Lecture notes. Written from the material of Berkeley Stat 156, licensed CC BY-NC 4.0. These are notes, not a transcript: the material has been reorganised and rewritten. This adaptation carries the same licence, and the original is linked above.
12. Paper Presentation Assignment¶
What this covers¶
This chapter is not a lecture. It reproduces the course's paper-presentation assignment, one of Stat 156/256's two graded pieces of work: each student records a video presenting two published causal-inference papers. It is kept here, rather than dropped, because a reader who has worked through this book has the tools to attempt the same exercise on a paper of their own choosing.
The assignment¶
You present two published papers, each in its own recorded video presentation of about 20 minutes. The goal is twofold: to engage thoughtfully with academic research in causal inference, and to hone presentation skills. Use slides or a virtual whiteboard to carry notation, mathematical expressions, graphs, and figures — the talk should stand on its own as a piece of exposition, not just a reading of the paper.
What counts as an admissible paper¶
- The paper must be related to causal inference.
- It must be published after 2012.
- At least one of the two papers must be a journal paper (as opposed to a conference paper).
Recommended venues¶
Journals: Annals of Statistics, Journal of the American Statistical Association, Journal of the Royal Statistical Society Series A, B, and C, Biometrika, Biometrics, Annals of Applied Statistics, Econometrica, Journal of Econometrics, Data & Policy, Journal of Machine Learning Research, PNAS, Science, Nature.
Conferences: AISTATS, ICML, NeurIPS, and the ACM Conference on Fairness, Accountability, and Transparency (FAccT).
Grading criteria¶
Presentations were evaluated on three criteria, in the course's own terms:
- Content. Clearly describe and motivate the problem setting, and situate the research in the context of prior work. Spend a significant portion of the talk on the paper's proposed contribution, both conceptually and in mathematical detail, and report any empirical findings. Engage critically with the paper: offer your own commentary on its novelty, strengths, weaknesses, and significance.
- Presentation. The talk should be accessible and informative to someone who has not read the paper but has taken the causal inference course. Emphasize the key aspects of the research, give intuition or motivation where they are complex or non-obvious, define new terms, use consistent language, and pace the talk appropriately.
- Relevance. The paper must meet the admissibility criteria above — causal inference, published after 2012, and so on. A paper that does not loses points on this criterion regardless of how well it is presented.
Sources¶
VideoGuidelines.md(Stat 256 Paper Presentation Guidelines), Berkeley STAT 156/256, Fall 2024 — the whole of this chapter. The source is itself a model reconstruction of a PDF with no text layer, so treat the wording above as a faithful paraphrase of that reconstruction rather than a verbatim quotation of the original assignment sheet.