Going Beyond What Works

Is the humanitarian aid system doing the right things? And, is it doing the right things right? Can data help increase the effectiveness of humanitarian aid? On the face of it, these questions seem redundant. One would assume that improved data would surely result in more effective humanitarian aid delivery. The answer, however, is “maybe” at best: more data does not always translate into better humanitarian action or quality of care.

In 2014, the budget for humanitarian aid amounted to $24.5 billion. The Global Humanitarian Assistance Report (2015) noted that there was still a shortfall of 38 per cent in terms of unmet need. With the current refugee crises, resources from traditional official development assistance (ODA) budgets have been shifted to humanitarian aid. Nevertheless, there is a need for more funding and the shortfall is unlikely to be met in the next decade. It is, therefore, not surprising that aid and donor agencies are concerned about the effectiveness and impact of their assistance. Currently, 93 per cent of people living in extreme poverty live in countries that are affected by humanitarian crises. Clearly, each dollar of aid needs to be used to help alleviate their suffering.

As the world prepares to meet at the World Humanitarian Summit in Istanbul in May 2016, it is an opportune time to consider if humanitarian aid is being distributed effectively. Indeed, the humanitarian sector already employs a variety of ways to understand whether this is being achieved. In this respect, the humanitarian system has come a long way from an earlier era, in which the intention and action of providing assistance were deemed to be sufficient. Most humanitarian aid programmes today, to the extent possible, conduct rapid assessments, collect routine programme data, carry out perception surveys, and perform real-time monitoring and evaluation, all of which contribute critically to understanding and implementing assistance activities.

The current data revolution is allowing data to be collected more easily and enhancing the ability to illustrate, visualize and analyse data. Data collected in faster, better and smarter ways can help inform many aspects of humanitarian aid. The table below summarizes some ways in which different kinds of data can be used to better inform the humanitarian system. It is clear that better data is required to understand if humanitarian aid is principled, timely, appropriate and effective.

Sources of data applicable to humanitarian aid, with advantages and disadvantages (Puri and Dhody, 2016).



Helps to inform

Other notes


Satellite images

Changes in the physical condition of a geographic area, a visual estimate and illustration of the extent of destruction, density of construction, land use.

Landsat high-resolution images require interpretation and ground truthing.


Aerial photographs, other remote sensing images

Surveillance, cartography and drawing maps, displacement-related movements, altitude, elevation, topography.

Can be combined easily with GPS data.

Activity is unobtrusive and, depending on resolution, can be used easily in geographic information systems (GIS). Remote sensing images help to collect data on inaccessible or dangerous objects. May replace expensive, on-the-ground data and ensure that areas are not disturbed.


Spatially disaggregated and explicit data (maps)

Assessments of soil; roads; access; use; boundaries of properties, including villages, states and other administrative units; altitude; population density; climate maps; ethnicity; and migration patterns. Maps showing indices such as disaggregated poverty have also become common.

Include topographic and soil maps. Usually require other methods for construction such as remote sensing but also census-type data collection (e.g. for property rights and boundaries).


Surveys (social, thematic, perception surveys at the household or individual level)

Livelihoods; behavioral patterns; eligibility for programmes; perceptions; socioeconomic indicators; physical/economic/social access; use; income, assets and impacts on welfare, including changes in these levels depending on whether data is time series, panel data, cross-section data or repeated cross-section data.

Requires careful qualitative work beforehand and afterwards to interpret; training for data collectors and careful piloting of instruments to ensure questions are conveying what is being asked; and data cleaning. M-health and e-health initiatives are especially well suited to surveys and are taking on the challenge of faster, better and smarter data. In most cases, this requires careful data entry and algorithmic checks, which can, if mobile phones or PDAs are used, be done on the survey itself. Cross-sectional and repeated time series or panel data have attributes that are specifically useful for understanding changes over time.

It is important to understand attrition in data and reasons for it. It is also important to construct these surveys carefully to understand incidence and direction of bias.


Administrative and programme data collected by humanitarian agencies, government departments or organizations; at the level of the administrative unit

Eligibility; average socioeconomic attributes; administrative, operational data management and delivering processes; laws and changes in legislation.

Typically broad ranges of data that can be used for subpopulations. Illustrates main changes or attributes. Can help target subsections of the population.


Individual, structured or semi-structured interviews

Perceptions, beliefs, customs, reasons for behavior change, some determinants of actions, social status, processes and exchange. Also helps to explain unobservable selection bias as well as participation and non-participation.

Usually required before, during and after most quantitative data collection so quantitative data can be anchored, collected well and interpreted clearly. Also required for questionnaire design and sampling.


Case studies

Qualitative understanding of local drivers and dynamics including processes.

Case studies are especially important to understanding processes and behaviors and provide deeper insights into what quantitative data may be telling us.


GPS data

Can help determine locations for cities, markets, hospitals, and schools as well as boundaries for properties and areas.

Can be combined into GIS with other data and helps to combine aerial data with on-the-ground data. GPS data can provide spatial coordinates that may be combined with mosaics of satellite images to then make these useful with other types of data. 


Management information systems

Delivery capture, implementation fidelity, extent to which project targets have been met.

These are usually used with process data to understand implementation fidelity.


Despite advances in data collection, we need more than simple data to inform action. Three kinds of data, used for specific purposes, are described below.

Data for rapid action. What are the needs of the crisis-affected population? Who are the most vulnerable? What capacities exist locally? Where should assistance be directed in a humanitarian crisis? What can help inform the scale and scope of the response? During the outbreaks of Ebola in West Africa and avian flu in Asia, there was a strong push for transforming the speed and quality of data collection to improve the response. This involved baseline post-emergency data, as well as vulnerability, capacity and resource requirements. In natural disasters and acute conflicts, the free flow of data is essential for better coordination between agencies and a more effective response. Different agencies have different standard operating procedures, and it is clear that the quality of data collected by most organizations varies widely.

Data for measuring impact. Indeed, we need to start measuring and understanding if and how much humanitarian aid is making a difference. For example, the World Food Programme is asking important questions about the extent to which prevention and treatment of moderate and acute malnutrition is contributing to measurable improvements in nutrition levels and overall health. Such evaluations that relate the changes that are being caused by the programme (also called impact evaluations) are increasingly being conducted. Nevertheless, a wide evidence gap remains in this area: an overall assessment completed in 2014 by the International Initiative for Impact Evaluation (3ie) found that there have been fewer than 50 impact evaluations of humanitarian aid programmes, although more than $100 billion has been invested in this area in just over a 10-year period (Puri and others, 2014).

Data for improving delivery. To ensure that assistance reaches the most vulnerable people, it is important that we understand field realities and “last mile” challenges. Developing effective models of delivery requires periodic analysis of data and feedback from beneficiaries. For instance, in resource-poor settings, community health workers are often relied on to deliver much of the health care at the primary level. Capacity constraints often impede the ability of underpaid and undertrained community health workers to provide a full package of services in an effective manner. Task-shifting and incorporating patient experts in the delivery of care are some innovative ways in which adherence to HIV treatment has been achieved in South Africa and elsewhere.

Consultations preceding the World Humanitarian Summit have only peripherally touched on the subject of evidence. The synthesis report of the consultation process speaks of an important aspect of this evidence—the need for better data—but many questions remain unanswered. How should programmes be delivered? What incentives will enable community health workers to deliver better care? How can humanitarian agencies and staff be encouraged to adopt and work with innovations that may have been proven effective? Do cash incentives always work or can social recognition play an alternatively effective role? To what extent should community participation be encouraged in designing and targeting programmes? What amount of training and incentives is optimal to increase effectiveness on the ground? Are there other ways in which we can ensure that the programmes are delivered to t he last mile, given the security concerns of humanitarian workers? How can humanity, impartiality and effectiveness be ensured?

Clearly, increasing data collection alone will not be sufficient. The humanitarian community needs to think efficiently and effectively about data to inform both doing the right thing and doing it right. This will require clear thinking about the type and purposes of data collected rather than just collecting data for data’s sake.


Global Humanitarian Assistance Programme (2015). Global Humanitarian Assistance Report 2015. Bristol, UK, Development Initiatives. Available from 

Puri, Jyotsna and Bharat Dhody (2016). Missing the Forests for the trees? Assessing the Use of Impact Evaluations in Forestry Programmes. In Sustainable Development and Disaster Risk Reduction, Juha I. Uitto and Rajib Shaw, eds. Tokyo, New York, Springer.

Puri, Jyotsna and others (2014). What methods may be used in impact evaluations of humanitarian assistance? 3ie Working Paper, no. 22. New Delhi, International Initiative for Impact Evaluation (3ie). Available from http://www.3ieimpact.org/media/filer_public/2014/12/08/wp_22_humanitarian_methods_working_paper-top.pdf.

World Humanitarian Summit secretariat (2015), Restoring Humanity: Synthesis of the Consultation Process for the World Humanitarian Summit. New York, United Nations. Available from https://www.worldhumanitariansummit.org/key-documents#major-reports-linking.