Modern technologies have opened up
possibilities to investigate the consequences of humanitarian crises faster. 510
Global, a data innovation team set up by the Netherlands Red Cross, has been looking
into ways to use big data for humanitarian goals. They have developed a data-driven
model to identify priority areas for humanitarian intervention related to
natural disasters faster. The effectiveness of their work showed from the actions
taken during typhoon Haima. This category 5 tropical superstorm hit the
Philippines in October 2016, ravaging big parts of the country. 14 people died
during the storm and estimated damage to property was $76.9 million (Jalad, 2016). However, things could have been a lot worse.
The typhoon had been anticipated for a while and as soon as it hit the
Philippines, humanitarian aid organizations received detailed information from 510
Global. Within hours they managed to map the areas hit by the typhoon, showing
clearly the damage done and thereby helping the Philippine Red Cross to target
priority zones (510 Global, 2016).
Typhoons Haima (2016) and Nina (2016) in the
Philippines (“Graduation Portal,” 2017) functioned as a test environment for 510
Global’s data model, and outcomes showed that the data tool can possibly be
used for different natural disasters in different countries. Applying such a
data model in disaster stricken areas would save time and money in the
provision of aid. However, for the model to be applied, additional research is
needed into the decision making process of humanitarian organizations. This is
where this study would come in.
Many studies into the coordination structure
and decision making process of corporate organizations have been done, but
these did often not include HO’s or NGO’s. A knowledge gap can be identified
here when it comes to the way HO’s do their planning, decision-making and
coordination processes. In this thesis, research will be done into these
processes and historical cases in which data-driven decision making has been
used will be analyzed. The main problem to focus on will be integrating a
big-data model like that of 510 Global into a humanitarian organization.
It may be obvious that research into this topic
carries high societal relevance, along with academic insights. Quoting 510
Global itself: “Smart use of big data will positively impact faster & more
(cost) effective humanitarian aid” (510 Global, 2017). Finding a solution to the integration of a
data model into HO’s will not only create a grand corporate resource, it will also
have direct impact on real life humanitarian crises, with the potential of saving
and improving the lives of many humans.
The next chapter will present a literature
review on the state of the art literature, followed by an introduction of the
(sub) research question(s). Research methods and tools will be discussed
afterwards and the proposal concludes with an outline of the expected time
Literature research: four challenges for big data as a tool for
Humanitarian big data strategies like the one
introduced by 510 Global have a lot of potential, but are not without
challenges. Several obstacles obstruct successful application of big data by
humanitarian organizations. In the following literature research existing data is
used to find out about these obstacles and problems. This results in the
identification of a knowledge gap in this area, from which the main research
question for this thesis is derived.
Haak (2017) identifies three main challenges for the use of big data as a tool for
humanitarian aid. Firstly, the collection
of reliable and representative data. Secondly, the correct analysis and processing of the data. And thirdly, the successful integration of the data in
organizational practices. While the first two challenges mainly concern the
domain of ICT, about which has been written a lot already (Raymond & Card, 2015), the third challenge has so far been
relatively untouched. For the big data models to be of any use to the final
humanitarian decision makers, it is essential that those models get integrated
into their organization structure (Vitoriano, Montero, & Ruan, 2013). A review of the available literature provided
four reasons why this integration has not been successful so far. These reasons
will be discussed in the next paragraphs.
collaboration between parties involved
In the aftermath of large scale disasters many
different actors are involved in providing aid to the stricken area and its
inhabitants. Both local actors (municipalities), national actors (governments,
army) and international actors (NGO’s, United Nations) try to plan, manage and
execute aid under conditions of high uncertainty and pressure (Ortuño et al., 2013; UN/ISDR & UN/OCHA, 2008; Van de
Walle & Comes, 2014). Historic cases showed that collaboration between these actors during a
disaster does not always go smoothly (Whipkey & Verity, 2015). Different actors bring their own information
systems, logistics, management and decision processes, which prevents the
parties from adapting a collaborative approach (Van de Walle & Comes, 2016). Cultural barriers between the local
organizations and foreign aid providers also impedes progress (Twigg, 2015). Due to these collaboration issues the
provided big data cannot be used to its full potential.
The gathering of big data is a complex process
that produces a highly technical model which might be incomprehensible for
non-technicians. It is therefore crucial that the produced information gets
delivered to the humanitarian decision maker’s in the right format (Van de Walle & Comes, 2016). Unfortunately, this is not always the case.
Communication of data often happens top down, on a technocratic and centralized
level, without taking in consideration the decision maker’s needs (Cinnamon, Jones, & Adger, 2016; Van de Walle & Comes, 2014). Also, when multiple data providers are active
simultaneously, they might send an overkill of data. A continues stream of
surveys, questionnaires, reports and maps can create an “Information disaster” (Van de Walle & Comes, 2014), wasting resources on information that cannot
be fully processed (Read et al., 2016). Therefore, incompatibility of the data is the
second cause that prevents humanitarian groups from integrating the data in its
3. Lack of a general policy framework
A third reason why data has not been
successfully integrated in practice yet is the lack of a general framework that
states how big data should inform decision making and how and when big data
should be used (Whipkey & Verity, 2015). The humanitarian organizations need a
foundation that prescribes the policies, protocols and practices considering
the use of big data (Chan, Bateman, & Olafsson, 2016). Since humanitarian aid groups have a high
fluctuation in staff members from crisis to crisis, practices cannot just be
remembered by the staff members, but should be written down in common toolkit
that can be used by all organizations (Raymond & Al Achkar, 2016). Only with such a framework can the
information shared by the data producers be used and integrated by the
The final factor that should be taken into
consideration when regarding the unsuccessfulness of integrating big data, is
the data responsibility of the humanitarian groups. According to Al Achkar (2016) data responsibility encompass the following: “the ability of organizations to be ready to responsibly and effectively
deploy and manage data collection and analysis tools, techniques and strategies
in a specific operational context before a disaster strikes”. In other
words, the competencies and ethical standards to safely handle data, in order
to achieve their mission: providing humanitarian aid (Chaumba & Van Geene, 2003). Many humanitarian organizations do not
possess such data responsibility yet and will need training before they can
integrate big data.
and research question: communication
of disaster response data
By reviewing the available literature on this
topic, it has become clear that the integration
of big data into common practices is at the moment the greatest challenge
for humanitarian aid providers. Four reasons have been found that address this
challenge and explain why integration has so far been unsuccessful. A solution
to this problem, however, has not been identified by any research yet. As Twigg
(2015) describes it: “There is
relatively little guidance on integration of data; much of what is available is
limited to general principles, and examples of good or bad practice are rarely
documented or shared.” As a result, a knowledge gap for this topic can be
identified, namely, responding to each of those four reasons and finding solutions
for them. This opens up possibilities for more research. From data provider 510
Global’s point of view, the second reason, concerning data incompatibility, is
most interesting. Therefore this thesis will focus on this issue. Concluding, the
following research question for the thesis proposal is formualted:
“How should the communication of disaster
response data be changed so that it is compatible with the decision making
structure of humanitarian organizations?”
approach and methods: case study
To answer the main research question, it is
essential to obtain a detailed overview of the decision making processes inside
humanitarian organizations. To create a valid and reliable model of the
decision making structure, one can look into the practices of real life cases.
Since the theoretical concept of humanitarian big data implementation does not
work in the real world yet, a case study
approach will be used to analyze real HO’s (Yin, 2009). By studying several cases, common practices
and requirements to big data output can be identified. In addition, a design oriented approach will be adapted
in a later stage of the thesis to create a design for the decision making model
of HO’s (Hevner et al., 2004). Combining the outcome of these two approaches
will fill the void in the functioning of the socio-technical system of
humanitarian big data.
By looking at the main research question on a
lower level of abstraction, seven sub-research questions (SRQ) can be formulated.
These questions will be answered over the course of seven phases. In this
chapter each of the phases will be discussed in order, describing the required
data, research method, tools and deliverables in every phase. The reader is
recommended to follow the research flow diagram while reading, which can be
found at the end of this chapter.
Phase 1: Preparation
The first phase encompasses the preparation of
the thesis, in which the writer gets familiarized with the topic and its
challenges. By performing a literature review, a knowledge gap is identified.
The research problem is defined, as well as the main research question and its
sub- research questions. The deliverable of this phase is the thesis proposal,
which marks the start of the thesis project.
Phase 2: Desk research
The research starts by gathering more information
on the topic, specifically in two areas. Firstly, data about historic cases of
humanitarian organizations that used big data is needed, to answer the first
In which historic cases has big data been used by humanitarian organizations?
By answering this question, historic cases are
identified that can be used for the case study in the next phase. It is useful
to make a distinction between cases that successfully used big data, and cases
that were unsuccessful. These cases serve as deliverable and starting point for
SRQ3 and SRQ4.
Secondly, information on the internal structure
of humanitarian organizations needs to be found, so that the second
sub-research question can be answered.
How are humanitarian organizations internally organized?
Knowing how humanitarian organizations
internally communicate and delegate work is valuable information and a
requirement to design a framework on their decision making, which will be done
in phase 4. One cannot expect that all HO’s are organized in the same way, it
is therefore important to identify resemblances and differences. The HO’s can
then be grouped based on their internal structure. The needed data can be
acquired relatively easy, through common desk research. By researching
literature both sub-research questions can be answered. Literature repositories
will serve as the main tool.
Phase 3: Case study
In the next phase, the first actual research
approach is executed: the case study. Two types of methods will be used: real
life simulations, and interviews. The simulations will happen in a
semi-controlled environment, in which past cases of disaster management are
re-enacted. Some kind of group decision room will be needed as a tool. By
observing the behavior of the team members of the humanitarian organization,
their working practices can be identified. The practices of different case
studies can be compared to find commonalities and define general practices. In
unsuccessful cases of big data integration one can observe why the big data was
not used successfully. These observations will answer the third sub-research
How did humanitarian organizations use big data in their decision making?
Interviews will give further insight into the
practices of HO’s, and it will also provide qualitative data on the wishes and
preferences of HO’s on the provision of big data. This is needed for the next
What requirements do humanitarian organizations pose to big data?
Interviews require a pre-defined list of
questions and a place to meet with the interviewee. The deliverable of this
question, a set of requirements on big data, will be used in phase 5, when
defining the recommendations and changes to data output.
Phase 4: Design
The fourth phase is the design phase, in which
a framework on the decision making structure of humanitarian organizations is
developed. Both information on the organization structure and decision making
practices are needed. This is provided by the desk research done for SRQ 2 and
the data from the case study in SRQ 3. Both deliverables function as input for
the fifth sub-research question.
What does the decision making structure of humanitarian organizations look
Naturally, a design method is used here to
design the framework. Simple modelling software can be used to create an
overview of the steps and processes within the humanitarian organization that
lead to their decision making. The deliverable of this phase will be a
framework that shows the functions and place of disaster response data in the
decision making structure. Since it will be unlikely that every HO adapts the
same decision making structure, designing a single framework would only address
a few HO’s. The outcome of this SRQ will therefore probably be in the form of
several frameworks, that share resemblances with each other, but are different
based on the found variances from SRQ 2 and SRQ 3.
Phase 5: Data analysis
In the fifth phase all gathered data comes
together to finalize the research project. The framework of the decision making
structure and the set of requirement on big data are qualitatively analyzed and
combined to answer the sixth sub-research question.
SRQ 6. How can the data provider improve its
data output and communication methods?
The answer of this question will be in the form
of a report, giving recommendations to the data provider (510 Global) how to
improve their data output and communication methods. The report will function
as a guideline the data provider can follow to decide on their behaviour and
methods. Based on with which HO the data provider is dealing, the report will
suggest different approaches for data output and communication.
Phase 6: Validation
Before a final
conclusion can be drawn, the designed framework will need to be validated. The
framework should be applicable to a wide variety of different humanitarian
organizations. After completing the fifth phase, 510 Global is supposed to
possess the means to work with any kind of decision making structure. But to be
sure that this hypothesis holds, validation is necessary. In other words:
SRQ 7. How can the solution be generally applied to all humanitarian organizations?
The suggested method
to answer this final sub-research question is to again make use of the available
case studies. This time the goal is not merely to observe the HO’s but to
actively participate and take on the role of data provider. The guideline
provided in the report, which was the result of SRQ 6, now needs to be followed
in order to determine the communication methods and data output for each case.
If the designed framework and guideline are valid, the outcome of each case
should be the successful integration of big data by the HO. This should also be
so for the cases that were unsuccessful in the past. If the framework and
guideline turn out to be invalid, one can define new practices and requirements
that were overlooked in phase 3. As a result, this validation process functions
as an iteration back to phase 3. A drawback of this method that the scope of
this thesis will only include time for one iteration. So in case the framework
is still not valid after the second try, additional research is required to
finalize the advice to 510 Global.
Phase 7: Finalizing
In the conclusive phase, the aggregated answers
of all sub-research questions will provide an answer to the main research
should the communication of disaster response data be changed so that it is
compatible with the decision making structure of humanitarian organizations?”
Recommendations to both the data provider and
the HO’s that will use the big data will be made, based on the validated
framework and guideline from SRQ 7. With this action the goal of the thesis has
been reached; providing a tool to smoothen the communication of data and help
integrate the data into the decision making structure of HO’s.
On the next page you
can find the research flow diagram, which shows the seven successive phases
with their sub-research questions (blue), deliverables (green) and research
methods (orange). Each answered sub-research question will provide a
deliverable that will be the input of the next SRQ in a future phase. Each
phase will be covered by a separate chapter in the thesis report. A global time planning of these phases and
the execution of the thesis project can be found in the final chapter of this