Table 1. Odds Ratios for Donating Across Treatments.

Manipulation		A) GR	B) TPI+	C) TPI–	D) GR × TPI
Invitee (receives a donated invitation)		7.006*			0.327
		(0.030)			(0.262)
Has previously received donated invitations		0.712			1.021
		(0.686)			(0.982)
Seeds
	Observes 0–75		11.414*	(baseline)	(baseline)
			(0.043)
	Observes 76–150		1.341	0.047	0.136
			(0.787)	(0.215)	(0.101)
	Observes 151+		0.219	0.003	0.015*
			(0.280)	(0.198)	(0.022)
Invitees
	Observes 0–75				(baseline)
	Observes 76–150				19.907*
					(0.041)
	Observes 151+				89.948*
					(0.026)
High payment		64.103**	2.532	0.858	3.235
		(0.007)	(0.300)	(0.930)	(0.295)
Time waited (in hours)		0.972*	0.992	1.019	0.976
		(0.023)	(0.577)	(0.619)	(0.075)
Previous participations		0.690	0.784	1.347	0.454
		(0.379)	(0.622)	(0.848)	(0.171)
Baseline odds		4.305	5.323	152.785	268.707***
		(0.181)	(0.100)	(0.130)	(0.000)
Number of observations		516	371	175	554
Number of participants		252	277	133	266
Wald		5 df, 11.93*	6 df, 6.66	5 df, 2.49	8 df, 11.98
		(0.036)	(0.354)	(0.778)	(0.214)

Two-sided tests: *p<0.05, **p<0.01, ***p<0.001.

The table reports odds ratios and p values (in brackets) from random-intercept logistic regression models for A) seeds and invitees in the no-observation treatment by number of donated invitations received; B) seeds in the observation and no-observation treatments by number of donated invitations observed; C) seeds in the observation treatment by number of donated invitations observed; and D) seeds and invitees in the observation treatment by number of donated invitations observed by invitees compared to seeds. Results show that receiving and observing donations initially increases the willingness to help others, and that invitees are less susceptible to a subsequent decline in helping.