Case Study

Photos and Donation


The poor in Kenya cannot afford health insurance, and so must pay for their medical expenses out of pocket. Unfortunately, the savings rate for health expenses is very low. This multiplies the impact of health-related emergencies, as people often do not have enough money to cover the cost of medical bills even when they have saved. Given that a nationally subsidized health insurance costs about $60 per year, there is an opportunity to create a peer-to-peer charitable giving platform to promote co-funding of health insurance for a Kenyan family. The Global Health and Development team is helping to optimize the design of such a platform.


As part of the design process, we tested the impact of differing facial expressions on donation. Based on the literature, one would predict that donors would be more willing to give to a recipient with a sad expression compared to a recipient with no photo or a neutral or happy expression. However, we were hoping that the additional impact of the photo would be small if the interface design without photo (in the control group) were sufficiently compelling, because it is costly for the platform to collect and display photos of every recipient.

We tested a prototype of the Mbrella mobile site on Amazon Mechanical Turk. On the donation page, participants saw no photo, or a recipient with a sad, neutral, or happy expression.

We measured frequency of donation, as well as the average donation amount in each condition.


It turns out that there are no significant differences in donation amount or donation propensity whether participants saw a photo or not. However, neutral and positive photos display a pattern indicating that they could potentially decrease donation amounts, and a sad expression performs slightly better than a neutral photo. While the results are counterintuitive, they are in the direction we hoped for.

While the results might suggest that a donation solicitation is no more effective with a photo than without one, it is important to note that participants in all conditions saw a lengthy and emotional description about the recipient (see below). It is possible that this description created empathetic feelings on its own, leaving little room for the photographs to have an impact on donations. This will be the subject of future tests.

Why it matters

Including a photo with a recipient expression on the Mbrella site seemed to have no effect on donations in this particular test. This is potentially because the interface design of the control group is so compelling that it already creates a strong connection between donor and recipient, and hence the photo creates no additional impact. However, if we were to include a photo, a sad photo would likely elicit the highest average donations on our site.