Beamer Presentation
Build clean, content-first academic slides in LaTeX Beamer that are reproducibly linked to the paper. The default style is metropolis-themed, 16:9, with the same notation.tex and tabs//figs/ artifacts as the paper.
Operating Principles
- Slides share the paper's source of truth. Same
notation.tex, same tabs/*.tex, same figs/*.pdf. Rerun the analysis once; both PDF and slides update together. Never re-typeset a number on a slide.
- One claim per slide. If you need more, split.
- Headline, then evidence. The frame title states the claim; the body shows the evidence. Reviewers in the audience are skimming.
- Structure to time. A 20-minute talk is fundamentally different from a 90-minute seminar. Pin the time budget early; it determines depth, not just length.
- Backup, don't overload. Move detail into the appendix; reach for backup slides during Q&A. Better to be asked than to drown the talk in detail.
Decision Policy
This skill follows the repo-wide Agent Policy.
ASK before proceeding (blocking):
- Time budget (15-20 min conference / 45-60 min seminar / 75-90 min job-market talk / 10-15 min policy briefing).
- Audience expertise (specialists / broader economists / non-economists).
- The headline (one sentence the audience must remember).
- What goes in the appendix vs. the main deck.
DEFAULT + flag (use this default; tell the user how to override):
- Theme
metropolis, 16:9 aspect ratio, metroset{numbering=fraction, progressbar=frametitle}.
- Frame title states the claim, not the topic ("Treatment increases outcome by 8%" not "Results").
- Imports
notation.tex, tabs/, figs/ from the paper directory — single source of truth.
- Backup slides start at
\appendix; one tightly-focused slide per anticipated question.
- End on takeaways; do not end on "Questions?".
DOCUMENT and proceed (write into the deck preamble comment):
- Time-budget allocation (slides per section).
- Which paper artifacts the deck imports.
- Any slide-only number that does not come from
tabs/ or figs/ (rare; flag for review).
PROCEED items: slides imported from the paper's notation file; latexmk build; \Cref{} for cross-refs; \alert{} for the headline color (matched to figure palette).
Pre-flight Checklist
- Format. Conference (15-20 min), seminar (45-60 min), job-market talk (75-90 min), policy briefing (10-15 min), thesis defense?
- Audience. Specialists in the same subfield, broader economists, or non-economists?
- What's the headline? State it now in one sentence.
- What evidence is irreducibly part of the talk? (Identification, main result, robustness summary at minimum.)
- What goes in the backup? (Detailed robustness, mechanism evidence, alternative specifications.)
- Visual aids. Are there figures more compelling than tables? Almost always: yes.
Time Budget by Format
15-20 min conference talk
└── ~12-15 content slides
Motivation (2) Question (1) Strategy (2) Results (4-5)
Robustness (1) Conclusion (1) +backup
Rule of thumb: 1-2 minutes per slide; never more.
45-60 min seminar
└── ~25-35 content slides
Add: literature, institutional setting, mechanism, validation,
extended robustness. Spend longer on the identification frame.
75-90 min job-market talk
└── ~40-55 content slides + 15+ backup
Audience expects the full chain: motivation, contribution, model,
data, identification, results, mechanism, robustness, policy,
conclusion. Plan for 30+ min of Q&A.
10-15 min policy briefing
└── ~6-10 content slides; figures only.
Lead with the policy-relevant headline; minimize regression tables.
Project Layout (paper + slides share assets)
project/
├── paper/
│ ├── paper.tex
│ ├── tex/
│ │ ├── notation.tex # SHARED with slides
│ │ ├── preamble.tex
│ │ └── (sections)
│ ├── tabs/ # SHARED
│ ├── figs/ # SHARED
├── slides/
│ ├── slides.tex # \input{../paper/tex/notation.tex}
│ ├── preamble-beamer.tex
│ └── (frame files, optional)
└── code/
The slide deck imports the same notation and references the same tabs/ and figs/ paths. This is DIME's single source of truth principle applied across artifacts.
Frame Recipes
Frame title = the claim, not the topic
BAD: "Results"
BETTER: "Treatment increases outcome by 8% (p < 0.01)"
Build complex claims incrementally
\begin{frame}{Treatment effect grows over time}
\only<1>{\includegraphics[width=\textwidth]{figs/fig_event_static.pdf}}
\only<2>{\includegraphics[width=\textwidth]{figs/fig_event_with_ci.pdf}}
\only<3>{\includegraphics[width=\textwidth]{figs/fig_event_full.pdf}}
\medskip
\only<3>{\textbf{Takeaway:} effect rises monotonically through year~5.}
\end{frame}
Tables on slides should fit in 6-8 rows
For talks, regenerate a slide-friendly table from the same regression with esttab keep()/drop() options or a separate slides_tabs/ folder.
Common Pitfalls
- Reading the slide. The slide is for the audience; the script is for you.
- Tables with 12 rows of fixed effects. Cut to the headline coefficient + clearly labeled FE/N.
- Equations no one will read. Show the estimating equation if it carries the identification; otherwise drop it.
- Figures from screenshots of regression output. Always export from code.
- A title slide with five logos and seven affiliations. One affiliation, one date, one venue.
- "Questions?" as the final slide. End on the takeaway; questions are implicit.
- 9-point font in a footnote no one will read. Cut the footnote.
Additional Resources
reference.md — extended patterns: theme tweaks, metropolis customization, equation reveal patterns, Q&A appendix structure, multi-author title page, animated coefficient plots.
examples/ — copyable Beamer files:
examples/preamble-beamer.tex — packages, theme, custom colors, frame title style
examples/slides.tex — master deck \input{}-ing notation + frame files
examples/frame_motivation.tex — motivation/question slide
examples/frame_strategy.tex — identification slide with equation
examples/frame_results.tex — main result slide with figure + takeaway
examples/frame_robustness.tex — robustness summary slide
examples/frame_conclusion.tex — takeaways slide
Cross-Skill Routing
- For shared notation and the paper-side LaTeX scaffolding →
latex-econ-model skill.
- For slide-friendly versions of regression tables →
latex-tables skill (keep(), drop(), alternative mtitles()).
- For figures used on slides →
econ-visualization skill (width = 10 in, height = 5.5 in for 16:9).
- For the underlying paper draft →
academic-paper-writer skill.
References
1---2name: beamer-presentation3description: Builds academic economics presentations in LaTeX Beamer that share the same `notation.tex` macros and `tabs/`/`figs/` outputs as the underlying paper, so a single rebuild keeps slides and paper in sync. Defaults to clean themes (`metropolis` / `default` with custom colors), 16:9 aspect ratio, time-budget-aware structure (15/20/45/60/90 min), incremental reveals, and DIME-aligned reproducibility (figures/tables produced by code, never screenshotted). Cross-links to `latex-econ-model`, `latex-tables`, `econ-visualization`, and `academic-paper-writer`. Use when the user asks for conference talk slides, seminar presentations, job-market talks, thesis defense slides, or wants to convert a paper into Beamer.4---56# Beamer Presentation78Build clean, content-first academic slides in LaTeX Beamer that are reproducibly linked to the paper. The default style is `metropolis`-themed, 16:9, with the same `notation.tex` and `tabs/`/`figs/` artifacts as the paper.910## Operating Principles11121. **Slides share the paper's source of truth.** Same `notation.tex`, same `tabs/*.tex`, same `figs/*.pdf`. Rerun the analysis once; both PDF and slides update together. Never re-typeset a number on a slide.132. **One claim per slide.** If you need more, split.143. **Headline, then evidence.** The frame title states the claim; the body shows the evidence. Reviewers in the audience are skimming.154. **Structure to time.** A 20-minute talk is fundamentally different from a 90-minute seminar. Pin the time budget early; it determines depth, not just length.165. **Backup, don't overload.** Move detail into the appendix; reach for backup slides during Q&A. Better to be asked than to drown the talk in detail.1718## Decision Policy1920This skill follows the repo-wide [Agent Policy](../../AGENT_POLICY.md).2122**ASK before proceeding** (blocking):23241. Time budget (15-20 min conference / 45-60 min seminar / 75-90 min job-market talk / 10-15 min policy briefing).252. Audience expertise (specialists / broader economists / non-economists).263. The headline (one sentence the audience must remember).274. What goes in the appendix vs. the main deck.2829**DEFAULT + flag** (use this default; tell the user how to override):3031- Theme `metropolis`, 16:9 aspect ratio, `metroset{numbering=fraction, progressbar=frametitle}`.32- Frame title states the claim, not the topic ("Treatment increases outcome by 8%" not "Results").33- Imports `notation.tex`, `tabs/`, `figs/` from the paper directory — single source of truth.34- Backup slides start at `\appendix`; one tightly-focused slide per anticipated question.35- End on takeaways; do not end on "Questions?".3637**DOCUMENT and proceed** (write into the deck preamble comment):3839- Time-budget allocation (slides per section).40- Which paper artifacts the deck imports.41- Any slide-only number that does not come from `tabs/` or `figs/` (rare; flag for review).4243`PROCEED` items: slides imported from the paper's notation file; `latexmk` build; `\Cref{}` for cross-refs; `\alert{}` for the headline color (matched to figure palette).4445## Pre-flight Checklist4647- **Format.** Conference (15-20 min), seminar (45-60 min), job-market talk (75-90 min), policy briefing (10-15 min), thesis defense?48- **Audience.** Specialists in the same subfield, broader economists, or non-economists?49- **What's the headline?** State it now in one sentence.50- **What evidence is irreducibly part of the talk?** (Identification, main result, robustness summary at minimum.)51- **What goes in the backup?** (Detailed robustness, mechanism evidence, alternative specifications.)52- **Visual aids.** Are there figures more compelling than tables? Almost always: yes.5354## Time Budget by Format5556```5715-20 min conference talk58└── ~12-15 content slides59 Motivation (2) Question (1) Strategy (2) Results (4-5)60 Robustness (1) Conclusion (1) +backup61 Rule of thumb: 1-2 minutes per slide; never more.626345-60 min seminar64└── ~25-35 content slides65 Add: literature, institutional setting, mechanism, validation,66 extended robustness. Spend longer on the identification frame.676875-90 min job-market talk69└── ~40-55 content slides + 15+ backup70 Audience expects the full chain: motivation, contribution, model,71 data, identification, results, mechanism, robustness, policy,72 conclusion. Plan for 30+ min of Q&A.737410-15 min policy briefing75└── ~6-10 content slides; figures only.76 Lead with the policy-relevant headline; minimize regression tables.77```7879## Project Layout (paper + slides share assets)8081```82project/83├── paper/84│ ├── paper.tex85│ ├── tex/86│ │ ├── notation.tex # SHARED with slides87│ │ ├── preamble.tex88│ │ └── (sections)89│ ├── tabs/ # SHARED90│ ├── figs/ # SHARED91├── slides/92│ ├── slides.tex # \input{../paper/tex/notation.tex}93│ ├── preamble-beamer.tex94│ └── (frame files, optional)95└── code/96```9798The slide deck imports the same notation and references the same `tabs/` and `figs/` paths. This is DIME's [single source of truth](https://dimewiki.worldbank.org/Reproducible_Research) principle applied across artifacts.99100## Frame Recipes101102### Frame title = the claim, not the topic103104```105BAD: "Results"106BETTER: "Treatment increases outcome by 8% (p < 0.01)"107```108109### Build complex claims incrementally110111```latex112\begin{frame}{Treatment effect grows over time}113 \only<1>{\includegraphics[width=\textwidth]{figs/fig_event_static.pdf}}114 \only<2>{\includegraphics[width=\textwidth]{figs/fig_event_with_ci.pdf}}115 \only<3>{\includegraphics[width=\textwidth]{figs/fig_event_full.pdf}}116117 \medskip118 \only<3>{\textbf{Takeaway:} effect rises monotonically through year~5.}119\end{frame}120```121122### Tables on slides should fit in 6-8 rows123124For talks, regenerate a slide-friendly table from the same regression with `esttab` `keep()`/`drop()` options or a separate `slides_tabs/` folder.125126## Common Pitfalls127128- Reading the slide. The slide is for the audience; the script is for you.129- Tables with 12 rows of fixed effects. Cut to the headline coefficient + clearly labeled FE/N.130- Equations no one will read. Show the estimating equation if it carries the identification; otherwise drop it.131- Figures from screenshots of regression output. Always export from code.132- A title slide with five logos and seven affiliations. One affiliation, one date, one venue.133- "Questions?" as the final slide. End on the takeaway; questions are implicit.134- 9-point font in a footnote no one will read. Cut the footnote.135136## Additional Resources137138- `reference.md` — extended patterns: theme tweaks, `metropolis` customization, equation reveal patterns, Q&A appendix structure, multi-author title page, animated coefficient plots.139- `examples/` — copyable Beamer files:140 - `examples/preamble-beamer.tex` — packages, theme, custom colors, frame title style141 - `examples/slides.tex` — master deck `\input{}`-ing notation + frame files142 - `examples/frame_motivation.tex` — motivation/question slide143 - `examples/frame_strategy.tex` — identification slide with equation144 - `examples/frame_results.tex` — main result slide with figure + takeaway145 - `examples/frame_robustness.tex` — robustness summary slide146 - `examples/frame_conclusion.tex` — takeaways slide147148## Cross-Skill Routing149150- For shared notation and the paper-side LaTeX scaffolding → `latex-econ-model` skill.151- For slide-friendly versions of regression tables → `latex-tables` skill (`keep()`, `drop()`, alternative `mtitles()`).152- For figures used on slides → `econ-visualization` skill (`width = 10 in, height = 5.5 in` for 16:9).153- For the underlying paper draft → `academic-paper-writer` skill.154155## References156157- Shapiro, [How to Give an Applied Micro Talk](https://www.brown.edu/Research/Shapiro/pdfs/applied_micro_slides.pdf) — still the canonical reference.158- McKenzie, [Tips on Giving an Effective Conference Presentation](https://blogs.worldbank.org/impactevaluations/tips-giving-effective-conference-presentation).159- Beamer User Guide — https://ctan.org/pkg/beamer.160- Metropolis theme — https://github.com/matze/mtheme.161- DIME Analytics, [Reproducible Research](https://dimewiki.worldbank.org/Reproducible_Research) — single-source-of-truth idea applied to slides.