Correction: ΔY1 _||_ D | X0,X1

@andrew.heiss.phd it is maybe too late and uses SWIGs rather than DAGs, but it speaks to your question raised here bsky.app/profile/andr...

Causal Graphs for Conditional Parallel Trends Difference-in-Differences (DiD) is a widely used research design that often relies on a conditional parallel trends (CPT) assumption. In contrast to settings with unconfoundedness, where causal graphs...

For the proof that we are allowed to do all that, practical recommendations and much more, see the paper arxiv.org/abs/2604.12818

Comments are very welcome!

(iv) with treatment–covariate feedback, all dynamic effect estimates are biased, reflecting omitted variable bias or “wrong world control bias”
(v) pre-trends do not diagnose post-treatment violations of CPT, while short-term effects can remain unbiased even with treatment–covariate feedback.

(i) pre-treatment covariates yield parallel pre-trends and unbiased short-term effects but biased dynamic effects
(ii) pre-outcome controls provide the same results as full sequence
(iii) in the absence of treatment-covariate feedback, strategies involving post-treatment variables are unbiased ...

We generalize to arbitrary time periods and provide a table that shows the minimal valid adjustment set under different causal structures.

This contains some useful insights that we illustrate in a simulation 👇

With three periods, graphs become more messy and independencies must be stichted together, but provide interesting insights.
If time Xt -> Dt, but Dt -> Xt+1, short term effects are identified, but dynamic effects not.

We illustrate why outcome dynamics rule out PT, unless more restrictions are imposed.

Y0 becomes a dilemma node that can only block one of two open paths between D and ΔYt

Δ-SWIGs are transformed Single World Intervention Graphs, which are themselves transformed DAGs. Good news: d-separation is all you need => path-blocking skills from DAGs apply.

E.g. ΔYt _||_ D | X1,X2 such that PT E[ΔYt | D=1, X1,X2] = E[ΔYt | D=0, X1,X2] holds

Also, conditioning on Y0 => M-bias

NEW PAPER 🚨
"Causal Graphs for Conditional Parallel Trends" with @henripf.bsky.social

It connects causal graphs and the modern Diff-in-Diff literature by introducing Δ-SWIGs as a graphical tool to reason about controls in DiD settings under standard additively separability assumptions.

Thread 🧵

I am hiring PhD students and/or Postdocs, to work on the theory of explainable machine learning. Please apply through Ellis or IMPRS, deadlines end october/mid november. In particular: Women, where are you? Our community needs you!!!

imprs.is.mpg.de/application
ellis.eu/news/ellis-p...

🚨Job alert🚨

Premium tenure-track (ass to full prof) in "𝐌𝐋 𝐌𝐞𝐭𝐡𝐨𝐝𝐬 𝐢𝐧 𝐁𝐮𝐬𝐢𝐧𝐞𝐬𝐬 𝐚𝐧𝐝 𝐄𝐜𝐨𝐧𝐨𝐦𝐢𝐜𝐬" @unituebingen.bsky.social made possible by @ml4science.bsky.social

Spread the word and feel free to reach out if you have any questions about the position or the environment.

Link: shorturl.at/QHZ5G

Me too 😄 I just pitched a vague idea to my research/art assistant and he came back with this.

The OutcomeWeights #RStats package now has a logo and a new vignette illustrating how Double ML improves covariate balance over "Single ML" RA or IPW.

Check it out:
mcknaus.github.io/OutcomeWeigh...

#causalSky #causalML

Wir suchen jemanden, der im WS 25/26 unseren vakanten Lehrstuhl für Statistik und Quantitative Methoden in den Wirtschaftswissenschaften vertritt. Auch die Dauerstelle wird bald ausgeschrieben. Wir freuen uns über nette, engagierte Kollegen! Gerne teilen... Link im nächsten Post.

Our cluster Machine Learning for Science is up for 7 years more funding!

Excellent news! The "Machine Learning for Science" cluster is an incredible public good for researchers @unituebingen.bsky.social interested in ML in all its facets.
Great job by @philipp.hertie.ai, @ulrikeluxburg.bsky.social and the cluster team.

The members of the Cluster of Excellence "Machine Learning: New Perspectives for Science" raise their glasses and celebrate securing another funding period.

We're super happy: Our Cluster of Excellence will continue to receive funding from the German Research Foundation @dfg.de ! Here’s to 7 more years of exciting research at the intersection of #machinelearning and science! Find out more: uni-tuebingen.de/en/research/... #ExcellenceStrategy

Junior Professorship (W1) with tenure track (W3) in Statistics and Empirical Economics #Econsky

🧵New survey paper: "Inference with Few Treated Units"
Luis Alvarez, Bruno Ferman and Kaspar Wüthrich

Tired of referees saying your standard errors are wrong?

This survey will help you understand if you really have a problem — and, if so, how to fix it!

One of my favorite parts is running OLS within the DoubleML package of @philippbach.bsky.social and colleagues.
Of course this is unnecessarily complicated, but instructive.

How can we design algorithms that maximize social welfare, rather than profits? This paper merges multi-armed bandits and adversarial learning with optimal tax theory and welfare economics. @maxkasy @NicoloCB @Rcolomboni buff.ly/pxPY8nj

@dariia.bsky.social

Dariia Mykhailyshyna - Workshops for Ukraine Feedback on the past workshops (if you want to learn how to make wordclouds, check out Text Data Analysis workshop below)

If you find this interesting and want the recording of the session, you can still get access for a donation via sites.google.com/view/dariia-...

One of my favorite parts is running OLS within the DoubleML package of @philippbach.bsky.social and colleagues.
Of course this is unnecessarily complicated, but instructive.

One year ago I gave a #CausalML Workshop for Ukraine 🇺🇦
We hand-coded DoubleML and causal forest in very few lines of code to exactly replicate their package outputs.
If you better understand theory through coding like me, check it out.
You find the R notebook now online: shorturl.at/uM82n
#RStats

Foundations of Machine Learning Research on machine learning, experimental design, economic inequality, and optimal policy

🤖 Interested in machine learning, economics, and the state of AI?🤖

In September, I will teach a 1-week intensive version of my course on foundations of ML (maxkasy.github.io/home/ML_Oxfo...) in our summer school.

Apply here: ouess.web.ox.ac.uk/september-su...

Spread the word!

I am a fan of this one, though it is not diverging but converging: doi.org/10.1016/j.ec...
It is so obvious where the policy change happens that it is not even indicated...

Celebrating Philip G. Wright’s legacy: directed acyclic graphs and instrumental variables - Royal Economic Society The Econometrics Journal published a special issue celebrating the seminal contributions of Philip G. Wright (1928).

#EctJ published a SI celebrating the seminal contributions of Philip G. Wright to causal inference in economics. Both our editorial & reproduction of Wright (1928) are free to read.

@resmedia.bsky.social @p-hunermund.com @borusyak.bsky.social @instrumenthull.bsky.social

res.org.uk/celebrating-...

Interesting new paper: arxiv.org/pdf/2503.09907

improves on both rdrobust and rdhonest!

Quite compelling... 1/n