Editor’s note: The World Food Programme (WFP) regularly conducts surveys throughout the world to assess the food security of different areas. WFP uses two measures to access food security: the Food Consumption Score (FCS) and Reduced Coping Strategies Index (rCSI). Traditionally surveys have been conducted through face-to-face interviews.
Quirk’s recently had the chance to interview Matthieu Sauvage-Mar, the director of client services at GeoPoll, to learn more about how their team worked with the WFP to adapt the FCS and rCSI to work on mobile phones, using SMS surveys. This is an edited version of the interview.
Q: How has the use of mobile research in emerging markets changed in the past few years?
A: The world of mobile research is still very new in emerging markets. While developed nations have moved quickly past basic mobile phones to embrace smartphones reliant on Internet, emerging markets have not. However, the sheer number of mobile phones available in developing nations is awe-inspiring. In Sub-Saharan Africa, for example, mobile subscribers have increased 18 percent per year for each of the past five years.
This widespread connectivity makes research done directly via mobile increasingly representative of the general population, which is crucial to gaining an accurate understanding of any population.
Why did the WFP look to mobile surveys when searching for data on food security in the Democratic Republic of Congo?
TheWFP was unable to send in-person survey enumerators to the North Kivu region due to the conflict that was taking place there. With the growing prevalence of mobile phones in the area, we thought there was the option of testing mobile surveys and seeing if results from mobile surveys were comparable to those found in face-to-face surveys. WFP found that not only were mobile surveys able to identify areas of food insecurity but also that the time and cost of mobile surveys were significantly lower than face-to-face surveys.
How was the mobile sampling process different from past WFP studies?
For face-to-face surveys, WFP used two-stage cluster sampling to obtain a sample that was representative of a wide portion of the population and enumerators went door-to-door to collect responses. In the Democratic Republic of Congo, GeoPoll has a database of over 7.9 million mobile phone subscribers, with 200,000 in North Kivu. The GeoPoll sample was obtained by a simple random sample of the North Kivu database. After respondents were randomly selected, they were sent an opt-in message which informed respondents that surveys would be free to respond to, lowering the economic barrier to participation.
What were some survey highlights?
For the rCSI, which consists of five questions designed to flag “extreme” behaviors such as reducing adult food consumption, we found that there was no significant difference in results from face-to-face and mobile surveys. In addition, mobile surveys did show certain expected results, such as a significant difference in scores for displaced and non-displaced populations. This was an exciting result for us, as it showed that mobile surveys were able to quickly and accurately identify areas in need.
Furthermore, the speed at which we were able to gather survey results demonstrated that mobile surveys are a valuable tool in evaluating crises.
What are the top challenges when conducting research in an emerging market?
There are many unknowns when working in emerging markets and each one presents its own unique set of challenges. Which days of the week are best for polling? Which times of day? How much do phone penetration and literacy matter? How about multiple languages? How much does offering an incentive help? Who is best partner as we scale?
All these questions need to be answered to build a successful infrastructure for collecting accurate, reliable data. One of the biggest challenges is making that first step and establishing relationships with the local telecom providers who can connect us to consumers. We also work carefully with the data we do collect, weighting and extrapolating to ensure that our sample is representative of the population we are trying to measure.
What’s next for mobile data collection?
With the spread of mobile phones throughout the developing world, we see a future where we can reach any population on the planet and learn about their wants, needs and concerns. This won’t be easy and a big priority in mobile data research is thinking about how best to use all of the methods available to mobile, whether it is through text messaging, voice messaging or the Internet.