Your occasional update on our network, upcoming events, professional development and the world of impact evaluation
Dear IEPN members,
April seems a bit late to say happy new year, but this is our first newsletter for 2026. We hope your year started with lots of impactful impact evaluation. The IEPN secretariat and conference committee have spent the first months of the year planning for an IEPN conference in the second half of 2026. We are seeking ideas and inspiration from our members. Please fill out the two‑minute conference survey to help us shape the best conference for you.
Our 2026 seminar series commenced in February with 2 presentations on structure learning and Bayesian adaptive trials (BATs). The seminars, presented by Professor Sally Cripps and Dr Anna Lopatnikova, generated lots of interesting questions from IEPN members. You can review the recordings and continue the discussion in our members’ Teams group. If you are not in the Teams group yet, please contact Tony Bertoia (tony.bertoia@edresearch.edu.au) to be added to the group.
We will be announcing more seminars soon, and we look forward to seeing you there.
Sincerely,
The IEPN Secretariat
Network update
IEPN Conference 2026
Planning for the first IEPN conference is underway. Thank you to everyone who expressed interest in joining the conference committee. We have a fantastic group of 12 network members who have joined the committee and brought great ideas with them. Our conference committee is:
- Alejandra Mendoza Alcantara
- Ansari Jainullabudeen
- Ashleigh Collier
- Bowen Fung
- Emma Tomkinson
- Gosia Bucki‑Smith
- Harrison Hansford
- Jakob Voo
- Luka Musicki
- Sam Williams
- Tony Bertoia
- Campbell McNolty.
We will have conference dates and more details to share with the network soon. In the meantime, you can help the committee by filling out a two‑minute survey.
Conference survey
The conference committee is seeking your ideas to shape the conference. Please complete our three‑question survey to help us plan a conference you will love.
Please be as ambitious as you like with your suggestions!
Training and professional development
Conference season is coming
In addition to the IEPN conference, several other impact evaluation‑related conferences have recently announced their keynote speakers. Here are the ones on our radar. If you are attending any of these, keep an eye out for your fellow IEPN members.
- Australian Conference of Economists in Canberra, 7–10 July
- The Labour Econometrics Workshop in Perth, 16–17 July
- Australian Public Finance Workshop in Canberra, 30–31 July
- Australian Evaluation Society International Evaluation Conference in Darwin, 14–18 September
Reading and links
If you find this reading list interesting or you would like to submit something for the next edition, please contact us at evaluation@treasury.gov.au
Bayesian adaptive trials for social policy: If you want to go deeper into the topic of our recent seminar, 2 of our presenters are co‑authors on this paper. It covers some of the same material as their seminar presentation.
LLMs love DiDs: Paul Goldsmith‑Pinkham blogs about using AI for academic research. It turns out, when AIs are asked to design an impact evaluation they reach for a difference‑in‑differences design more often than not.
Learning to implement a continuous DiD with the help of AI: Similar to Paul Goldsmith‑Pinkham, Scott Cunningham has been testing the boundaries of AI’s Impact Evaluation abilities. In this post, Scott steps through his approach to understanding a new DiD method, with the help of Claude Code.
The problem with event studies: In a recent blog post, Andreas Backhaus, discusses the potential bias in some event studies, even when pre‑trends are parallel. His discussion unpacks the findings from a recent paper on the wage penalty paid by mothers by Simon Bensnes, Ingrid Huitfeldt and Edwin Leuven.
For your machine learning friends: This article is pitched to data scientists, but it’s a cogent explanation of the reasons that models that make excellent predictions are not always useful for decision making. Spoiler, it’s because they don’t identify causal relationships.