This experimental study tests the role of conditional cooperation for user's donations to a large online platform. Specifically, we vary the text used in solicitation banners in two randomized A/B tests. The treatment varies the communicated number of other users who donated in the past. In a second test we attempt to shed light on the persistence of conditional cooperation when relating the number of donations to the number of users of the online platform.
External Link(s)
Citation
Linek, Maximilian and Christian Traxler. 2018. "No Conditional Cooperation? Trials in a large online Fundraising Campaign." AEA RCT Registry. November 13. https://doi.org/10.1257/rct.3543-1.0.
This experimental study tests the role of conditional cooperation for user's donations to a large online platform.
Intervention Start Date
2018-11-09
Intervention End Date
2018-11-11
Primary Outcomes (end points)
(1) number of donations, (2) number of donation page views (clicks on the solicitation link)
Primary Outcomes (explanation)
Secondary Outcomes (end points)
(3) the number of x-clicks
Secondary Outcomes (explanation)
the number a user closes the donation solicitation in her browser
Experimental Design
Two A/B tests will be conducted, both of which will vary the text content of the solicitation to donate. First, we vary the communicated number of other users who donated in the past. Second, we attempt to shed light on the persistence of conditional cooperation when relating the number of donations to the number of users of the online platform.
Experimental Design Details
Randomization Method
Randomization done by cooperation partner
Randomization Unit
Browser sessions of users of the online platform
Was the treatment clustered?
No
Sample size: planned number of clusters
no cluster
Sample size: planned number of observations
1,200,000 browser sessions
Sample size (or number of clusters) by treatment arms
assignment to either treatment or control group is conducted randomly with equal probability (50%)
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)