Online Causal Inference Seminar

A regular international causal inference seminar. Sign up to our mailing list to receive announcements.

All seminars are on Tuesdays at 8:30 am PT / 11:30 am ET (3:30 pm UTC through Oct 31; 4:30 pm UTC from Nov 1 through year-end).

Zoom link and other details are provided below. Past talks are available here. Recordings of past webinars are available on our YouTube channel (subscribe to get notified!).

Upcoming Seminars

• Tuesday, Nov 03, 2026: Young researchers' seminar
  - Time: 8:30 am PT / 11:30 am ET / 4:30 pm UTC / 12:30 am (+1day) Beijing time
  - Details: Zoom link, Meeting ID: 968 8371 7451, Passcode: 414559
  - Speaker 1: Gellért Perényi (EPFL)
  - Title: Causal Inference from Self-controlled Case Series
  - Abstract: The self-controlled case series (SCCS) design is widely used to study effects on rare events, particularly in medical settings. The design requires data only from individuals who have experienced the event of interest, so-called cases, and makes within-person comparisons to control for confounding. This makes the design useful when only case data are available, for example from surveillance systems or clinical records. However, the conventional motivation for the design does not justify a causal interpretation.
  In this work, I describe the SCCS design from a causal inference perspective, and give sufficient assumptions to identify the causal effect of treatment on an outcome. I will explain when and why the design permits identification of causal effects, and how restricting the analysis to cases need not introduce bias. I will then present statistical tests for the existence of a treatment effect that remain valid under only a subset of the identification assumptions. These assumptions are more plausible in several practical settings.
  - Speaker 2: Lars van der Laan (Stanford University)
  - Title: Automatic Debiased Machine Learning for Functionals of M-estimands
  - Abstract: Debiasing is often needed for inference when nuisance functions are estimated using flexible machine learning methods. Automatic debiased machine learning (autoDML) seeks to automate first-order bias correction for low-dimensional targets, bypassing problem-specific derivations. We develop a general autoDML framework for estimands that are smooth functionals of a generic nonparametric M-estimand. The key structural insight we exploit is that the local curvature of the risk defining the M-estimand induces a Hessian inner product under which the Riesz representation of the target’s local sensitivity determines a canonical debiasing direction. This direction can itself be estimated through an empirical Riesz criterion constructed from the Hessian of the defining loss and the derivative of the target functional. The framework recovers existing regression-based autoDML procedures as special cases while extending to smooth, possibly nonlinear, functionals of general M-estimands. We derive a functional von Mises expansion, establish conditions for pathwise differentiability, and characterize the corresponding nonparametric efficiency bounds. We develop one-step, targeted minimum loss-based, and sieve plug-in estimators, and show that for linear targets and quadratic risks a doubly robust error structure arises naturally. We illustrate the framework through estimation of long-term survival probabilities from short-term follow-up data under a beta-geometric survival model.

• Tuesday, Nov 10, 2026:
  - Speaker: Emilija Perković (University of Washington)
  - Time: 8:30 am PT / 11:30 am ET / 4:30 pm UTC / 12:30 am (+1day) Beijing time
  - Details: Zoom link, Meeting ID: 968 8371 7451, Passcode: 414559
  - Title: Lost in Aggregation: The Causal Interpretation of the IV Estimand
  - Abstract: Instrumental variable methods are one of the standard tools for estimating causal effects from observational data when confounders cannot be measured. The treatments in many prominent IV studies are aggregates: years of education, GDP, caloric intake. Only the sum is observed, and its components may have different effects on the outcome. This talk asks what the IV estimand means in that setting. We first show that the causal effect of an aggregate treatment is not well defined on its own. An intervention on the sum must also specify how the components change, and different specifications give different effects. We formalize this choice as an aggregate-constrained component intervention distribution (ACID) and characterize when the IV estimand coincides with the resulting causal effect. We connect the result to treatment variation irrelevance, ambiguous manipulations, and the long history of aggregation problems in economics.
  Joint work with Danielle Tsao, Krikamol Muandet, and Frederick Eberhardt.
  [Paper][Slides][Video]

• Tuesday, Nov 17, 2026:
  - Speaker: Jonathan Roth (Brown University)
  - Time: 8:30 am PT / 11:30 am ET / 4:30 pm UTC / 12:30 am (+1day) Beijing time
  - Details: Zoom link, Meeting ID: 968 8371 7451, Passcode: 414559
  - Title: Simple Standard Errors for Complicated (Quasi-)Experimental Designs
  - Abstract: We develop standard errors for estimates of causal effects in settings where the distribution of treatment exposures is known but potentially exhibits complicated dependence patterns across units. Such dependence arises, for example, in settings with spillovers/interference across units or when the treatment assignment is spatially correlated. Our variance estimator generalizes several properties of the Neyman variance estimator for completely randomized experiments to general designs: it is unbiased for an upper bound on the variance from the design-based perspective; the conservativeness of this upper bound depends on the heterogeneity of treatment effects; and it is also valid under certain forms of super-population sampling of the potential outcomes. Our estimator reduces to standard approaches in simple designs, such as the Neyman variance estimator in completely randomized experiments and the cluster-robust variance estimator in cluster-randomized experiments. We show that the proposed estimator is admissible among the class of quadratic-form variance estimators, and derive asymptotic conditions under which it is consistent for an upper bound on the asymptotic variance of the treatment effect estimator. In simulations and applications based on two empirical studies, one with peer effects and another with a spatially correlated treatment, we find that the proposed variance estimator performs well, and in some cases yields standard errors that are 2-3 times shorter than those of existing alternatives.
  - Discussant: P. M. Aronow (Yale University)
  [Paper][Slides][Video]

• Tuesday, Nov 24, 2026: No seminar. Happy Thanksgiving!

• Tuesday, Dec 01, 2026:
  - Speaker: Tom Claassen (Radboud University)
  - Time: 8:30 am PT / 11:30 am ET / 4:30 pm UTC / 12:30 am (+1day) Beijing time
  - Details: Zoom link, Meeting ID: 968 8371 7451, Passcode: 414559
  - Title: Anytime-anywhere FCI+ : a true ‘anytime’ algorithm for causal discovery with high-density regions
  - Abstract: Applying causal discovery algorithms to real-world data can be challenging, especially when unobserved confounders may be present. In theory, constraint-based approaches like FCI are able to handle this, but implicitly rely on sparse networks to complete within a reasonable amount of time. However, in practice many relevant systems are anything but sparse. For example, in protein-protein interaction networks, there are often high-density nodes (‘hub nodes’) representing key proteins that are central to many processes in the cell. Other systems, like electric power grids and social networks, can exhibit high-clustering tendencies, leading to so-called small-world networks. In such cases, even basic constrained based algorithms like PC can get stuck in the high-density regions, failing to produce any useful output. To tackle this, the famous Anytime-FCI (Spirtes,2001) was designed to interrupt the search at any stage and then produce a sound causal model output a.s.a.p., but unfortunately can still require a long time to finish. The same holds for more recent alternatives like ICD (Rohekar et al.,2021).
  In this talk I will present a radically new search strategy for causal discovery in graphs with high-density regions. It allows to build up the causal graph on the fly, updating the model after each validated edge removal, while remaining sound throughout. It exploits the intermediate causal model output to efficiently reduce size and number of the conditional independence tests, and automatically prioritizes ‘low-hanging fruit', leaving difficult (high-density) regions until last. The resulting ‘anytime-anywhere FCI+’ is a true anytime algorithm that is not only faster than its traditional counterparts, but also more flexible, and can handle arbitrary edge removals as well, opening up new possibilities for e.g. targeted cause-effect recovery in large graphs.
  [Slides][Video]

• Tuesday, Dec 08, 2026:
  - Speaker: Eli Ben-Michael (Carnegie Mellon University)
  - Time: 8:30 am PT / 11:30 am ET / 4:30 pm UTC / 12:30 am (+1day) Beijing time
  - Details: Zoom link, Meeting ID: 968 8371 7451, Passcode: 414559
  - Title: AI-assisted design and analysis of experiments with unstructured treatments
  - Abstract: Randomized experiments involving complex, unstructured treatments such as text, images, or video, are increasingly common in social science research. The goal of many of these studies is to estimate the effect of a structured, interpretable attribute that generalizes beyond the treatments used in the experiment (e.g., the presence of a linguistic or visual feature). There are two key challenges when conducting such studies: observed effects of the attribute of interest (i) may be confounded by other correlated attributes, leading to poor internal validity; and (ii) may not generalize beyond the treatments used in the experiment, leading to poor external validity. We propose a framework for designing experiments with unstructured treatments that leverages generative AI or large databases to sample treatments in a way that explicitly navigates the tradeoff between internal and external validity. We show how to conduct conservative asymptotic inference with a difference-in-means estimator, and develop a calibrated AI-assisted estimator that incorporates zero-shot AI predictions to improve precision without introducing bias. Throughout, we work in a finite population framework, where treatment assignment is the only source of randomness, and treat AI generations and predictions as black-boxes with no guarantees on their performance. We illustrate and evaluate our approach by designing two hypothetical experiments, one with text and another with images.
  - Discussant: Naoki Egami (MIT)
  [Paper][Slides][Video]

Recordings of our past webinars are available on YouTube. Follow us on YouTube to stay notified!

Mailing List

To stay up-to-date about upcoming presentations and receive Zoom invitations please join our mailing list. You will receive an email to confirm your subscription. If you are already subscribed to our mailing list and would like to unsubscribe, you may unsubcribe here.

Suggest a speaker

If there is anyone you would like to hear at the Online Causal Inference Seminar, you may let us know here.

Opportunities in Causal Inference

Please check out our opportunities in causal inference page for conferences, workshops, and job listings! If you would like us to list an opportunity, please email us at onlinecausalinferenceseminar@gmail.com.

Youtube channel

Recordings of our past webinars are available on YouTube. Follow us on YouTube to stay notified!

Instructions for Attendees

You can join the webinar by clicking the link on the webpage. If you signed up to the mailing list, you will receive an email with the link before the webinar begins. On Tuesday, you should join the seminar shortly before the start time 8:30 am PT.

Please participate during the seminar!

Due to high demand, we will host the seminar as a Zoom webinar. As an attendee, you will not be able to unmute yourself. If you have questions about the content of the talk, please submit the questions using the Zoom Q&A feature. Time permitting, and depending on the volume of questions, the moderator will either ask your question for you or confirm with you to ask the question yourself and unmute you at a suitable time. In some meetings, the collaborators of the speaker will be online to address your questions in Q&A. Note that Q&A will be moderated by us so you will only be able to see some of the questions of the other attendees. If you want to send messages to the moderators during the seminar, please use the Zoom chat feature. 

Zoom instructions

If you have not used Zoom before, we highly recommend downloading and installing the Zoom client before the meeting. Additional instructions on how to use Zoom during a webinar can be found here. Note that for the online causal inference seminar, we do not require registration in advance so you will be able to join by simply clicking the link on this webpage or in the email.

If you have further questions, please drop us an email at onlinecausalinferenceseminar@gmail.com