Data collection and analysis tools for food security and nutrition

participate in, contest or otherwise challenge 
their decision-making (in terms of inputs, logic 
or outcomes). Consequently, there is likely to 
be power asymmetry, for instance, between 
the AI system developers, service providers or 
third parties, and those who interact with the AI 
systems. Moreover, AI systems may fail to give a 
comprehensible explanation of their underlying 
decision-making process to the affected 
individuals. This opacity and power asymmetry 
not only expand opportunities for potential 
exploitation, but may erode the sociotechnical 
foundations of justice, morality and human 
rights (Yeung, 2018). Some researchers, such as 
Baú and Calandro (2019), have recommended 
a human rights-based approach to digital 
technology. Furthermore, if the decision-making 

---

76 
]
DATA COLLECTION AND ANALYSIS TOOLS FOR FOOD SECURITY AND NUTRITION
process of the digital system is hidden from the 
person directly affected by the outcomes, then 
the person may not trust the system. This is why, 
for instance, research for explainable AI is being 
conducted (see, for example, Rudin, 2019).
Currently, there are few FSN efforts aimed 
at increasing the interpretability and 
explainability of AI systems. Khan and Hoffmann 
(2003), for instance, propose and describe 
a menu construction using an incremental 
knowledge acquisition system (MIKAS). The 
diet recommendation system asks experts 
to provide an explanation for each of their 
actions, to include the explanation in the 
system’s knowledge base. Interpretability 
and explainability of algorithms ought to be 
prioritised beyond performance and error 
rates (Côté and Lamarche, 2021). Algorithm 
developers, model builders and domain experts 
could provide explanations for the application’s 
decisions for inclusion in the system’s 
knowledge base and output. Open-source 
initiatives can also contribute to interpretability, 
transparency and explainability of systems. For 
instance, the details of a model can be fully 
described within source code. It is, however, also 
important to be aware that information other 
than the source code may be required to fully 
understand a model, including the nature of 
data, documentation, etc. (Sampson 
et al., 2019). 
Digital technologies that are transparent and 
give users freedom of choice are desirable.
In order to mitigate the risks associated with 
digital technologies, it is also valuable to build 
the capacity of users. For instance: providing 
users with full information, including on risks 
and biases; educating users about their digital 
rights and responsibilities; ensuring that users 
are trained or supported to handle relevant 
technologies; creating an enabling environment 
for users to access the required digital 
infrastructure and digital resources; etc. It is 
important to include stakeholders in the needs 
analysis, design, piloting and implementation 
of digital technologies. When users are involved 
in the process, they are more likely to provide 
contributions to the system development process 
and trust and accept the realized systems 
(Maguire, 2001).
Another concern associated with digital 
technologies is who owns the FSN digital data, 
who has access to it, and who has control over 
its use and implementation. Issues of ownership, 
access to and control of data can lead to risks 
associated with inequitable data access, power 
asymmetry, negative exclusive property regimes 
over data, exclusion (wilful or not) of certain 
types of data, unethical tracking and targeting 
(for instance, through AI-powered unethical 
target advertising), and market dominance by 
organizations and bodies that control the data 
(
SEE BOX 29
). In the process, digital technologies 
could affect the cultural fabric and identity of 
FSN stakeholders (Klerkx, Jakku and Labarthe, 
2019) – for instance, what it means to be a 
farmer (Burton, Peoples and Cooper, 2012; 
Carolan, 2017), and a possible change in the 
culture of farming from a hands-on approach to 
data-driven management (Butler and Holloway, 
2016; Carolan, 2017). Moreover, there are 
cyber-security risks associated with digital 
technologies in FSN (for instance, for smart 
farming as described by Barreto and Amaral 
[2018]). Users and respondents in FSN may be 
concerned about the privacy, protection and 
misuse of their data. They may fear that their 
data may be used to exploit them, may be used 
against them, may end up in the wrong hands, 
or may put them in precarious positions in the 
future. Some researchers (e.g., Clapp and Ruder, 
2020) have argued that digital technologies can 
reinforce existing systems which are considered 
economically, socially and ecologically 
unsustainable and favour specific FSN players 
(Rijswijk et al., 2021).

---

[
77
4 
NEW AND EMERGING DIGITAL TECHNOLOGIES FOR FSN DATA
BOX 29:
CHALLENGES WITH DIGITALIZING SERVICES AND ACCESS: THE CASE OF INDIA’S AADHAAR 
IDENTIFICATION NUMBER
India’s Aadhaar (literally “the foundation” in Hindi) programme, intended to provide a unique 12-digit identification 
number to 1.3 billion Indian residents, was launched in 2009 as a voluntary biometric ID system to smooth delivery of 
public services, such as food assistance and welfare benefits, and reduce fraud. However, since 2014, the biometric 
ID system under Aadhaar is being made compulsory to access more and more basic services and entitlements. 
Failure to obtain an Aadhaar number has sometimes hindered residents’ access to fundamental benefits such as 
rice or wheat at subsidized prices, an important source of food security for many Indians, access to pensions, school 
admissions for children and so on, and this is why proper functioning of the system is essential. Implementing this 
system presents a number of challenges that have led to shortcomings. Some of these shortcomings are caused 
by limited availability of the necessary IT infrastructure, including electricity, to operate the biometric ID systems, 
especially in rural areas. In addition, if someone is unable to go in person for biometric identification, the benefits 
cannot be accessed. While delegation systems exist on paper, in practice they rarely work. This disproportionately 
affects the elderly and the disabled, and those from remote villages. Furthermore, repeated reports of data leaks 
have raised concerns for the privacy of personal records, which is particularly worrying as the Aadhaar identification 
number is not only a condition to receive social support, but increasingly linked to private transactions, including tax 
payments.
The issues presented in the implementation of the Aadhaar programme should be used as a learning experience to 
exercise caution in the adoption of new digital technologies when these are linked to fundamental access to food and 
social protection, as possible technology, infrastructure and capacity constraints can deeply affect the realization of 
the right to food for the neediest and exacerbate inequalities.
Source: Khera, R. (2019) 
In response to the risks associated with data 
ownership, access and control, a responsible 
research and innovation (RRI) approach to 
digital transformation has been proposed for 
use, for instance, in agriculture (Barrett and 
Rose, 2022). The RRI approach is based on 
four main principles: anticipation, inclusion, 
responsiveness and reflexivity. Similarly, Rose 
and Chilvers (2018) propose a more systemic 
approach to map innovations associated with 
digitalisation in agriculture; broadening of 
notions of inclusion in RRI to include a diversity 
of stakeholders; and evaluating responsible 
innovation frameworks in practice to determine if 
innovation processes can be made more socially 
responsible.
It is also important to formulate and enact 
laws, regulations and policies on ethics, 
consent, privacy, data protection, ownership, 
fair competition and copyright. Governments 
and regional and international organizations 
should involve stakeholders in defining and 
implementing appropriate data standards and 
policies in order to minimize the potentially 
negative consequences of data access and 
sharing. Ge and Bogaardt (2015) studied a 
number of data harvesting initiatives in agrifood 
chains to identify the key governance issues 
to be addressed. Examples of data protection 
and privacy laws and regulations include the 
European Union’s General Data Protection 
Regulation (
https://gdpr-info.eu/
) and the Data 
Protection Act of the United Kingdom (of Great 
Britain and Northern Ireland) (
https://www.
legislation.gov.uk/ukpga/2018/12/contents/
enacted
). Such laws and regulations are often 
subject to the oversight of an independent 
authority to ensure compliance and protection 
of individual rights. At a broader level, the UN 
Global Pulse has developed Privacy Principles 

---

78 
]
DATA COLLECTION AND ANALYSIS TOOLS FOR FOOD SECURITY AND NUTRITION
in consultation with experts from various 
sectors. The United Nations Secretary-General’s 
Independent Expert Advisory Group on a Data 
Revolution for Sustainable Development has 
recommended the development of a global 
consensus on principles and standards 
concerning legal, technical, privacy, geospatial 
and statistical standards to, among other things, 
facilitate openness and information exchange 
and promote and protect human rights (FAO, 
2017; UN, 2015) It is worth noting the UN High-
Level Committee on Programmes (HLCP) is 
looking into a global data governance framework 
(
https://unsceb.org/session-report-369
).
It is particularly important for FSN actors 
to protect potentially vulnerable segments 
of society. For instance, Kraak 
et al. (2020) 
propose various actions to protect young 
people from irresponsible digital marketing 
that could negatively impact diets and lifestyle 
choices. Among these proposed actions are 
recommendations that technology firms develop 
policies to protect the digital privacy rights of 
young people; enforce standards for digital 
platforms that support responsible marketing 
to children and adolescents; and ensure that 
digital marketing and media policies are posted 
on the firms’ public websites. Kraak 
et al. 
(2020) also propose that governments develop 
comprehensive national legislation, regulations 
and policies that protect digital privacy and 
restrict the use of all forms of digital marketing 
to children and adolescents; collaborate with 
international and regional bodies to develop 
cross-border policies to regulate transnational 
digital marketing and media practices; monitor 
and evaluate how transnational companies are 
using digital marketing and social media and 
enhance accountability for their practices. (More 
details on governance of FSN data are presented 
in Chapter 5).
It cannot be overstated that early and continuous 
inclusion and involvement of all relevant 
stakeholders is key to the acceptance and 
success of new technologies in the FSN sector. 
Stakeholders include, but are not limited to, 
governments, industry, consumer groups, NGOs, 
farmers and other smallholder producers. 
Although upstream and downstream sectors 
influence the adoption of technologies by 
farmers, they can also learn from farmers so 
that the technologies implemented take into 
account the requirements of the farmers (OECD, 
2001). In order to ensure that everyone is in a 
position to benefit from new technologies and 
that technology-based efforts do not reinforce 
the digital divide, it is important to ensure that 
digital technology implementations are adapted 
to the needs, requirements and contexts of all 
users and stakeholders, especially vulnerable 
groups and those in developing countries who 
have less digital access (due to low internet 
connectivity, for example) and human capital 
(for instance, related to low literacy levels). 
Of course, support to ensure access to and 
utilisation of technologies should indeed be 
provided for all stakeholders, especially for those 
who are vulnerable. Furthermore, during the 
conceptualisation, design and implementation 
process of such efforts, it is also important 
to take into account indirect and long-term 
effects of the digital technologies. Moreover, it is 
instructive to create spaces for FSN stakeholders 
to reflect on how digitalization will affect existing 
FSN innovation systems (Bronson, 2019; Klerkx, 
Jakku and Labarthe, 2019) and to consider a 
policy-driven strategic overview of FSN needs 
and priorities (Regan, 2021).
Involving users and stakeholders in the design 
and implementation of digital applications early 
and throughout the process, it becomes possible 
to anticipate and address the associated risks 
and needs (Rijswijk 
et al., 2021) and significantly 
increases the likelihood that they will accept, 
value, own, support and trust the respective 
technologies. Ortiz-Crespo 
et al. (2021) describe 
a user-centred design process that was used 
to develop a system called Ushauri, to provide 
farmers in Tanzania with agricultural advice. 
Furthermore, with the required capacity 
(such as skills, infrastructure such as open-
source tools), local individuals and groups can 
themselves build digital technology platforms. 
In fact, Carolan (2022) argues that participation 
or inclusivity extends beyond simply making 
sure that voices are heard and that, inclusivity 
includes empowering individuals to build their 

---

[
79
4 
NEW AND EMERGING DIGITAL TECHNOLOGIES FOR FSN DATA
own digital platforms – in contrast to exclusive 
intellectual property regimes.
If experts, users and stakeholders are not 
involved in the design, development and 
implementation of digital technologies, other 
risks may arise, for instance in data collection, 
analysis and interpretation. For example, in the 
case of automated analysis or if there is a lack of 
analytical expertise, without the participation of 
experts and/or local stakeholders there is a risk 
of misinterpretation or overgeneralization. (This 
risk arises, for example, when the computational 
models/algorithms used in the technologies 
do not take into account the social, economic, 
cultural and natural complexities of the target 
people or country). Although machine learning 
can help to improve prediction in nutrition-
related research (for instance in cardiovascular 
risk prediction [Rigdon and Basu, 2019]), 
procedures for model validation in nutrition 
research are often not sound or not well reported 
(Christodoulou 
et al., 2019), hampering an 
objective model comparison in real world case 
studies. Methodologies for model development 
and validation should therefore be more 
carefully designed and reported (Christodoulou 
et al., 2019); or improved upon (Espel-Huynh 
et al., 2021). In this regard, experts can play 
a key role in identifying algorithms that have 
optimal performance and are appropriate for 
specific prediction problems. When designing, 
developing, implementing and researching 
digital technologies, therefore, it is important 
to involve relevant experts to give inputs for or 
during the various data cycle stages (including 
data collection, building the underlying models 
and performing analysis).
Digital technologies should offer FSN services 
and FSN content that are based on and adapted 
from trusted sources, and that take into 
consideration local contexts in order to meet the 
unique needs and preferences of different user 
groups (FAO, 2013b). For equity and inclusivity, 
the global community and international 
organizations should actively and continually 
engage and support the sustainability of 
indigenous knowledge, innovations and capacity 
from the grassroots and local levels, and 
vulnerable groups, thus contributing to empower 
local communities.
On a final note, it is interesting to note that some 
of the new digital technologies, when made 
accessible, can be used to support and enhance 
stakeholder engagement, promote inclusion, 
and support coordination in FSN efforts. These 
facilitative technologies include crowdsourcing, 
crowdsensing, online social media and mobile 
computing.
QUALITY OF DATA
Data quality entails elements such as accuracy, 
completeness, timeliness, validity and 
consistency. While digital technologies can 
enhance data quality (for instance, by validating 
accuracy and ensuring timeliness), there is also 
potential for digital technologies to affect data 
quality negatively. Data collection from users 
or respondents through technologies such as 
online social media, crowdsourcing and other 
mobile computing-based applications is relatively 
subjective and, therefore, subject to factors such 
as deception and carelessness. It has also been 
reported that data collected from 
citizen science
efforts tends to be noisy, that is unreadable by 
analysis programmes (Kelling 
et al., 2015). It 
may be useful to complement such user-focused 
digital technologies with other digital technologies 
or methods that are more objective.
It is worth noting, however, that over-reliance 
on numeric data (on the false presumption 
that such data is more objective) may lead to 
a scenario where data or information remain 
largely incomplete. In many cases, qualitative 
data captures key information about local 
contexts where FSN interventions are or will be 
undertaken in ways that cannot be represented 
by numbers. As was noted in Section 4.1.2, some 
of the new and emerging digital technologies 
support the processing of qualitative data in 
the form of images, videos, audio recordings 
and text. Such technologies can be used 
for data collection (e.g. online social media, 
crowdsourcing and other mobile computing-
based applications), data storage (e.g. cloud 
computing, big data), data analysis (e.g. machine 

---

80 
]
DATA COLLECTION AND ANALYSIS TOOLS FOR FOOD SECURITY AND NUTRITION
learning via sentiment analysis) and data 
dissemination (e.g. information visualization, 
online social media). However, a number of 
digital technologies still only collect or process 
numeric data, given that qualitative data 
collection, processing, codification and storage 
may involve complex processes and be highly 
demanding in terms of time and resources. Still, 
it is important to consider that over-reliance 
on digital technologies that collect, or process 
only numeric data may downplay important 
nuances that can be gleaned from qualitative 
data, and therefore, it is useful to also use 
digital technologies that can effectively manage 
qualitative data.
Moreover, another potential source of inaccurate 
data can derive from distractions in the 
respondents’ uncontrolled settings, which can 
affect the quality of data collected. IoT and 
sensors can give false or misleading readings 
(for instance due to environmental complexities), 
which can translate into potentially detrimental 
agricultural and nutritional decisions and 
actions by farmers and policy-makers. However, 
research (such as Hariri, Fredericks and 
Bowers, 2019), is in progress to overcome these 
limitations, making big “poor-quality” data 
more valuable than small “high-quality” data. 
As such, digital technologies used in real-world 
settings should be constantly monitored, tested, 
calibrated and enhanced and, in some cases, a 
combination of digital technologies or methods 
should be used to ensure data quality.
INTEROPERABILITY OF DATA
Interoperability makes it possible for different 
systems to share, exchange and understand data. 
This is critical when efforts are being made to 
integrate different systems, which, in turn, is key 
to making digital technologies and systems widely 
useful. Interoperability may be necessary at any 
stage in the data cycle. For instance: users may 
want their respective digital applications to be able 
to fetch and analyse FSN data from diverse big 
data or 
cloud computing
sources.
Interoperability initiatives often involve tasks 
such as developing standards and specifications 
(such as using ontologies to provide global 
term identifiers and frameworks to define 
and categorize FSN-relevant terms); building 
mappings for many different sets of standards 
and specifications; curating multiple domains of 
vocabulary; and harmonizing existing vocabularies 
or curating new terms in them. It is important to 
note that initiatives, such as FAO’s AGROVOC,
17
FoodOn (Dooley 
et al., 2018) and CGIAR’s Crop 
Ontology,
18
(described earlier), are efforts that 
contribute to interoperability.
CAPACITY, EQUITY, SCALABILITY AND 
SUSTAINABILITY
Digital technologies involve relatively high 
investment costs, and are expensive for 
some organizations, for farmers with lower 
socioeconomic status, and for other vulnerable 
FSN stakeholders. Some organizations that carry 
out data collection and analysis are finding the 
cost of requisite technological infrastructure 
prohibitive (Sivarajah 
et al., 2017), in addition to 
lacking sufficient personnel with skills in core 
data competencies (for example, in data analysis, 
information visualization, interpretation and 
decision making). Vulnerable FSN stakeholders 
might also not have the capacity to use the 
technologies or interpret data results, or may 
lack altogether access to internet connection and 
digital devices. The use of digital technologies in 
such scenarios may lead to or reinforce existing 
inequalities, such as the digital divide, and to the 
unequal distribution of the benefits of new digital 
technologies, favouring those who can already 
afford them. Furthermore, if technologies are 
implemented without the inputs of vulnerable 
FSN stakeholders, they may become even more 
disconnected from and further marginalised.
It is therefore important to invest in the necessary 
technology, infrastructure and research 
necessary to improve data interoperability and 
data quality, as well as access to and affordability 
17 For further information, see 
https://www.fao.org/agrovoc/about
.
18 For further information, see 
https://bigdata.cgiar.org/digital-
intervention/crop-ontology-2/
.

---

[
81
4 
NEW AND EMERGING DIGITAL TECHNOLOGIES FOR FSN DATA
of technology. It is also important to build and 
enhance human capacity. For instance, by 
training in core data competencies (e.g., data 
collection, data analysis, information visualization, 
interpretation and decision making). Several 
institutions are supporting this training in the FSN 
domain. For example, FAO is offering training in 
such areas through the FAO eLearning Academy 
(Remotely Sensed Information for Crop Monitoring 
and Food Security - Techniques and methods for 
arid and semi-arid areas (
https://elearning.fao.
org/course/view.php?id=155
). Other means for 
building and enhancing human capacity include 
educating users to support the data lifecycle 
process; enhancing user- and indigenous-capacity 
to improve data quality; and educating data 
owners and data producers about privacy, consent, 
data usage, data ownership and the rights they 
have. All stakeholders – data owners, producers 
and respondents – should be informed about 
the purpose of collecting, processing and using 
data and whether the data will be shared with 
other parties. Collaboration can also be useful in 
addressing some capacity concerns. The potential 
benefits of collaboration include: ensuring 
interoperability of technology standards and 
architectures; defining appropriate data standards 
and policies on data access and data sharing; 
pooling digital resources and infrastructure; 
implementing shared services in a synergic 
manner; sharing best practices and mutually 
beneficial information; developing context-relevant 
and user-relevant technological interventions; 
and limiting the potential for technology to be a 
disincentive to meaningful production. In addition, 
efforts on interoperability of data and systems 
can lead to the realisation of open-source tools 
and materials which, in turn, can reduce some 
capacity costs. Responsible automation, as 
described earlier, can also help to alleviate some 
of the capacity challenges.
Furthermore, almost all digital agriculture 
initiatives have to contend with the challenges 
of scaling (how to include locations, users, etc.) 
and sustainability (how the initiatives can extend 
beyond the current funding, etc.) (Florey, Hellin 
and Balié, 2020; Kos and Kloppenburg, 2019). 
Some of the recommendations to overcome these 
challenges include: demonstration of the benefits 
of using digital technologies and tools to support 
decision-making, adoption of interdisciplinary 
approaches and interconnectedness, recognizing 
the need for learning, feedback, partnerships, and 
joint action in multi-stakeholder settings within 
the context of FSN innovation systems (Florey, 
Hellin and Balié, 2020 p.135; Schut 
et al., 2016; 
Shepherd 
et al.,
2020).

---

82 
]
Chapter 5
INSTITUTIONS 
AND GOVERNANCE 
FOR FSN DATA 
COLLECTION, 
ANALYSIS, AND USE
82 
]
Italy, 15 October 2018, FAO Headquarters – CFS annual Plenary.
©IFAD/Daniele Bianchi

---

[
83
5 
INSTITUTIONS AND GOVERNANCE FOR FSN DATA COLLECTION, ANALYSIS, AND USE
P
revious chapters in the report have made 
the case for the importance of using data 
to inform decisions; discussed the type 
of data needed at various levels in the wide 
ecosystem that determines food security and 
nutrition; commented on their current availability 
and the most important gaps; presented 
examples of valuable initiatives that contribute 
at each step of the cycle; presented an overview 
of the major constraints and bottlenecks that 
still affect FSN data systems worldwide; and 
introduced the enormous potential that resides 
in new and emerging digital technologies.
One theme that emerges from this discussion is the 
increased complexity of modern data systems, with 
many actors involved. Nowadays, in every practical 
context – including FSN policymaking at global, 
national or local levels – the collection, processing 
and use of data to reach effective, evidence-
informed decisions involves a distributed (and often 
fragmented) process, with responsibilities held by 
different individuals and institutions, at different 
levels. Ensuring the proper coordination and 
collaboration among the various actors involved 
throughout the data cycle is fundamental to the 
success of any solution. This presents significant 
challenges for the design of an effective data 
governance system and creates an opportunity to 
take a systems approach not only to describe what 
is meant by food security and nutrition (Clapp 
et 
al., 2021; HLPE, 2020), but also when addressing 
the roles that data collection, dissemination 
and analysis play in ensuring food security and 
adequate nutrition for all.
The multiplicity of actors involved in generating 
and using data for public good, together with 
the special nature of data in the digital era, 
creates complex challenges for data governance. 
This is a very active area of investigation, and a 
consolidated view of which the most appropriate 
governance mechanisms are and which 
institutions should lead and coordinate them is 
still far from being crystallized. In fact, there is 
not even an agreed-upon of data governance. 
The DAMA Guide to The Data Management 
Body of Knowledge defines it as, “The exercise 
of authority and control (planning, monitoring, 
and enforcement) over the management of 
data assets.” (DAMA International, 2009, p.37). 
Abraham, Schneider and vom Brocke define it 
as “a cross-functional framework for managing 
data as a strategic enterprise asset”, highlighting 
its broad scope to include the specification 
of “decision rights and accountabilities for 
an organization’s decision-making about its 
data” and the formalization of “data policies, 
standards, and procedures” (Abraham, Schneider 
and vom Brocke, 2019, p.425-26). These 
definitions refer to data as an asset owned by a 
specific firm, company or organization (reflected 
implicitly in the expression “decision-making 
about
its data”), that has clearly established, 
full authority and control over the data. We find 
these definitions too narrow to be applied to 

---

84 
]
DATA COLLECTION AND ANALYSIS TOOLS FOR FOOD SECURITY AND NUTRITION
FSN data, which encompasses a wide range 
of types of data, including data “owned” by 
governments, data “owned” by private entities 
and, quite importantly, data apparently owned by 
no one, which is potentially available to anyone 
who has the skills to access it from the internet. 
The evolving data landscape that is taking shape 
as the digital revolution continues, especially 
following recent global events such as the 
COVID-19 pandemic, introduces new challenges 
for data governance, highlighting the need for it 
to transcend boundaries – of firms, organizations 
and even national governments. As noted by the 
Center for Strategic and International Studies 
(CSIS), in the publication “Data Governance 
Principles for the Global Digital Economy”: