Data collection and analysis tools for food security and nutrition

Partnership in Statistics for Development in the 
21
st
Century (PARIS21), shows how the increased 
commitment on the part of external partners to 
support statistics has been mostly directed to 
economic and demographic statistics, with less 
focus on environmental and agricultural statistics 
(OECD, 2019). According to a recent report on 
The 
implementation of the Cape Town Global Action 
Plan on Sustainable Development Data (World 
Bank, UNSD and Paris21, 2022), two-thirds of NSOs 
in International Development Association (IDA) 
countries
12
experienced either moderate or severe 
delays in budget disbursement in the last fiscal 
year, which hampered the implementation of their 
work programmes, and nearly 70 percent of them 
prioritized the need to address funding shortages in 
business and agricultural census programmes over 
the next three years.
Solutions have been sought in exploring ways to 
reduce the cost of data generation, for instance, 
through increased reliance on secondary data 
rather than collecting primary data, imposing, 
however, additional requirements in terms 
of analytic capacity to ensure that data from 
different sources is integrated properly and avoid 
compromising the quality of the data series and 
comparability over time.
Another solution has been recourse to services 
for data collection, analysis and dissemination 
offered by private companies and professionals. 
While useful to partially fill the data gaps, 
such initiatives may raise various concerns, for 
example regarding privacy, data access and data 
governance.
13
Additionally, increased reliance 
on private data services may further erode the 
relevance and independence of NSOs.
A third solution, entailing the adoption of new 
technologies for data generation and collection, 
may certainly help (see more in Chapter 4). 
However, new technologies usually require initial 
investments and sustained support to ensure 
that the technologies are effectively used. One 
important aspect related to finance that has 
prevented useful innovations from becoming 
a permanent feature of data generation has 
been the difficulty in securing stable funding to 
keep the operations in place. When innovations 
have been promoted through externally funded 
projects, despite positive results, lack of 
sustained funding has halted their large-scale 
implementation.
In some cases, attempts to reduce costs to cope 
with limited resources may have detrimental 
consequences on data quality and relevance. 
In sampling-based inferences, such as when 
conducting farm or population surveys, or 
12 These are countries considered eligible for support, according 
to the criteria established by the International Development 
Association, based on per capita gross national income being below 
an established threshold, or lacking the creditworthiness needed 
to borrow from the International Bank for Reconstruction and 
Development (IBRD). Currently, 74 countries (39 in Africa, 14 in East 
Asia, 6 in South Asia, 4 in Europe and Central Asia, 8 in Latin America 
and the Caribbean, and 3 in the Middle East and North Africa) are 
eligible. For a full list, see 
https://ida.worldbank.org/en/about/
borrowing-countries
.
13 See for example: 
https://www.oecd.org/digital/trusted-
government-access-personal-data-private-sector.htm
.

---

3 
CONSTRAINTS, BOTTLENECKS (AND SOME SOLUTIONS) FOR EFFECTIVE USE OF FSN DATA
conducting food-composition studies, one 
way to reduce costs is by reducing sample 
size, but these reduced samples may be too 
small to produce indicators at the needed level 
of precision. In other cases, particularly for 
time-sensitive decisions, the need to secure 
the necessary funding may delay the survey 
operations to the extent that the usefulness 
of the information they provide may be 
compromised (
BOX 17
).
BOX 17:
THE HIGH COST OF FSN-RELEVANT SURVEYS
Population surveys that provide key information on respondents’ dietary intake and nutritional status may 
require enumerators to perform individual nutrition assessment (collecting anthropometric, biochemical, clinical 
assessment and dietary intake data). Training the enumerators and implementing the necessary field operation is a 
costly and labour-intensive process.
Similarly, food production surveys that seek to reach small farmers and fishers in interior areas require the 
mobilization of many enumerators over large distances, all of which increase overall survey costs. While newer 
methods, such as the use of smartphones, may reduce the time spent in face-to-face data collection, and therefore 
potentially reduce the number of needed enumerators, it is important to evaluate disparities in the ownership of 
digital devices and access to technology and knowledge among the vulnerable group, including women and small 
farmers.
In countries or even the regions with multiethnic populations, many languages are spoken and understood. This 
adds a layer of complexity to the process of data collection (such as the validation of tools in different languages, 
verifying the language competencies of the enumerators, etc.) and is expensive. When these demands arise in the 
context of existing financial constraints, feasibility is usually prioritized over representativeness.
In many countries, the cost of validating dietary assessment tools, such as food-frequency questionnaires or 
screeners with objective biomarkers, has been a major constraint and resulted in limited validation efforts. 
This has often cast doubts on the quality of the data and, thus, on the validity of results arising from the dietary 
surveys. Validation of self-reported dietary intakes, estimation of micronutrient intakes or levels of toxicity 
require biochemical analysis. This is an expensive, resource-intensive process that requires elaborate logistical 
arrangements, which are prohibitive in many projects. The lack of objective validation of dietary intake remains a 
consistent challenge in interpreting dietary data.
Finally, dietary data needs further processing in terms of nutrient analysis. Such analysis, followed by the creation 
of comprehensive food composition databases, is an expensive undertaking and unaffordable for many low-income 
countries.
Inadequate research 
infrastructure
Insufficient funding and the lack of well-trained 
human capital also result in 
inadequate research 
infrastructure
at the national level to support 
every stage of the data cycle (
FIGURE 2
). Beyond the 
insufficiency of human and financial resources, 
inadequate research infrastructure influences 
how institutions set their priorities and actions 
for research. Under-funded NSOs, overwhelmed 
by competing priorities, tend to focus less on 
food production statistics and certainly not (due 
to underfunding and to a lack of capacity for 
system thinking) on generating statistics from 
across FSN-relevant sectors (agriculture, social 
protection, health, industry and trade) or covering 
the six dimensions of FSN. This is especially so in 
[
51

---

52 
]
DATA COLLECTION AND ANALYSIS TOOLS FOR FOOD SECURITY AND NUTRITION
developing countries where inadequate funding 
further stresses organizational capability and 
makes it necessary to prioritize just one aspect 
of FSN. In such countries, the inadequacy of 
research infrastructure is evident in the lack of 
research quality frameworks and methodological 
expertise for timely, relevant and sufficient data 
collection and validation; lack of prior data; lack 
of data processing and analysis capabilities; 
and poor practices in data dissemination and 
communication (Filter 
et al., 2022; Jones et 
al., 2017). Finally, infrastructure and resource 
constraints also hinder 
data-digitalization
efforts, 
further limiting data availability and accessibility.
The lack of adequate modern data infrastructure, 
especially in low-income countries, also limits 
effective data collection, analysis and use. Due to 
lack of access to broadband infrastructure in some 
developing regions, such as Sub-Saharan Africa 
and South Asia, where Internet usage gaps are as 
high as 49 and 64 percent, respectively (Lishan and 
Minges, 2018). 
Social gradients
also influence 
the placement of cellular and mobile services 
and, thus, the penetration and quality of services 
in remote areas. Social divides in digital access 
and literacy is a further impediment to reaching 
disadvantaged stakeholders, such as women in low 
and middle-income countries and smallholders 
(LeFevre 
et al., 2021). Thus, while technological 
advances may reduce costs and widen the reach 
of surveys and help to fill some gaps in data 
availability, the social divide may lead to the 
underrepresentation of those with poorer digital 
access and literacy (LeFevre 
et al., 2021). This can 
result in policies and interventions that are based 
on data generated from skewed sampling, which 
may not serve unrepresented stakeholders who 
may have the greatest need for data-driven policy 
and support (Bell 
et al., 2017; LeFevre et al., 2021). 
Therefore, the adoption of newer technologies 
without considering the local context and the 
impact of their use on users and beneficiaries can 
further exacerbate inequalities, as illustrated in 
boxes 18 to 21.

---

3 
CONSTRAINTS, BOTTLENECKS (AND SOME SOLUTIONS) FOR EFFECTIVE USE OF FSN DATA
BOX 18:
THE COMPLEXITY OF NUTRITION ASSESSMENTS
Nutrition assessments
are a resource-intensive undertaking and, therefore, many of the constraints and bottlenecks 
discussed in this chapter limit the complete range of assessments (including anthropometry, biochemical and 
clinical and dietary intake). As previously mentioned, resource constraints can affect the availability of data and 
capacity constraints can affect the quality of the data available. In particular, biochemical and clinical assessments 
are resource-intensive and, therefore, multiple constraints acting in parallel result in a serious lack of data in this 
regard.
An important domain of nutrition assessment is the accurate estimation of dietary intake in populations. Data in 
this area are inconsistent, outdated, national food composition databases are incomplete, due to lack of support 
for institutions involved in developing the databases; all these factors challenge the accuracy of nutrient intake 
estimations in various countries and prevent their utilization by multiple users. The lack of comprehensive food 
composition databases with adequate representation of both plant and animal, aquatic and land-based foods 
consumed in the country, means that many countries rely on the databases of neighbouring countries or global 
databases to estimate nutrient intakes. The use of inaccurate food composition data may lead to erroneous research 
results, flawed policy decisions (particularly in nutrition, agriculture and health), misleading food labels, false health 
claims and inadequate food choices (Charrondière, 2017).
The Malabo Montpellier Panel report (2017) clearly states that “African governments continue to lack the data 
necessary to effectively combat malnutrition”, as “few national governments collect the data required to inform 
decision makers about what people eat, and there is no functioning global dietary database.” (Malabo Montpellier 
Panel, 2017, pp. 11–12). A recent review on global dietary surveillance (Micha et al., 2018) confirms the non-
availability or inadequacy of country-specific food composition tables (FCT) and food composition databases (FCDB) 
as one of the major challenges linked to the limited availability of global dietary data which are needed for a wide 
variety of purposes, including modelling, designing and implementing context-specific dietary policies to reduce 
disease and disparities at national and regional levels. Strengthening regional collaboration and establishing 
reference laboratories may provide a cost-effective solution. Another issue which must be tacked in nutrition 
assessment, is the lack of representation of 
indigenous and forest foods
in food composition databases. This hinders 
the accurate evaluation of dietary intakes in indigenous populations (FAO, 2013a). INFOODS also tackles constraints 
in paucity of food composition data.
BOX 19:
ON FOOD SAFETY DATA
Low- and middle-income countries often lack resources to invest in improving their own national food safety 
regulatory frameworks and, therefore, rely on Codex standards as the basis for such legislation. However, Codex 
standards may overlook practices that are common in small-scale food production and their connected value 
chains (Humphrey, 2017). Both the European Food Safety Authority (EFSA) and Codex Alimentarius have databases 
containing food safety parameters, but these are not available as open access. Food safety data, specifically, may be 
regarded as sensitive to a country as levels above maximum limits can result in export bans and affect trade. Also, 
financial and human resources for food safety monitoring programmes are major constraints in enabling timely and 
relevant data collection related to food safety.
[
53

---

54 
]
DATA COLLECTION AND ANALYSIS TOOLS FOR FOOD SECURITY AND NUTRITION
BOX 20:
THE WOMEN EMPOWERMENT IN AGRICULTURE INDEX
The Women’s Empowerment in Agriculture Index (WEAI) has been designed to track gender equality and the 
transformation of gender norms (Malapit et al., 2019). The granularity of data allows for disaggregation by age-
group; gender; location; agroecological region; urban, peri-urban or rural residence; ethnicity and socioeconomic 
and occupational class. This, in turn, also allows for in-depth understanding and targeted action. Sampling that 
allows for such disaggregation along the food supply chain facilitates understanding of the contribution to food 
production from both formal and informal sectors, and their disaggregated food consumption patterns. When 
disseminated efficiently to the relevant stakeholders, this information can facilitate the involvement of the vulnerable 
groups in decision-making and aid in their ownership of targeted initiatives. Such efforts are important to promote 
equity in access to FSN data for policies and decisions at grassroot and local levels, taking into account local diversity 
and context.
BOX 21:
SATELLITE TECHNOLOGIES FOR IMPROVED DROUGHT ASSESSMENT (SATIDA)
To improve reach, granularity and affordability in data collection, some countries have developed accessible digital 
technologies for monitoring food security that help bridges many of the constraints referred to in this section, 
improving the granularity of the data while applying a simple and affordable process. One such example is the 
SATIDA (Satellite Technologies for Improved Drought Risk Assessment) project, which was developed to support 
Doctors without Borders. At the regional and national levels, timely and granular data that allow for evaluation of 
impact of innovative value-chain solutions and factors that can improve their uptake are also lacking (Committee on 
World Food Security, 2021).
HUMAN RESOURCE CONSTRAINTS
The lack of adequate human capital within public 
institutions responsible for FSN data generation, 
curation and dissemination, is often cited as a 
major constraint to data collection and analysis 
in many countries. Human resources and staffing 
have a huge impact on the availability of sufficient, 
timely and high-quality data.
Constraints related to data 
collection
The need for well-trained personnel in data 
collection using traditional survey methods has 
been acknowledged time and again (Krosnick, 
Presser and Husbands, 2015). Dietary data 
collection, for example, requires specific skills, 
including the ability to select and properly use 
the most appropriate dietary assessment data 
collection instrument, to assist respondents 
in estimating portion sizes, and to ensure 
completeness of the reporting.
Although new technologies can facilitate data 
collection, they do not eliminate the need for 
considerable numbers of adequately trained 
competent personnel (Aweke 
et al., 2021). 
Technology used to interview people from remote 
locations, such as computer assisted telephone 
interviews (CATI) or internet-based technologies, 
might reduce the need for human resources, as 
might automating some of the routine or time-
consuming tasks, but does not replace them 
entirely. For example, reliable measurement of 
certain outcomes, such as anthropometry and 
the measurement of local food environments, 
will always require the physical presence of 
enumerators at the location. Furthermore, 
harnessing the newer technologies to organize, 

---

[
55
3 
CONSTRAINTS, BOTTLENECKS (AND SOME SOLUTIONS) FOR EFFECTIVE USE OF FSN DATA
analyse and disseminate multidimensional data 
usually requires technical skills that existing data 
collection and analysis staff may not have. Effective 
implementation of these technologies also 
demands soft skills, including management and 
leadership. The lack of these skills among existing 
staff can collectively affect institutional capabilities 
and arrangements for data processing, analysis 
and dissemination. The training required to develop 
these capabilities may be difficult to put in place 
and take time, and this could limit or delay the 
adoption and use of these new technologies (ILO, 
2016, 2020).
While the importance of well-trained personnel 
in data collection and analysis for FSN is 
acknowledged, resource constraints make it 
imperative to balance between the need for 
specialization and the 
sustainability of training 
and capacity-building
efforts. Capacity-building 
programmes such as those included under the 
EAF-Nansen Programme, where students are 
provided a stipend and trained in Norway with the 
opportunity to collaborate with the host institution 
upon return, is one example of a sustainable 
capacity-building programme (
https://www.fao.
org/in-action/eaf-nansen/news-events/detail-
events/en/c/1309584/
). Many European Union 
Funding Programmes also have consideration for 
the sustainability of the capacity-building efforts 
they fund. Despite these efforts, the sustainability 
of capacity building is oftentimes challenged 
by shortcomings in local environments, such 
as lack of job opportunities, poor remuneration 
and existing environments which do not provide 
autonomy. This results in the brain-drain that 
afflicts the Global South.
To address constraints in data analytical 
capabilities, the FAO provides statistical support to 
member countries. The success of these initiatives 
is documented with countries in the Southeast 
Asian Region have shown the highest gains in 
terms of statistical competency over the last 
decade (OECD, 2019). However, the ultimate impact 
of the support provided to build capacity is limited 
by the narrow assessment of capacity of national 
statistical systems.
With reference to the challenges posed by the 
diffusion of new technologies in agriculture, Florey, 
Hellin and Balié (2020) highlight that:
1.
Many binding constraints faced by smallholder 
farmers are associated with basic capacity issues. 
For instance: smallholder farmers “are not 
organized collectively, they have limited experience 
of market negotiation, and little appreciation of 
their capacity to influence the terms and conditions 
upon which they engage with the market, and they 
have little or no information on market conditions, 
prices, and quality of goods (Shiferaw 
et al., 2011).
2.
In geographies where markets for increased 
inputs do not exist because the private sector 
initiative and participation have not been 
sufficiently stimulated (Ricker-Gilbert 
et al., 2011; 
Ghins 
et al., 2017), pushing for higher-yielding 
technologies (such as modern crop varieties) to 
increase productivity merely ensures that input 
prices can be more readily controlled by the low 
number of agro-dealers. As a result, the market 
power exercised by too few operators will lead to 
depressed farm-gate prices because of continuing 
high input prices.
3.
There are many farmers for whom increasing 
productivity and greater access to markets are 
not a priority, instead, they focus on off-farm or 
non-farm activities with a view to temporarily or 
permanently exiting from farming (Mausch 
et al., 
2018).
Constraints related to the lack of 
data processing, analytical and 
dissemination capabilities
The reliability and availability of FSN data are 
often limited due to (i) lack of capabilities in 
data processing and analysis and (ii) lack of data 
analytical capabilities.
The analysis of dietary assessment data, for 
example, requires specific skills, such as the 
ability to choose an appropriate food composition 
table given the list and detail of dietary intake 
data and the ability to match food listed in food 
composition databases with the description of food 

---

DATA COLLECTION AND ANALYSIS TOOLS FOR FOOD SECURITY AND NUTRITION
items included in the data collected, even when 
there is no perfect match. Similar considerations 
can be made with respect to the analysis of food 
security data from surveys. For instance, in the 
initial implementation of the FIES food insecurity 
in survey tool, FIES data was collected in a large 
number of surveys throughout the world before 
a sufficient number of trained analysts had the 
time to acquire the necessary analytic skills to 
process the data properly, especially in low- and 
lower middle-income countries. Consequently, 
various reports were produced in which results 
were misleading, as they were based on incorrect 
assumptions made during analysis.
Concrete examples of how processing capabilities 
affect the quality of dietary data are seen in 
performing dietary intake assessment, food 
composition analysis and biomarker assessments 
relating to micronutrient intake and food toxicity 
assessments. To facilitate data processing and 
analyses, various automated procedures have been 
proposed, which presents opportunities, but also 
risks (
SEE BOX 22
).
BOX 22:
OPPORTUNITIES AND RISKS IN THE USE OF AUTOMATED DATA ANALYSIS
Recent technological advances in dietary assessment have integrated the various steps in dietary analysis, 
using dietary analysis platforms that have offline and online capabilities (
https://www.fao.org/infoods/infoods/
software-tools/en/
). This reduces the potential for errors arising from manual data entry and its subsequent 
transcription. However, many of these software that allow for modular usage of local food composition databases 
are not open-access, and their lack of affordability limits widespread uptake in low- and middle-income countries. 
Another limitation is that they require capabilities in the appropriate use of food coding in dietary intake analysis. 
Standardizing data coding as part of quality assurance and data processing is another important step that may not 
be properly addressed owing to lack of expertise, specifically when quality research frameworks do not exist. For 
instance, standardization of food coding is an important step in dietary analysis that matches foods in the dietary 
assessment obtained with foods in the nutrient database. As diets are complex and the variety of foods consumed is 
greater than those reflected in the food database, matching of foods is challenging and requires expertise, including 
knowledge of the local cuisine. Additionally, foods consumed simultaneously, like coffee with milk, are given codes 
that identify these recurring combinations. The combination codes, when appropriately used in the database can 
aid holistic dietary pattern and quality analysis and reveal more visible and accountable patterns that may impact 
nutrition security and health (Mason et al., 2015).
An attempt by the European Food Safety Authority (EFSA) to provide a common link to data sources across different 
food safety domains is the FoodEx2 project (Nikolic and Ioannidou, 2021). FoodEx2 provides descriptions of a large 
number of individual food items aggregated by food groups and broader food categories within a hierarchical 
structure. The Food Ex2 facet descriptors included in the classification system are also mapped to national food 
composition database compilers from 14 European countries. This expands the dataset to include harmonised 
information on the most common composite recipes of European countries and harmonised information on food 
supplements and provides an updated food composition database with over 1 750 foods (Roe et al., 2013).
56 
]
Apart from the data processing abilities, several 
constraints related to analytical capabilities have 
been identified in FSN areas. Analytical challenges 
can involve deficiencies in data measurement 
capabilities (measurement techniques, independent 
from human resources training) or insufficient 
capabilities in data analysis based on limitations 
in computing software. One example with regards 
to lack of analytical capabilities is the challenges 
faced in assessing dietary biomarkers. While the 
use of dietary biomarkers improves the accuracy 
of dietary intake estimations, its implementation 

---

[
57
3 
CONSTRAINTS, BOTTLENECKS (AND SOME SOLUTIONS) FOR EFFECTIVE USE OF FSN DATA
requires extensive sample collection, storage, 
transportation, processing and analytical abilities. 
Micronutrients and toxicity analysis in food require 
sophisticated laboratory equipment and related 
methods that are prohibitively expensive to the 
LMICs. This lack of food and biochemical analytical 
capabilities results in incomplete nutrient lists in 
the food composition tables of many countries.
A related constraint is the insufficient 
data 
analytic capacity
(that is, powerful computers) 
needed to process large amounts of available 
data and information. The collection and use of 
multidimensional big data sets also introduces 
complexities that may require upskilling of the 
current staff.
Insufficient capacity to effectively disseminate, 
interpret and communicate data limits the utility of 
the data and hinders advocacy efforts for continued 
investment in FSN-related data collection. After 
data collection and analysis, results are often 
communicated only in the form of tabulated data, 
with relatively 
little interpretation and analysis
(FAO, 2015; OECD, 2019). While awareness is 
growing of the importance of supporting data 
use with proper analytical briefing on how the 
data are obtained from elementary information 
(Hicks 
et al., 2019; Vaitla et al., 2018; Sethi and 
Prakash, 2018), the lack of such products can 
hamper data-informed policymaking and targeted 
interventions to address the problem (FAO, 2015). 
Moreover, skills restricted strictly to statistical 
domains may be insufficient with the emergence 
of advanced technologies in data production 
with increased complexity, and the involvement 
of new data providers and users. There is also 
a lack of emphasis on data communication and 
dissemination. Additionally, lack of availability of 
the information in local languages hinders data 
utilization by creating language barriers. Given 
that too few NSOs in developing countries monitor 
the use of their data (Sethi and Prakash, 2018), it 
is difficult to gauge the actual utility of the data. 
It is important to obtain this data and estimate 
bottlenecks that prevent effective data usage to 
strategize remedial measures.
INADEQUATE INSTITUTIONAL 
ARRANGEMENT AND DATA 
GOVERNANCE
This section describes issues relating to data 
governance that arise from the lack of stakeholder 
engagement, lack of coordination among agencies 
and lack of transparency and appropriate 
regulatory frameworks.
CONSTRAINTS THAT LIMIT 
STAKEHOLDER ENGAGEMENT
The usability of data is limited when stakeholders 
have not been involved in the survey planning 
and there is inadequate dissemination or access 
to information on what data are available and 
how they can be used by the stakeholder. These 
constraints to the access and use of data for 
improved decision-making make it difficult to 
advocate for further funding and commitments 
towards the collection and analysis of FSN data.
Specific concerns with regards to 
human rights 
and privacy
arise when stakeholders are not 
involved in the collection of data, specifically among 
vulnerable populations, including indigenous 
populations. (These issues are discussed in detail 
in Chapter 5). Adequate representation of diversity 
and inclusion of minorities and the ability to 
disaggregate data for specific populations are also 
closely related to lack of stakeholder engagement 
and the limitation this poses to the utility of the data 
in decision-making in these contexts. Inadequate