This course offers a rigorous mathematical survey of causal inference at the Master’s level. Inferences about causation are of great importance in science, medicine, policy, and business. This course provides an introduction to the statistical literature on causal inference that has emerged in the last 35-40 years and that has revolutionized the way in which statisticians and applied researchers in many disciplines use data to make inferences about causal relationships. We will study methods for collecting data to estimate causal relationships. Students will learn how to distinguish between relationships that are causal and non-causal; this is not always obvious. We shall then study and evaluate the various methods students can use — such as matching, sub-classification on the propensity score, inverse probability of treatment weighting, and machine learning — to estimate a variety of effects — such as the average treatment effect and the effect of treatment on the treated. At the end, we discuss methods for evaluating some of the assumptions we have made, and we offer a look forward to the extensions we take up in the sequel to this course.
LB
A good course. Lot's of insights on Propensity Score Matching. They show good references to those willing to read some articles. Although quick classes, exercises are easy and very practical.
MV
Assignments are a mess, and apparently haven't been fixed for years after multiple complaints. Otherwise a good course, although not better than the one from U of PA, which was more accessible IMO.
PV
Great course. Really interesting and condensed content. However, It was difficult to follow lectures without any kind of reading and there wasn't any support on the discussion forums.
BS
!!!! very useful, professor is very professional, and course has high value!
Showing: 20 of 41
Not having access to slides and materials negates any interest in proceeding with this course.
It was difficult to follow lectures without any kind of reading
The first week is a throw-away, as there are no slides, just a talking head throwing notation at you. The second week at least has a blackboard, but then the assessment is broken.
It is impossible to learn statistics without slides in the first week.
Great course. Really interesting and condensed content. A perfect course for analysts and data scientists. I will be recommending this to a few of my colleagues.
For some reason there are no slides in week 1 but don't worry there are slides from week 2 onwards
I find it incredible that a course discussing a topic using complicated subscripted variables such as Y sub i sub Z would not use the equivalent of a whiteboard, but would instead try to communicate these concepts vocally. Using a professor who speaks in a monotone.
I am also surprised that apparently no attempt was made to make the suggested readings from the literature available on the web.
Please add slides or some teaching aids. This course is otherwise not usable
Talking head is not the best way to present for presenting such subjects.
Guy reading lectures? Give me a break. This has to be the worst course ever.
It is a very good intro to CI with proofs and references to recent developments.
However, I have to subtract some stars because the quality in material preparation of this course is not up to usual Coursera standards: for the first week there are no slides (so it's hard to follow), and some answers in the exams are not correct. This has been pointed out on this course's discussion forums, but nobody involved in the preparation of this course replies on its discussion forums.
I can't feel any efforts the lecturer had made to help us understand the topic.
A good course. Lot's of insights on Propensity Score Matching. They show good references to those willing to read some articles. Although quick classes, exercises are easy and very practical.
The selection of material is excellent and the professor covers an amazing amount of ground in a handful of lectures. Currently, however, there are many superficial problems with the course, including repeated errors in the quizzes and lectures that are confusing because the slides are missing.
Topic/syllabus/reference material: 5 stars - a great intro to CI (Rubin's approach)!!
Learning material: 2 stars (talking head, slides not provided, typos).
Assessment: 1 star (not particularly engaging and full of mistakes like correct answers scored as incorrect or calculations expected to be done with data different from that provided).
In my experience, this is a course where knowledge is obtained in another way and from outside the course. Confusing and there is no proper, ethereal exposure. This is my exclusive opinion. And for me, it is very sad to take an absolutely useless course, which is why I decide to drop out so as not to waste time.
A talking lecturing on mathematics and statistics without any equations or slides. It's impossible to follow and the worst form of pedagogy. Coursera and Columbia should remove this course. It's embarassing.
More exercises would be better!
Nice overview of topic but many quiz answers were wrong and haven't been corrected in over 5 years.
The layout of the slides is easy to lose people. There are too many errors in the quiz and no one has ever tried to correct them even though some students have been pointing them out for years.
the teaching videos are kind of boring