A WORKED SAMPLE REPORT
Follow one dataset through its comparison, fitted estimate and diagnostic checks. Every reported number below comes from the shared sample or its saved DiD-Acto analysis.
Source: the shared rent-control benchmark, a simulated dataset created for practice. The estimates are computed by DiD-Acto; this is not a study of an actual rent-control policy.
SAVED ANALYSIS · EXPLORATORY · QUALIFIED
95% confidence interval: −$92.58 to −$84.48
Monthly median rent in the treated areas was about $88.53 lower than the level implied by the comparison group’s change, averaged over 2021–2022.
This is a difference in monthly median rent, averaged across ZIP areas and post-adoption months. It is not a percentage change or the rent paid by every household.
Treated-area rents still rose by $14.29 between the pre- and post-period averages. Comparison-area rents rose by $102.82. The estimate captures the difference between those changes.
01 / FOLLOW THE DATA
The sample follows 100 ZIP areas for 48 months. Half adopt in January 2021; the other half stay untreated. Each row records one area and month.
| Group | Before: 2019–2020 | After: 2021–2022 | Change |
|---|---|---|---|
| 50 treated ZIP areas | $1,619.56 | $1,633.86 | +$14.29 |
| 50 comparison ZIP areas | $1,666.94 | $1,769.76 | +$102.82 |
These four means reproduce the fitted TWFE coefficient in this balanced, unadjusted common-adoption design. For staggered rollouts or covariate-adjusted studies, the estimator may use different comparisons and weights.
02 / UNDERSTAND THE METHOD
All treated areas adopt at the same time, and the panel is complete. The planner selected TWFE with ZIP and month effects. It adjusts for persistent differences between areas and common monthly movements, using the recorded treatment indicator.
The plan uses no covariate adjustment. It declares no anticipation or spillovers and a common untreated trend, matching the benchmark’s stated design. In your own study, the rollout history supplies the reasoning for those choices.
The estimated effect is −$88.53, with a standard error of $2.04. The 95% interval runs from −$92.58 to −$84.48; all of that range represents lower rent relative to the comparison change.
Clustering by ZIP allows observations within an area to be correlated over time. The interval uses a t reference with 99 degrees of freedom.
The benchmark documents an imposed effect of −$85. That value falls inside this run’s interval. The known effect makes this sample useful for checking the calculation against its design.
03 / READ THE CHECKS
| Check | Saved finding | How to read it |
|---|---|---|
| Data and support | 4,800 observations; no validation errors or warnings; no trimmed cells. | All 100 areas contribute across the 48-month panel. |
| Joint pretrend test | F(5, 99) = 1.676; p = 0.147. | The five tested leads, months −6 through −2, are not jointly rejected at 5%. Read this with the earlier trend pattern and the study history. |
| Placebo in time | +$0.97; p = 0.870. | Moving adoption two months earlier and using pre-treatment data produces a small, imprecise placebo effect. |
| Cross-method agreement | Not assessed: no second estimate with the same overall target. | The event study describes a monthly path; it is not an independent replication of the overall effect. |
| Formal sensitivity | Not produced for this common-adoption quick profile. | This report contains conventional cluster-robust uncertainty, not an HonestDiD sensitivity interval. |
The full event study estimates 23 pre-treatment coefficients. One monthly interval, at event −5, excludes zero at 5%; the declared joint check of five recent leads does not reject. Looking at the complete pattern avoids selecting one monthly test as the whole story.
This is an exploratory quick-profile analysis. The saved report records a qualified result because the staggered-heterogeneity warning was not assessed: it requires a diagnostic-only TWFE result, while this common-adoption run uses TWFE as its primary. Cross-method agreement and formal sensitivity are also absent, as described above. The original report retains the exact statuses and reasons.
A confirmatory planning check was made before fitting and could not supply the required independent comparator and formal sensitivity for this design. The sample was then planned explicitly for exploration. One analysis run was completed; the dataset, target and model were not selected by comparing outcomes.
04 / EXPLAIN THE FINDING
In this simulated rent-control sample, monthly median rent rose less in the treated areas than in the comparison areas after January 2021. DiD-Acto estimates an average relative reduction of $88.53 over 2021–2022 (95% CI −$92.58 to −$84.48). The recent joint pretrend and pre-period placebo checks did not reject at 5%. This is an exploratory result for the benchmark’s common-adoption design.
For a study of an actual policy, the next conversation would focus on why areas adopted, whether housing composition changed, and whether nearby comparison areas were affected. Those facts determine how closely this design fits the question.
Use the shared sample test-bench/syn_real_estate_rent_control_common.csv. The unit is zip_id, time is month, outcome is median_rent, and treatment is treated_now. Register and inspect it, then explain the treated and comparison groups. Show me an exploratory plan with uncertainty and available diagnostic checks before running. After I approve, compare the saved result with the worked sample report.
05 / FOLLOW THE EVIDENCE
The original report files come directly from the saved run. This webpage adds a reading guide and charts; the download preserves the engine’s report and status fields. Report rendering used the stored result with zero estimator calls.
0e9cbbb588e59425130be49b06c39e5345f65498. Dedicated sample workspace; no private analyst data.run_29478d8d9c2b4059b02afe08fa1b7f118b2722ab3b9c19402f2e4c935daf4453c3874dd92429ffa3b572Download the analysis specification · Download file checksums and provenance
The monthly trend chart and before/after table are calculated directly from the unchanged CSV. The headline, confidence interval and diagnostics come from the saved report JSON. The site build verifies source checksums and reproduces the four-mean contrast without refitting.