Interactive coefficient translator and descriptive Texas ZIP-code map
Scenario calculator
Translate a change in planned-event exposure into the coefficient-implied conditional association in transformed restaurant revenue. Both the outcome and applicable event exposure use natural-log-plus-one transformations.
Selected model
Event count: 0 → 1
Association in transformed revenue (R + 1)
+7.40%
Transformed exposure multiplier2.000×
Selected coefficientβ = 0.103
Transformed revenue multiplier1.074×
Coefficient interval translation
Uses the reported 95% coefficient interval.
Change in logged outcome
β × [ln(X₁ + 1) − ln(X₀ + 1)]
Transformation check
Raw doubling does not generally equal a 2× transformed change.
Calculation trace
Interpretation guardrail
Outcomeln(Revenue + 1)
Exposureln(Event + 1)
Claim typeConditional association
The percentage is a coefficient translation for transformed revenue (R + 1), not an exact percentage change in raw restaurant revenue and not a causal forecast.
Texas ZIP-code spatial map
Explore observed restaurant revenue and planned-event exposure across the 1,369 ZIP codes represented in the supplied 2019–2022 panel. Use the continuous year and month bars or play the monthly animation. Values are descriptive and do not estimate ZIP-specific effects.
Legend
Identifier audit: the supplied data contain 41,879 unique SafeGraph place keys, whereas the manuscript reports 41,099 restaurants. The map uses the supplied records as observed; this discrepancy should be reconciled before citing an establishment count outside the dashboard.
Verified coefficient explorer
Browse the estimates supported by the calculator. Confidence limits are copied from Tables 2 and 3; the dashboard does not back-calculate standard errors.
Manuscript estimates
95% confidence intervals based on standard errors clustered at the ZIP-code level.
Group
Estimate
β
95% CI
Source
41,467,966Observed restaurant-days
1,369ZIP codes in supplied data
2019–2022Study period
293,588Unique planned events reported
Method and interpretation
The dashboard translates published coefficients and maps observed summaries without re-estimating the study.
Estimating framework
The manuscript estimates high-dimensional fixed-effects linear models of ln(Revenue + 1) on log-plus-one event exposure and weather controls. The baseline includes restaurant and day fixed effects; standard errors are clustered by ZIP code. Restaurant-week and restaurant-month specifications provide more demanding temporal controls.
Unit of analysis: observed restaurant-day.
Event exposure is defined at the ZIP-code-day level.
Zero observed revenue and zero event counts are retained through log-plus-one transformations.
Results are conditional associations because local time-varying confounding cannot be ruled out.
Exact calculator formulas
ratio = (X₁ + 1) / (X₀ + 1)
Δ ln(R + 1) = β × ln(ratio)
association in R + 1 = 100 × (ratioβ − 1)
Spatial interpretation
The map aggregates observed SafeGraph revenue and PredictHQ exposure from the supplied panel. Revenue is aggregated over restaurant-day observations; event counts and attendance are deduplicated at the ZIP-code-day level before monthly, annual, and full-period aggregation. ZCTA boundaries are 2020 Census geography and approximate postal ZIP areas. Mapped values are descriptive, not geographic causal effects, forecasts, or estimates of attendee spending.