In July, Yemen was declared a Level 3 Emergency – the highest priority level in the global humanitarian system. At WFP, we knew we needed real-time information to track needs on the ground. Mobile surveys had worked well for us in emergencies in Iraq and the Ebola epidemic. So WFP managers in Yemen decided to launch a phone survey-based food security monitoring system.
But, would mobile phone surveys work in Yemen?
Concerns were immediately voiced: Would people readily take survey calls in the midst of a conflict? Did people have access to electricity to recharge their phones? Would the phone network function well enough let us call all parts of the country? Would people trust us?
Guess what- people not only responded to our calls but many responded consistently! We started calling people at the end of July, reaching about 2,400 people with each survey round. Many respondents- 60%- have stuck with us month to month. This means we can offer trend analysis on food security indicators- check it out in our monthly food security updates.
So, what have we learned from using mobile surveys in Yemen?
- Working with a professional call center has been a great asset.Due to the conflict, we’re placing calls from a call center outside the country. Our operators had to go through a ‘learning’ phase where they picked up Yemeni dialect — food items seem to have quirky names and vary somewhat from Arabic spoken in other countries.
- Random digit dialing works, just be patient! We used random digit dialing because we did not have access to a phone number database for Yemen. So we created our own! A computer dialed up random numbers, using the mobile network operators’ prefixes and generating the final five to seven digits. If a phone number worked and the call went through, it was immediately routed to an operator. As you can imagine, this took some time, but after a few weeks, we figured out what bands of numbers are active in Yemen and ramped up the call volume. We always quickly get through to people in the capital, Sana’a, but it takes longer to reach enough people in less populated governorates. Patience has been a virtue. But we’re pleased to be at the stage where we can complete a round of data collection in 2 weeks in such a complex environment.
- A qualitative, open-ended question provides rich information. In addition to asking people a quantitative food consumption score, we have an open-ended question where we ask people to describe the food security situation in their community. We analyze these responses using tools such as pattern sentiment analysis and word clouds. In the November Yemen bulletin, our pattern sentiment analysis showed a decline that we also saw in respondents’ reports of food consumption. This correlation not only shows that we are getting reliable data through open-ended questions but also that potentially, we could use people’s perceptions of food security to measure an actual deterioration in food security.
- Time of day for calling is important. In Yemen we learned, call at night. We had a hard time reaching respondents when we called during daytime hours. It turned out that people were keeping their phones switched off during the day to preserve the battery in order to make calls in the evenings. When the call center noticed that 90% of the calls were going through during evenings, they allocated more operators to evening shifts. This helped us reach our targeted number of respondents.
Monitoring IDPs, Cyclones, and Operations
Yemen is another case where mobile phone interviews have enabled mVAM to reach households located in the most conflict-exposed areas, where we could not do a survey face-to-face. We are even able to reach internally displaced persons (IDPs)- currently over 1/3 of our sample. Interviewing IDPs is key in a conflict that has caused widespread displacement.
The system has also been flexible: in November, when the southern coast of Yemen was hit by two successive tropical cyclones, we were able to immediately place calls to affected areas of Al Maharah and Abyan. Mobile surveys have also allowed us to monitor our food assistance programmes. We are also currently calling 1,200 beneficiaries every month to ask about the food they received and track their food consumption. We’ve found our results to be in line with previous face-to-face surveys, confirming the reliability of remote mobile data collection.