We have all heard the phrase “correlation does not equal causation.” What, then, does equal causation? This course aims to answer that question and more! Over a period of 5 weeks, you will learn how causal effects are defined, what assumptions about your data and models are necessary, and how to implement and interpret some popular statistical methods. Learners will have the opportunity to apply these methods to example data in R (free statistical software environment). At the end of the course, learners should be able to: 1. Define causal effects using potential outcomes 2. Describe the difference between association and causation 3. Express assumptions with causal graphs 4. Implement several types of causal inference methods (e.g. matching, instrumental variables, inverse probability of treatment weighting) 5. Identify which causal assumptions are necessary for each type of statistical method So join us.... and discover for yourself why modern statistical methods for estimating causal effects are indispensable in so many fields of study!
KS
My work involves working with observational data. This course taught me to think in more formal and organized way on topics and questions of causal inference.
YS
This is a great course to me! This course really helps me have a better understanding of what constitutes causal effects. I really appreciate him for this course!
GB
Excellent video lectures. Challenging end of module quizzes. I found more challenging doing the practical exercises because I had no experience with R.
PH
I completed all 4 available courses in causal inference on Coursera. This one has the best teaching quality. The material is very clear and self-contained!
WL
Very easy to follow examples and great coverage for such an important topic! The delivery sometimes get repetitive and I wish we talked more about how the uncertainties are derived.
FF
The material is great. Just wished the professor was more active in the discussion forum. Have not showed up in the forum for weeks. At least there should be a TA or something.
CE
Works best on double speed (from settings menu of each video). Content is delivered in clear and relatable manner using interesting real world examples.
MV
A consise course on causality; watched on 2x speed because the instructor speaks rather slowly; really bad formatting of quiz questions.
FW
Great class！ I have learned a lot on causal inference to conduct experiment analysis at work. The R coding sessions and lectures on the logic/math behind are really helpful.
OB
Great course! I am glad i came accross it. Helped me a great deal with my project at work. I wish there were more courses by this professor.
YZ
It will be better to give reviews of related applications in specific AI areas (e.g, computer vision, NLP, etc.) at the end of each of the sections of the lesson.
WJ
Great introduction on the causal analysis.The instructor did a great job on explaining the topic in a logical and rigorous way. R codes are very relevant and helpful to digest the material as well.
Showing: 20 of 184
The material is great. Just wished the professor was more active in the discussion forum. Have not showed up in the forum for weeks. At least there should be a TA or something.
I completed all 4 available courses in causal inference on Coursera. This one has the best teaching quality. The material is very clear and self-contained!
Brief yet comprehensive crash course in causality, which introduces the controlling vs. setting, causal assumptions, incident user design, active comparator design, directed acyclic graph (DAG), backdoor path criterion and disjunctive cause criterion for selecting variables to control for, propensity score matching, IPW matching, doubly robust estimators, instrumental variable analysis, two-stage least squares (2SLS), intention-to-treat (ITT) analysis, as well as sensitivity analysis, with R packages and scripts available! Super helpful and highly recommended!
A must for anyone interested in causal effect estimation. The professor is throughout with the content, he doesn't go too fast and too slow, and relies heavily in explaining the intuition behind the methods. It'd be great if he could do a second course on this with the more advanced topics mentioned but left out, like sensitivity analysis for propensity score, IPTW and IV, that are requiered for those writing papers.
Excellent introduction course. Jason Roy is an incredibly talented teacher.
I enjoyed the course and learned basics of causal inference. What I missed was more exercises with R in order to gain more practical understanding of the material. In particular, it would be great to have exercises where you get some dataset and your task is to calculate given causal effect and you need to come up with an approach and to execute it. This would mimic more closely problems that you encounter in practice.
The course material is excellent, but the course description dramatically under-estimates the study time needed to complete the course. This is especially true for the R assignments if you are not already *very* comfortable in R. There are also many problems with link rot and software/version compatibility issues for the R exercises.
I would have given the course 4 stars were it not for the unforgiving nature of the R exercises.
Overall, I would recommend this course for someone if they are already quite comfortable in R, or are willing to pout in at least 20 hours of work for each of the R assugnments.
This is a very theoretical course with much math formula and less well-explained practical examples to better illustrate those formula. I came to this course hoping to learn about new ideas and techniques of experiment design for causal effect when randomized experiments are not possible. Unfortunately I did not achieve this goal. This is just my personal view. If you come with a different purpose, you might find this course more useful than I did.
Fantastic instructor with lessons accessible for both those with some background wanting to brush up and for newcomers. Note that the programming assignments are in R and one uses a fixed random seed so it will be difficult to complete the assignments in another language. That said, the data are available so you can play with the same concepts in another language outside of the assignments. Certainly recommend.
Excellent course! The lectures are very clear and easy to follow, and Professor Roy is really good at explaining the concepts in a simple way. The assignments in R are helpful for grasping the theoretical concepts. I would specially recommend this course to data scientist, who might be interested in complementing their predictive analytics skills with the the necessary ones to tackle questions about causality.
This is one of the best online course I have taken so far, Jason is a very good instructor and he explains everything clearly in an easy to understand manner. I have tried another course on a similar topic on Coursera but I simply gave up on the other one. This course provides concrete examples and exercises, it allows me to understand the topic in fine details. I highly recommend this course.
Excellent course and lecturer. The lecturer takes his time to explain everything in a smooth speed. Is easy to follow. Good exercises and quizzes. I am quite satisfied with the course. I am looking for more (advance) courses from the same lecturer about the same subject, but also other subjects.
I enjoyed the course a lot and I think I took a lot from it as well. The quizzes and computer projects were appropriate, and the resourcees posted were very useful.
Works best on double speed (from settings menu of each video). Content is delivered in clear and relatable manner using interesting real world examples.
This course is quite useful for me to get quick understanding of the causality and causal inference in epidemiologic studies. Thanks to Prof. Roy.
I learned so much from Dr. Roy by watching his great lectures. Thank you!
In the beginning the course to me was quite difficult, as it has a different perspective on statistics I was used to. Most people tend to say: "correlation is not causality". When it came to propensity scores, matching and so on the possibilities became more clear to me to apply these methods in practice. The pace of the videos is slow, so I played the videos in 1.5 of the time. What I missed was the ability to download the slides. The instructor would look into this, but we're still waiting several weeks later. Another thing I missed was any sense how many other students were in the course.
I expected more from this course. It gets too deep into the more advanced topics without using specific examples to showcase the main ideas. The instructor could also be more engaging, I had to watch the videos at x1.25 to be able to keep my attention on them
I am sure the instructor is very knowledgeable and excellent in front of a class. His style does not work online.
I thought that this was really an excellent course. Although it was presented solely as lecture slides, the professor really took great care in presenting motivations for techniques and then the techniques themselves - as well as conceptual material (e.g., potential vs. actual outcomes, conditioning vs. setting, etc.) that were crucial to grasping the big picture. My only, minor, suggested tweak might be to possibly make the quizzes a little more challenging - perhaps by giving some numerical problems to solve. But overall, really just a great class.