Table 3. Effect of temperature on the probability of self-reported mental health difficulties: Alternative specifications.

	(1)	(2)	(3)	(4)	(5)
	Day-of	Previous 7 Days	Previous 14 Days	Previous 21 Days	Previous 30 Days
<20°F	-0.0680**	-0.0237***	-0.0155***	-0.0112***	-0.0083***
	(0.0273)	(0.0079)	(0.0056)	(0.0041)	(0.0032)
20–30°F	-0.0101	-0.0026	-0.0038	-0.0014	-0.0002
	(0.0238)	(0.0053)	(0.0040)	(0.0030)	(0.0025)
30–40°F	-0.0097	-0.0066	-0.0075**	-0.0057**	-0.0044**
	(0.0191)	(0.0040)	(0.0030)	(0.0024)	(0.0020)
40–50°F	-0.0099	-0.0063**	-0.0052**	-0.0036**	-0.0025**
	(0.0135)	(0.0031)	(0.0024)	(0.0016)	(0.0013)
50–60°F	-0.0081	-0.0027	-0.0041**	-0.0026**	-0.0019*
	(0.0134)	(0.0024)	(0.0017)	(0.0013)	(0.0011)
70–80°F	0.0132	0.0054*	0.0040**	0.0035**	0.0020*
	(0.0126)	(0.0028)	(0.0020)	(0.0015)	(0.0011)
≥80°F	0.0040	0.0026	0.0048*	0.0044**	0.0032**
	(0.0183)	(0.0036)	(0.0027)	(0.0020)	(0.0015)
N	3,060,158	3,060,158	3,060,158	3,060,158	3,060,158
No. of counties	2,400	2,400	2,400	2,400	2,400

The dependent variable is a binary variable indicating whether the individual has self-reported mental health difficulties for all columns; column (1) looks at the day-of temperature in bins while columns (2)-(5) look at the temperature-day effect from previous 7 to 30 days. Standard errors are clustered at the county level in parentheses. State-month dummies are included to control for local seasonality; county, day-of-year, and year dummies are included to control for unobserved differences across counties, days of the year, and years. All regressions include individual covariates (age and its squared, gender, income, marital status, number of children, education, employment, health plan) as well as weather controls (precipitation, dew point temperature, sunlight). Temperature days in bin 60–70°F are omitted as the reference. Survey weight is applied.

*** p<0.01,

** p<0.05,

* p<0.1.