PS 8.3 — Draft Your JSSR Abstract and Poster Results Narrative Against the Style Guide
Course: ASSIP 8-Week Empirical Finance Camp · Week 8 · Problem Set 8.3 Covers: Ch 8.3 (Writing the Empirical Paper) — the four-move introduction (§2) compressed to the JSSR abstract, the one-sentence contribution (§3), the positioning literature review (§4), the headline-first results section (§5) compressed to your poster’s results panel, and the causal-language discipline (§7) — held to the binding table-and-prose standard of Appendix D (book/appendices/D-style-guide/; D.1 table layout and notes, D.2 coefficient/FE/cluster/sample disclosure, D.5 prose-style and causal-language rules). Builds on the frozen estimates from Ch 8.1 (specification curve) and Ch 8.2 (robustness), and the identification memo of Ch 7.5. Tool: nb8.3 (notebooks/week-08/nb8.3-publication-tables-figures.ipynb) — generates your headline regression table, the event-study figure, and the 200–300-word JSSR abstract to the Appendix D / JSSR standard, exported to LaTeX. You do not hand-format a single table or hand-type the abstract. What you turn in: a single Markdown (or LaTeX) file, abstract-and-results.md, containing two artifacts of your own ASSIP project — the JSSR-ready 200–300-word abstract (the one the poster’s title block and the JSSR submission both carry) and your poster results-panel narrative plus headline-table backing — plus a short reflection. This is a scaffolding assignment that produces a real deliverable for your own project. Every camper’s draft is different because every camper’s project is different. Students writing the optional companion full paper should still complete this PS; the abstract drafted here is the same abstract the companion paper carries — they cannot diverge.
Total: 100 points. Point values are stated per component and map onto the writing rubric Appendix E grades against. The two cross-cutting principles dominate the grade, both straight from Ch 8.3: front-load every verdict (the contribution leads the intro, the headline table leads the results, the topic sentence leads each paragraph — §1’s “say the important thing first, at every scale”), and match every verb to your weakest assumption (§7’s causal-language discipline). A draft that buries its contribution in paragraph four, or that lets “is associated with” drift into “causes” by the results section, loses points the same way a real manuscript loses a referee.
The one rule that makes this assignment mean anything. Your introduction must be readable by someone who will never read the rest of the paper (Ch 8.3 §2). A program officer skimming fifty abstracts, a professor deciding whether to assign your paper, a journalist hunting a number to quote — all of them stop at your introduction. If a reader who stops there cannot walk away with your question, your answer, and the appropriate degree of confidence in that answer, the introduction has failed its one job, and no polish on the body will rescue it. Write the front for the reader who never reaches the back.
A note on numbers. Unlike PS 7.3 (the pre-analysis plan, which is pre-results and therefore contains no estimates), this assignment is post-results: you have your frozen point estimate from Ch 8.1 and your robustness battery from Ch 8.2, and the whole point is to report them well. Report your real, frozen numbers — never numbers you wish you had, never numbers tuned after the fact. If you are drafting before your own analysis is final, you may use clearly-labeled illustrative numbers from nb8.3’s synthetic example (as the model deliverable does), but label them illustrative every time, exactly as the exemplar does. The cardinal sin is a table that has silently drifted from the analysis it claims to report — which is why nb8.3 generates your table from the frozen results rather than letting you retype them.
The model deliverable in book/appendices/E-solutions-manual/E-w8-ps8.3-solutions.md shows the A-grade standard on Maya’s HMDA fair-lending difference-in-differences (the project carried since Week 1, the identification memo of Ch 7.5, the robustness battery of Ch 8.2). Read it after you draft your own — not before — so it calibrates your standard without dictating your project. Your introduction should be as disciplined as Maya’s about your own question, not a paraphrase of hers.
Component 1 — The one-sentence contribution (15 points)
Before you write a paragraph, build the single sentence the whole introduction serves (Ch 8.3 §3). If you cannot write it, you do not yet have a paper — you have a regression and a hope.
(a) (8 pts) Write it, and name its job. Write your one-sentence contribution: the answer to what is true now that was not established before you did this work? Not what you did (“we run a difference-in-differences on HMDA”) but what is now known (“we provide the first quasi-experimental estimate of how state fair-lending examination programs affect the minority–white denial gap”). Then name which of the recognizable jobs from §3 it claims — first to credibly estimate, overturn/qualify an established finding, reconcile two literatures, measure a mechanism, or bring new data — and justify in one sentence that your literature review will support that claim rather than expose it (if you claim “first” and three papers did it already, you have just handed a referee a ten-second kill).
(b) (4 pts) Calibrate the verb to the design. State, in one sentence, why your contribution sentence uses the verb it uses, bounded by your weakest identifying assumption (§7). A clean staggered DiD with flat pre-trends earns “credibly estimate”; an OLS-on-observables with a plausible omitted variable you could not rule out earns only “document a robust association” or “provide suggestive evidence.” Quote your own contribution sentence’s main verb and name the assumption that licenses it.
(c) (3 pts) The stress test. Apply the §3 practical test: hand your contribution sentence to a classmate who has not seen your project and ask them to tell you, from that sentence alone, what you did and why it matters. Report what they said. If they asked “wait, what’s the question?” or “is that causal?”, revise the sentence and report the revision — those questions mean the sentence is doing the introduction’s whole job badly.
Component 2 — The four-move introduction (35 points)
Write your full introduction as the four moves of Ch 8.3 §2, in order: hook, one-sentence contribution (from Component 1, landing in the first or second paragraph), the “we find” preview, and the roadmap — with one bridge sentence of positioning into the literature. Roughly four to six paragraphs. Label each paragraph with the move it makes (you may delete the labels in a final version, but keep them for grading).
(a) (8 pts) The hook. Open with a real-world fact or tension — a number, a puzzle, a policy debate — that names the question and makes the reader care, in that order, fast. Do not open with “the literature has long studied.” The hook earns attention with consequence, not with a literature recitation. Two or three sentences. (Maya’s hook leads with the denial-gap number and the regulatory stakes, not with a citation.)
(b) (8 pts) The “we find” preview. Write the paragraph — usually beginning literally “We find that…” or “Our main result is…” — that previews the headline number, its sign, its rough magnitude in units a human can feel, and the key robustness fact, then names the assumption the interpretation rests on. Magnitude in human units means “about 1.8 percentage points, roughly a fifth of the pre-period gap,” not a bare coefficient. Students resist this paragraph because it feels like spoiling the ending; reread §1, there is no ending to spoil. Give the number up front; you spend the rest of the paper earning the right to have stated it.
(c) (6 pts) The positioning bridge. Write the one or two sentences that locate your contribution between the strands of literature, by contrast, not by listing — “our work sits at the intersection of two literatures that have not met: [strand A, which establishes X but cannot do Y] and [strand B, which does Y but cannot speak to Z].” This is a preview of Component 4; here it is the single bridge sentence that takes the introduction into the literature section.
(d) (5 pts) The roadmap. Write the short, functional last paragraph: “Section 2 positions our work in the literature; Section 3 describes the data; Section 4 lays out the design and our identifying assumption; Section 5 presents the main results; Section 6 reports robustness; Section 7 concludes.” A few sentences, no drama.
(e) (8 pts) Verb audit (the §7 discipline, made into a deliverable). After drafting, do the deliberate pass of §7: read only the verbs attached to your main result, across the hook, the preview, and the contribution sentence, and force every one down to the level your weakest identifying assumption permits. In a short paragraph, report the verbs you found and any you changed — e.g., “the preview originally said ‘the program reduces the gap’; my design is a DiD whose parallel-trends assumption is untestable beyond the pre-period, so I changed it to ‘the program reduced the gap, an estimate credible under parallel trends,’ with the assumption attached.” Catch drift (the abstract hedges, the results forget themselves) and catch over-hedging into mush (“may possibly perhaps tentatively suggest” has hedged itself out of any claim). Calibration cuts both ways: claim exactly as much as your design supports.
Component 3 — The positioning literature review (15 points)
Write the literature-review section of your paper — the part that draws the map so your gap is visible as a gap (Ch 8.3 §4). The failure mode this component grades against has a name: the annotated bibliography (“Smith finds A. Jones finds B. We contribute.”). That is a list, and a list does the reader no work.
(a) (9 pts) Group and contrast. Cluster the prior work into two or three strands — by question, by method, or by finding — and characterize each strand by what it established and what it could not, so the empty space on the map (your contribution) is constructed brick by brick out of the literature itself, not merely asserted. Cite primary sources with full bibliographic information per CONVENTIONS §6; if you cannot verify a citation, tag it [CHECK] rather than fabricate it (a hallucinated reference is the one error a referee catches in ten seconds and never forgives). Maya’s fair-lending positioning cites Sun & Gao (2019) on the same-sex-borrower margin as part of the strand that documents differential treatment.
(b) (3 pts) Every citation load-bearing. Confirm in one sentence that every paper you cite either builds part of the map’s wall or marks the edge of the hole — if a citation could be deleted without changing where the gap appears, delete it. A padded review signals you are counting references, not building an argument.
(c) (3 pts) Generous and precise, never dismissive. Confirm that you position prior work as “established X under conditions Y; we extend to where Y fails,” not “prior work was wrong.” The dismissive voice is the voice of someone about to be refereed by the author they just insulted.
Component 4 — The headline-first results section (25 points)
Write a results-section excerpt that leads with your headline table (Ch 8.3 §5). The cardinal sin here is the data dump: eight tables of equal weight, leaving the reader to find the important one. You found it for them in the introduction; do it again here.
(a) (12 pts) The headline table, generated by nb8.3, Appendix-D-compliant. Place your single main table — generated by nb8.3, not hand-typed — as the first object in the section. It must stand alone (Ch 8.3 §8): a reader seeing only the table can determine the sample, the estimator, the meaning of the coefficient, what the stars mean, and what fixed effects and clustering were used. Per Appendix D the table carries: 1. an informative title and complete notes defining the sample, units, estimator, and every symbol (D.1); 2. standard errors or t-statistics in parentheses directly beneath each coefficient, with the flavor of SE named and clustered on what (D.2) — “standard errors clustered by state in parentheses,” not a bare “(0.07)”; 3. significance stars defined in the notes (e.g., *** p<0.01, ** p<0.05, * p<0.10), never left to the reader’s guess; 4. a consistent, sensible number of significant digits (a coefficient is not more precise for six decimals); 5. dedicated rows disclosing fixed effects, controls, sample size N, and (within-)R^2 (D.2) so the specification is legible at a glance.
If you have a building-up table (column 1 raw, then adding controls and fixed effects, with your preferred specification flagged), so much the better — that is the standard nb8.3 produces.
(b) (9 pts) Walk the reader through it, in human units. In prose, point at the coefficient, state its sign, its magnitude in units a human can feel (“1.8 percentage points, about a fifth of the pre-period gap” — not just a number), and its statistical precision. Then interpret it economically: is the effect big enough to matter, not just big enough to be significant? Significance is not importance — a coefficient can be three-starred and economically trivial — and your prose must carry the economic interpretation the stars cannot. Lead with the headline; mention secondary results only after, and explicitly subordinate them.
(c) (4 pts) Causal language held to the design (the §7 audit, again). Confirm in one or two sentences that the verbs in your results prose match your weakest assumption — the place §7 warns drift bites hardest (“the results section forgets itself”). If your design is a DiD, the results say “the program reduced the gap” with the parallel-trends conditional somewhere in the section, never an unconditional “causes.” If your design is observational, the results say “is associated with” and stay there. Name the strongest verb you used and the assumption that earns it.
Submission checklist
Turn in one file, intro-and-results.md, plus the reflection below. Before you submit, confirm:
Reflection (ungraded but required, ~150 words). Answer Ch 8.3’s “Your Turn” prompt 3 for your project: take the headline table nb8.3 generated, cover up the surrounding prose, and list every thing a reader cannot determine from the table alone — the sample, the estimator, what the stars mean, the fixed effects, the clustering. Then say which of those you fixed in the table notes (following Appendix D) rather than in the paragraph, and why a table that needs its surrounding prose to be intelligible has failed its first and most important reader (the referee who goes to the tables first).
How this is graded
Appendix E grades against the writing rubric, with points mapped to the four components above. Two cross-cutting principles dominate, both straight from Ch 8.3:
Front-load every verdict. The contribution leads the introduction, the headline table leads the results, the topic sentence leads each paragraph (§1). A draft that builds suspense toward a reveal at the end — the high-school mystery-essay habit — loses points, because in a paper suspense is a bug: the reader is your judge, not your entertainment, and you make their job easy by stating the verdict first. Full marks require that a reader who quits after the first sentence of any unit has still learned the most important thing in that unit.
Match every verb to your weakest assumption. Your strongest verb is bounded by your weakest assumption (§7). A clean DiD with flat pre-trends earns “the program reduced the gap” with the assumption named; an OLS-on-observables earns only “is associated with.” The most-punished failure is drift — the abstract hedges correctly, the results section forgets itself, the conclusion claims “we have shown that X causes Y.” We run the same audit you ran on published authors in Week 5: underline every causal verb attached to the main result and ask “did the design earn this?” Run it on yourself before the grader does.
A note on the division of labor with Appendix D: this problem set tells you which moves to make and in what order; Appendix D is the binding spec for the table craft and prose style — exact table layout and notes (D.1), which coefficients and FE/cluster/sample lines to disclose (D.2), the prose-style and causal-language rules (D.5). Do not reinvent the standard here; point your draft at D and follow it. And do not hand-format your tables in a text editor — nb8.3 does it for you, reproducibly, from the frozen results, which is the entire point.
End of PS 8.3. When your intro-and-results.md is drafted, check your standard against the model deliverable in book/appendices/E-solutions-manual/E-w8-ps8.3-solutions.md — an A-grade introduction and a headline-first results excerpt for Maya’s HMDA fair-lending DiD, with a fully Appendix-D-compliant headline table and instructor grading notes on overclaiming, table craft, and the one-sentence contribution. Read it after you draft your own. The introduction you write here is the promise your whole paper defends; the results section is where, leading with the table, you begin to redeem it. Both live or die on the same two disciplines — say the important thing first, and claim exactly as much as your design supports, no more and no less.