A New Tool in Humanitarian Relief: Texting

Understanding how Ebola is affecting the food supply is far easier when a robot is running the survey.

A mobile phone salesman at a stall in Freetown, Sierra Leone's capital, in 2012  (Simon Akam/Reuters)

Pandemics, like war, have a higher cost than their death toll. On top of the 5,000 lives that Ebola has claimed, there are other sorts of victims in the six West African countries the virus has reached. The emergency erodes trust and infrastructures, threatening local economics and livelihoods.

One infrastructure that’s relatively hard to take down with disease, though, is the cellular phone system. Now, researchers are using it to check on the well-being of people living among the Ebola pandemic.

A UN World Food Programme (WFP) survey earlier this month found that households in Kailahun and Kenema—two districts in eastern Sierra Leone badly affected by the Ebola outbreak—are using “severe food coping strategies.”

“This means people are struggling to meet their basic food needs,” said Jean-Martin Bauer, a food analyst with the WFP. These coping strategies can include skipping meals, reducing portion sizes, and eating less-preferred foods.

The results of the survey are key to understanding who needs support and when, but the methods are important too. The poll was conducted by SMS and “interactive voice” calls—that is, by texting or “robocalling” questions to people who live in the two districts. This automated technique keeps researchers safe, and allows for multiple rounds of surveys to be sent out automatically over time.

“Our typical approach involves sending out roving teams of enumerators with clipboards (or handheld devices) to collect data through face-to-face personal interviews with respondents,” wrote Bauer in an email. “The process delivers valuable detailed information, but tends to be cumbersome.”

Phone-based surveys reduce some of that burden, Bauer said. He added that the WFP can now bring in new data regularly and issue reports on the matter monthly without needing an army of enumerators.

​“Reporting at that frequency is logistically complex to do face-to-face under typical operational contexts, even more so in the context of a public health emergency,” said Bauer.

The poll was made possible by the Denver-based company Mobile Accord. Its main product, Geopoll, conducts opinion and market research surveys exclusively through SMS texting and interactive telephone calls.

Because cellular networks are set up in many developing countries—and landline or broadband networks are not—Geopoll essentially makes widespread polling possible in many places.

(You may recognize Mobile Accord’s name: Earlier this year, it caught flak for building a “Cuban Twitter” clone as a contractor for the U.S. Agency for International Development.)

Mobile Accord, which works with cell providers to obtain customer data, can boast a specialized knowledge about how and where to conduct these types of polls. In 23 developing countries—including Sierra Leone and Liberia—it says it knows how to phrase questions well and when to ask them in the first place.

“Response rates for polls in Uganda are better between 1:00 and 3:00 p.m. in the afternoon, whereas in Kenya it’s 4:00 to 6:00 in the afternoon,” Steve Gutterman, the company’s president, told me when we talked this spring. “That varies by location because there’s religious and cultural variances.”

Geopoll often offers free mobile airtime or other bonuses in exchange for survey responses. In the WFP survey, respondents received airtime credit “as a token of appreciation." This is “in line with best practice for remote mobile phone surveys,” said Bauer.

Surveys like these aren’t perfect. “According to the World Bank, there were 44 mobile cellular subscriptions per 100 people in Sierra Leone in 2013,” says the WFP report. While that’s double the 2009 rate of ownership, most people who own cellphones are richer than average. And since Geopoll can only survey phone owners, its results in turn tend to skew toward respondents with higher incomes.
A previous study, conducted two years ago, had shown that Kailahun and Kenema were more food-secure than the Sierra Leonean average. Taken with this new survey, we can see that residents's food situation has degraded significantly—and that this degradation is so widespread it's reached the wealthier populations that Geopoll can survey.

In other words, many well-off people who were very food-secure before the Ebola outbreak are now skipping meals and dramatically shrinking portions. We’ve been able to gain that information through a creative use of widespread technology. But will we act on it?