Table 5.

Results of the regression model estimating the number of participants recruited per campaign event (number of observations=199)^a.

Predictors (lag in days)	Estimated number of recruited participants per campaign event	P value	95% CI
Mailed letters (15 days)	106.5b	.002	39.6 to 173.4
Mailed reminders (5 days)	16.9	.28	−13.5 to 47.2
Log paid Facebook reach per 1000 (2 days)	1.4b	<.001	0.6 to 2.1
Facebook unpaid posts (>2 days)	3.2b	.002	1.5 to 5.0
Wide-reach news media coverage (>2 days)	163.6b	<.001	149.0 to 178.2
Other recruitment means (>4 days)	−1.3	.05	−2.7 to 0.0
Partner communications (>2 days)	3.5	.16	−1.4 to 8.4
Snowball recruitment (>4 days)	2.5	.85	−24.4 to 29.3
Smaller-reach news media coverage (>2 days)	10.4b	.001	4.1 to 16.8
Intercept	4.3b	<.001	2.9 to 5.7

^aR²/R² adjusted=0.79/0.78.

^bP<.05.