Table 3.

Treatment Effect on COVID-19 Cases at Zip Code Level^*

				Mean (CI 95%)		OLS model			Num
Campaign	Outcome	Period	County treatment	Treatment	Control	Treatment (CI 95%)	p-value	RI p-value
Both campaigns	Asinh(Fortnightly Cases)	Dec/Jan 1–14	All	4.350 (4.302,4.398)	4.370 (4.323,4.417)	−0.035 (−0.062,−0.007)	0.013	0.009
			Low Intensity	4.359 (4.273,4.445)	4.358 (4.305,4.411)	−0.032 (−0.067,0.004)	0.080	0.097
			High Intensity	4.347 (4.295,4.399)	4.407 (4.325,4.489)	−0.039 (−0.075,−0.003)	0.033	0.038
Thanksgiving	Asinh(Fortnightly Cases)	Dec 1–14	All	4.333 (4.278,4.388)	4.298 (4.243,4.353)	−0.027 (−0.059,0.005)	0.097	0.108
			Low Intensity	4.284 (4.170,4.399)	4.256 (4.192,4.320)	−0.015 (−0.063,0.033)	0.535	0.498
			High Intensity	4.348 (4.285,4.411)	4.418 (4.313,4.523)	−0.039 (−0.082,0.004)	0.078	0.096
Christmas	Asinh(Fortnightly Cases)	Jan 1–14	All	4.368 (4.310,4.425)	4.442 (4.385,4.499)	−0.042 (−0.073,−0.012)	0.007	0.010
			Low Intensity	4.429 (4.312,4.547)	4.456 (4.391,4.522)	−0.048 (−0.091,−0.006)	0.025	0.043
			High Intensity	4.346 (4.280,4.412)	4.396 (4.281,4.510)	−0.036 (−0.080,0.008)	0.108	0.111

^*This table provides the control and treatment means at the zip code level, in addition to the estimate of the treatment coefficient in equation (2). The outcome is the inverse hyperbolic sine of the fortnightly cases, during a period which starts five to seven days after the event (Thanksgiving or Christmas). 95% CI are reported in parentheses. Standard errors are clustered at the zip level.