Integrating community voices in anticipatory action: a synthesis of complex qualitative data
Anticipatory action means providing the right assistance, in the right place and at the right time. Frameworks and commitments to achieve this, such as the Grand Bargain, accountability to affected people and the Transformative Agenda, all recognize the importance of ensuring community voices are “heard and acted upon”. This recognizes that at-risk and affected communities are the best judges of what they need and when, and that these communities should shape decision-making.
The Start Network is trying to ensure that at-risk communities can plan ahead of a crisis and not have to resort to negative coping mechanisms, such as skipping meals, borrowing money for food, or taking their children out of school. This ‘window of opportunity’ to act can differ from one part of a country to another.
Our recent study shows that in Senegal, communities in different regions enter the lean season – the period between planting and harvesting, when jobs and incomes decline, and food is scarce – at different times. This suggests that some might require assistance to cope with drought earlier than others. Furthermore, different intersecting vulnerabilities can mean that certain groups require different types of assistance at different times, as another recent Start Network study, on gender and disaster risk financing, shows.
How, then, do we ensure that we hit the right window of opportunity and provide the right assistance with anticipatory action? Data, and particularly qualitative data collected over several months throughout the year (longitudinal data), can help us understand not only when that time is, but also what kind of assistance is most needed. However, it can be very time consuming to make sense of so many data points in a way that helps decision-makers.
To address this, the Start Network has developed a methodology for synthesizing and visualizing large quantities of complex longitudinal data, as part of the ARC Replica programme in Senegal.
How we did it
To understand when it is the right time to provide assistance, we collected qualitative data in different parts of Senegal. These were divided according to the main source of livelihood or income. These areas – called ‘sentinel sites’ – included regions where pastoralism, groundnuts and millet were the main sources. In each sentinel site, we interviewed 1-4 people. In total, 22 community members were interviewed remotely each month and asked about six thematic areas: weather and agriculture; water supply; household health; market closures and food at markets; borrowing and buying on credit; and livestock. The result was a rich dataset outlining how households were coping before, during and after the lean season.
Next, we developed a colour-coding system in which each sentinel site was assigned one colour for each month. A site was coded green if no problems were reported for that month, yellow if minor problems were reported, orange for substantial problems (e.g., use of negative coping mechanisms) and red for serious problems (e.g., malnutrition). If one respondent mentioned something negative, the box was coded yellow; if more than one respondent reported something negative, it would be orange; and if all respondents reported problems, it was coded red. Where the responses were conflicting (e.g., one person reporting a substantial problem while another reported minor issues), a judgement was made by the coders on which colour to assign.
Figures 1 and 2 show the coded results. Each box in Figure 1 represents one month; these data were then converted into a wheel depicting the development in each sentinel site (Figure 2).
Figure 1: Colour coding of months
Figure 2: Example of a sentinel site wheel
Figure 1: Colour coding of months
Figure 2: Example of a sentinel site wheel
What did this process tell us?
There are multiple windows of opportunity for anticipatory action. Following communities over several months, we found that some enter the lean season earlier than others. We also found that malnutrition cases were highest in May and August, with more families struggling to provide three meals a day. Around March/April, households were starting to purchase supplementary feed for their livestock. At the same time, once grazing lands were exhausted, prices for livestock feed started increasing, compounding the negative effects on livestock. Each of these represents a window of opportunity for anticipatory action.
The interventions were timely, but the impact was sometimes short-lived. While respondents found the cash and food distributions provided by the government and the Start Network to be timely, they exhausted them very quickly. This was especially true in larger households, with some exhausting the assistance by the start of the lean season, and malnutrition was on the rise again by the peak lean season.
What does this mean for anticipatory action?
For interventions to be effective and sustainable in their impact, anticipatory action practitioners should consider three factors: timing, duration and targeting.
Timing should consider when communities are preparing for the lean season and interventions should happen before people resort to negative coping mechanisms. This may vary from one region to another, so contingency plans might need longer lead times to account for differences between regions. In Senegal, malnutrition cases seem to rise in May and August, which suggests two potential windows of opportunity for food security interventions. The first window is March, which is likely to be particularly effective if the harvest in the previous year was poor; the second could be in June, just before the start of the lean season.
The duration of an intervention is key for sustained impact. This means phasing out support over time, rather than providing one-off support (e.g., a cash transfer). This can help avoid a situation in which households exhaust their assistance too quickly and find themselves in a vulnerable position at the start of the lean season. Support can also be prolonged by providing targeted assistance at different times.
Interventions should provide different types of support, targeted to specific needs of communities. In Senegal, support that is focused on feeding livestock might be appropriate in March/April, when households start purchasing supplementary livestock feed. Support for planting, by contrast, might be most appropriate in April/May, before the start of the rainy season.
This type of complex analysis can help practitioners and decision-makers provide the right interventions and deliver them when they have the highest potential impact. This pilot was a first step in the learning process, however, and the Start Network plans to test it in other countries. For example, we are implementing this approach through ARC Replica interventions in Zimbabwe, where the Start Network and the World Food Programme are collaborating with the government to protect Zimbabweans against drought.
The value of people-centred risk analytics in strengthening disaster risk financing models has been demonstrated, and we hope that this kind of data can also supplement risk models in anticipatory action. The Start Network is also incorporating community voices in Building Blocks, a guidance tool for our members to develop their own disaster risk financing systems. As we test and refine this approach, we want to use larger sample sizes and improve the robustness of the method.
This blog was written by Susan Njambi-Szlapka, research and learning advisor, and Abi Jones, research officer, at the Start Network.