of analytic review questions and empirical applications of methods.
6/6
Work
• Problem sets (50%)
• There will be 9 problem sets, you are only responsible for 8 (we’ll drop
the lowest score).
• Mix of analytic review questions and empirical applications of methods.
• Go out on Thursdays and due the following Wednesday, formatted nicely.
6/6
Work
• Problem sets (50%)
• There will be 9 problem sets, you are only responsible for 8 (we’ll drop
the lowest score).
• Mix of analytic review questions and empirical applications of methods.
• Go out on Thursdays and due the following Wednesday, formatted nicely.
• Final collaborative project (40%)
6/6
Work
• Problem sets (50%)
• There will be 9 problem sets, you are only responsible for 8 (we’ll drop
the lowest score).
• Mix of analytic review questions and empirical applications of methods.
• Go out on Thursdays and due the following Wednesday, formatted nicely.
• Final collaborative project (40%)
• short (< 20 pg) research paper either applying or extending a method
from the class.
6/6
Work
• Problem sets (50%)
• There will be 9 problem sets, you are only responsible for 8 (we’ll drop
the lowest score).
• Mix of analytic review questions and empirical applications of methods.
• Go out on Thursdays and due the following Wednesday, formatted nicely.
• Final collaborative project (40%)
• short (< 20 pg) research paper either applying or extending a method
from the class.
• Milestones throughout term: find group by Sept 17th and submit
half-page proposal by Oct 10th
6/6
Work
• Problem sets (50%)
• There will be 9 problem sets, you are only responsible for 8 (we’ll drop
the lowest score).
• Mix of analytic review questions and empirical applications of methods.
• Go out on Thursdays and due the following Wednesday, formatted nicely.
• Final collaborative project (40%)
• short (< 20 pg) research paper either applying or extending a method
from the class.
• Milestones throughout term: find group by Sept 17th and submit
half-page proposal by Oct 10th
• Participation (10%)
6/6
Work
• Problem sets (50%)
• There will be 9 problem sets, you are only responsible for 8 (we’ll drop
the lowest score).
• Mix of analytic review questions and empirical applications of methods.
• Go out on Thursdays and due the following Wednesday, formatted nicely.
• Final collaborative project (40%)
• short (< 20 pg) research paper either applying or extending a method
from the class.
• Milestones throughout term: find group by Sept 17th and submit
half-page proposal by Oct 10th
• Participation (10%)
• Asking questions in lecture, section, asking/answering questions on
Ed/Slack.
6/6
Work
• Problem sets (50%)
• There will be 9 problem sets, you are only responsible for 8 (we’ll drop
the lowest score).
• Mix of analytic review questions and empirical applications of methods.
• Go out on Thursdays and due the following Wednesday, formatted nicely.
• Final collaborative project (40%)
• short (< 20 pg) research paper either applying or extending a method
from the class.
• Milestones throughout term: find group by Sept 17th and submit
half-page proposal by Oct 10th
• Participation (10%)
• Asking questions in lecture, section, asking/answering questions on
Ed/Slack.
• Not really intended to hurt your final grade.
6/6
