The capability approach for children

Several approaches and methodologies have been suggested to conceptualise and measure child wellbeing and child development. These approaches are rooted in different scientific disciplines, including nutrition science,^{(Reference Requejo, Strong and Agweyu18)} philosophy,^{(Reference Hassoun19)} psychology,^{(Reference Lippman, Anderson Moore and McIntosh20)} and economics.^{(Reference Gibbons, Sprong and Chzhen21)} Examples of these indices are the WHO’s D-Score^{(22,Reference McCray, McCoy and Kariger23)} or the UNICEF Early Childhood Development Index, the Nurturing Care Framework,^{(Reference Wertlieb24,25)} and the Child Flourishing Index.^{(Reference Clark, Coll-Seck and Banerjee26)}

Within the field of nutrition science, child growth has been conceptualised as the change in weight and height, that is, reflecting only the physical dimension. Our IUNS Task Force advocates for re-defining the concept of child growth in the context of the capability approach,^{(Reference Haisma, Yousefzadeh and Boele Van Hensbroek13)} to make it a multidimensional concept that includes the environmental context in which a child grows, since this context affects physical, social, and psychological child growth. Capabilities capture not only the actual situation of a child, but also their opportunities to grow in physical and non-physical dimensions, conditional on the status of their parents and caregivers and shaped by wider structural opportunities.

The capability approach guides researchers on how to build multidimensional indices. In capability theory, endowments are the resources or assets a person possesses, such as physical goods, personal traits, or social entitlements. These endowments can potentially be used to achieve various outcomes, called ‘functionings’. Functionings refer to a person’s achieved life outcomes, such as being healthy, having a good job, being safe, or being educated. However, simply possessing endowments does not guarantee the realisation of desired outcomes. There must be supportive social conditions, and the ability to convert resources into actual achievements depends on conversion factors – which can be personal (like physical ability or knowledge), social (such as cultural norms or discrimination), or environmental (including infrastructure or climate conditions).

In the original capability approach proposed by Sen,^{(Reference Sen27)} people are free to choose their functionings by converting their commodities into a vector of characteristics. For example, a bicycle is a commodity and ‘playful’ and ‘transportation’ are its characteristics. A bicycle can have a playful characteristic for a child that uses the bicycle only to play, but for children that use the bicycle to go to school, the bicycle also has a transportation characteristic. Capabilities refer to the practical possibility of individuals to achieve functionings. Sen and all capability approach advocates assert that focus should be on addressing capability deprivations and protecting and expanding capabilities.

In the capability approach for children, it is important to consider that (1) the capabilities of children are age-dependent (among other factors such as sex), that is, the capabilities change depending on the age of a child, and (2) there exists an inter-generational transfer of capabilities from parents and caregivers to children, and hence parents and caregivers influence the capabilities of children. For example, recent evidence indicates that there is an increased risk of a child not attaining expected educational standards at age 5 if their mothers were anaemic at 28 weeks of pregnancy.^{(Reference Olga, Sovio and Wong28)} In terms of age-dependency, changing historical environments and cohort effects imply that the capabilities of children and the conversion of commodities are different from those of adults, and evolve with age.

Thus, in the case of children, capability theory must be adjusted to account for the endowments and capabilities of their parents and caregivers and the structural opportunities or shortcomings that shape parents’ and caregivers’ endowments. The achievements of children – as being well-nourished – are termed functionings, and capabilities are defined as the opportunities to achieve these functionings; for example, the opportunity to have a nutrient-diverse meal is a capability that enables being well-nourished. Additionally, the conversion factors affecting children are influenced by those of their parents and caregivers. For example, if parents/caregivers are constrained in utilising their endowments (e.g. due to wars, economic sanctions, or the fact that natural disasters reduce food availability), providing a nutrient-diverse meal for their child could be extremely difficult. To incorporate the insight that children’s conversion factors are nested within or influenced by those of their parents (shaped by the contextual and structural factors), we expanded the original formalisation of capability theory. In its basic form, the capability set of a child ${Q_i}\left( {{E_i}} \right)$ is defined as the set of possible achieved functionings ${a_i}$ , where each ${a_i}$ is the result of applying a choice function ${f_i} \in {F_i}$ to the outcome of a conversion function $c\left( \cdot \right)$ acting on the child’s endowments ${e_i} \in {E_i}$ . Formally, this is written as:

$${Q_i}\left( {{E_i}} \right) = \left[ {{a_i} = {f_i}\left( {c\left( {{e_i}} \right)} \right),\;{\rm{for\;}}{f_i}\left( \cdot \right) \in {F_i}\;{\rm{and}}\;{e_i} \in {E_i}} \right]$$

Here, ${Q_i}\left( {{E_i}} \right)$ represents the capability set of child $i$ , ${a_i}$ is the achieved functioning, ${f_i}$ is the function representing the child’s choice, $c\left( \cdot \right)$ is the conversion function, and ${E_i}$ is the set of endowments available to the child. ${F_i}$ denotes the set of functionings that are available to be realised by the child. This basic structure assumes a direct transformation from endowments to functionings through conversion and choice.

However, when considering children, this framework needs to be expanded to reflect that children’s abilities to convert endowments into functionings are deeply dependent on their parents or caregivers. The capability sets of children are not formed in isolation, but rather through a nested system of dependencies. In this expanded model, the child’s conversion function itself becomes dependent on the conversion outcomes of the parent, and functionings of parents and caregivers become endowments for children, which children in turn can transform into capabilities through conversion factors.

Moreover, the capability approach not only includes the agency of parents/caregivers, but also recognises the agency of children and, that is, the child’s voice, choice, and participation in matters affecting their lives (e.g. education, play, relationships, cultural expression). This approach shifts policy from a top-down model (what adults think children need) to one that includes children’s perspectives (e.g. participatory methods in education, child-inclusive surveys, rights-based approaches). Thus, children are not only passive recipients of care and protection but active agents in shaping their environments, families, schools, and communities.

Formally, if parental endowments are denoted by ${E_p}$ , and the parental conversion function by ${c_p}$ , the parental endowments ${e_p} \in {E_p}$ will be transformed into enabling conditions for the child. The child’s own conversion function, ${c_i}$ , now takes two arguments: the child’s own endowments ${e_i}$ , and the converted parental conditions ${c_p}\left( {{e_p}} \right)$ . The resulting achieved functioning ${a_i}$ is now represented as a composite:

$${Q_i}\left( {{E_i},{E_p}} \right) = \left[ {{a_i} = {f_i}\left( {{c_i}\left( {{e_i},{c_p}({e_p})} \right)} \right),\!\!\;{\rm{for\;}}{f_i}\left( \cdot \right) \in {F_i},{e_i} \in {E_i}} ,{e_p} \in {E_p}\right]$$

This formulation acknowledges that the effective capabilities of children emerge from a layered structure in which parental capacities, resources, agency, and choices directly shape the child’s context. The child’s own conversion factors, and ultimately their functionings, are embedded within and shaped by the prior conversion of parental endowments. This recursive logic highlights the structural interdependence at the core of capability formation in early life and aligns with the real-world observation that children’s wellbeing and opportunities are fundamentally mediated through the conditions created by those who care for them.

Additionally, functionings in multidimensional child growth depend heavily on the child’s age, as functionings represent realised wellbeing and quality of life. For example, a bicycle, an endowment, can be used by a child for play (increasing leisure) or to enable the functioning of commuting to school efficiently. However, if the parents are not available for teaching the child to ride a bike, or if the child lives in an area without safe roads, or faces societal discouragement, these conversion factors may hinder this functioning. Another example is access to vaccinations: if the child or their parents and caregivers’ distrust medical institutions, cannot take time off work, or face logistical barriers, the functioning of being vaccinated may not be achieved. Thus, the actual functionings depend on the interplay of conversion factors affecting both children and parents.

The diagram below presents a multidimensional capability framework for child growth, illustrating how physical and non-physical child growth is shaped by interactions across ecological systems and contextual structures. At the centre lies the child, whose achieved functionings at a given time reflect their current wellbeing. These functionings are influenced by direct interactions within the micro-system (family, caregivers, teachers, peers), which are embedded within broader meso-, exo-, and macro-systems encompassing neighbourhood dynamics, institutional structures, and human rights frameworks (Figure 1).

Figure 1.

Theoretical framework for multidimensional child growth that combines an eco-biological perspective with the capability approach. Adapted from Yousefzadeh et al.^{(Reference Yousefzadeh, Biggeri, Arciprete and Haisma17)}

Commodities, resources, and services (entitlements) flow into this system, but their effectiveness depends on conversion factors – such as caregiver education, social norms, and local infrastructure – that influence how entitlements are transformed into real opportunities, or capability sets. These sets represent the genuine opportunities available to a child, which may or may not be realised as achieved functionings in the future. The framework emphasises that physical and non-physical child growth is a dynamic process, with feedback loops and reciprocal interactions continuously shaping capability expansion or reduction.

Operationalising the capability approach for children

Operationalising the capability approach for children implies translating the theoretical framework of the TMICG-IUNS into an empirical index. This can be challenging due to the uncertainty over which dimensions should be selected, and the fact that capabilities (opportunities) cannot be observed directly. In particular, the ‘unobservability’ of capabilities implies that statistical methods are needed to estimate the real opportunities that children have in multiple inter-related dimensions. The levels of achievement or deprivation for each of the capabilities and their associations with other factors would be instrumental to estimate the real opportunities that children have in multiple inter-related dimensions.

At the first step of operationalisation of CFCG, we aimed to identify the capabilities for child growth in a climate vulnerable setting, namely, Haor areas of Bangladesh.^{(Reference Chakraborty29–Reference Chakraborty, Darak and Haisma32)} Haor areas are geographically vulnerable due to long-term flooding. We investigated how the contextual factors related to ecological aspects, social barriers, and gender issues affected the capabilities that parents had to prioritise in their context to achieve healthy child growth outcomes. We conducted our fieldwork in two rounds. In the first round (conducted from December 2016 and January 2017), we conducted in-depth interviews and focus group discussions (FGDs) with the parents of children 2 years of age. Then in the second round, (February 2019) we used visual methods to co-create knowledge in our discussions with the parents to validate the list of capabilities that emerged from the first round. The analysis was done using an iterative process of inductive and deductive reasoning with the constitutive elements of the CFCG.^{(Reference Hennink, Hutter and Bailey33)} The findings generated a validated set of capabilities for child growth at the child, mother, father, and household level employing the emic perspectives of the parents living in Haor.

At the child’s level, the parents discussed the capability to stay away from disease and to eat well, the capability to stay happy and to be playful, and the capability to be born with God’s blessings and with the hereditary traits needed to grow. At mothers’ level, they mentioned the capability of staying healthy and nourished, the capability of staying away from violence, and the capability of practicing autonomy in allocating time for childcare. At fathers’ level, they described their struggles to save the lives of their wives and children, particularly during periods of pregnancy and delivery. At the household level, the parents discussed the capability to live in a safe shelter, the capability to be mobile, the capability to overcome their struggles with the earth, and the capability to have a source of safe drinking water. However, maternal and child survival were found to be at the core of the concept of child growth in the Haor context.^{(Reference Chakraborty29,Reference Chakraborty, Darak and Haisma32)} A range of external factors, such as poor communication between the spouses about the mother’s health condition, fear of a caesarean birth, inadequate advice by local doctors, underdeveloped infrastructure, and seasonal extremities contributed to late admissions to health care facilities. This interfered with the capability of fathers to save the lives of their wives and children. Mothers also lack the autonomy to seek health care by themselves due to insufficient access to financial resources, or to women-friendly social arrangements for health care after a caesarean delivery, limited knowledge about how to find the location of the health facility, and their gendered roles that restrict them to indoor activities. Thus, the findings of our operationalisation of CFCG generated insights regarding the capability deprivations of the parents that are relevant for child growth. Through our research, we explained how capabilities can be identified in a democratic or a participatory way by combining different methodologies from literature review and expert consultations to community-based participatory approaches.^{(Reference Chakraborty29,Reference Chakraborty, Darak and Hinke30)} At the community level, our data collection approaches focused on different interviewing techniques to draw a valid and complete list of capabilities for child growth in the context of Haor, Bangladesh. For example, in the first round of data collection, we employed a ‘doxastic interviewing technique’^{(Reference Berner-Rodoreda, Bärnighausen and Kennedy34,Reference Kvale35)} by one-way discussions in which the interviewer mainly asks questions to the participant. However, in the second round we applied an ‘epistemic interviewing technique’ using visual tools that involves a two-way interaction between participants and researchers to co-construct knowledge.^{(Reference Berner-Rodoreda, Bärnighausen and Kennedy34,Reference Brinkmann36,Reference Curato37)} The findings emphasised the importance of further assessment of the levels of achievement or deprivation for each of the capabilities and their associations with other factors to estimate the real opportunities that shaped the design of the quantitative research for the operationalisation of MICG in other contexts.

The application of the capability lens in southeastern Tanzania showed that capabilities for child growth emerge from parents’ emic understandings of what it means for a child to grow well in their context.^{(Reference Mchome38)} Drawing on ethnographic enquiry, in-depth interviews, and participatory approaches, these studies identify locally valued dimensions of growth and the contextual factors shaping them.

At the child level, parents emphasised the capability to remain free from serious illness – while accepting that ‘normal fevers’ form part of everyday childhood – as well as the capability to eat well, stay cheerful, and engage in play.^{(Reference Mchome, Bailey and Darak39)} Bodily appearance, including chubbiness (‘bonge’), skin and hair quality, and general liveliness, served as indicators of healthy growth. Parents assessed growth holistically, combining weight, height, appetite, energy levels, and developmental milestones. This multidimensional view became especially visible in their conceptualisation of height. While height was recognised as one aspect of growth, caregivers distinguished between hereditary shortness and a broader set of signs associated with stunting. A child could be tall yet considered ‘stunted’ if thinness, delayed milestones, poor hair or skin, or reduced activity were present.^{(Reference Mchome, Bailey, Darak and Haisma40)} Thus, linear growth alone did not capture the capability set associated with child wellbeing.

At the mother’s level, the capability to feed the child, maintain personal health, and secure adequate nutrition emerged as central.^{(Reference Mchome, Yousefzadeh, Bailey and Haisma41)} Mothers reported that heavy workloads, seasonal agricultural demands, and limited rest affected breastfeeding practices and the time available for childcare. Their capabilities were also shaped by constraints around decision-making power, limited access to income, and restricted autonomy in seeking healthcare. Gender norms and household structures played key roles in shaping these capability deprivations. Mothers who lacked control over food resources or support from spouses described difficulties in consistently providing the forms of care they valued.

At the father’s level, the studies highlight the capability to support maternal and child health by providing financial resources, facilitating access to medical services, and contributing to household food security.^{(Reference Mchome38,Reference Mchome, Yousefzadeh, Bailey and Haisma41)} Fathers’ behaviour – whether supportive or neglectful – acted as a conversion factor influencing mothers’ capability to feed, rest, and seek treatment for the child. When fathers struggled with unstable income, alcohol use, or divided commitments (e.g. in polygamous arrangements), maternal capabilities were reduced, limiting the opportunities available for children’s healthy growth.

At the household and community level, socio-cultural norms further shaped capabilities. The postpartum sex taboos explored by Mchome and colleagues revealed how growth is embedded within moral frameworks connecting parental behaviour with child wellbeing.^{(Reference Mchome, Bailey and Kessy42)} Adhering to postpartum abstinence was perceived as essential for protecting child growth, and deviations were believed to result in growth-related harm requiring traditional forms of treatment. These norms operated as part of the local reasoning around growth, influencing how parents interpreted risk, sought care, and responded to early signs of growth problems.

Taken together, the studies of Mchome et al. illustrate that capabilities for child growth in southeastern Tanzania are distributed across multiple actors – children, mothers, fathers – and embedded within a broader socio-cultural and environmental context.^{(Reference Mchome38–Reference Mchome, Bailey and Kessy42)} Child growth is evaluated through a composite of physical, behavioural, aesthetic, and developmental signs shaped through local knowledge systems, rather than through biomedical indicators alone. These insights highlight the value of identifying child growth capabilities through participatory and context-sensitive approaches and offer essential guidance for operationalising multidimensional child growth in capability-informed frameworks in other settings.

Extensive literature indicates which dimensions are relevant for a MICG. Nussbaum^{(Reference Nussbaum43)} identified core physical and non-physical dimensions for adult wellbeing that should be considered in the capability approach. These dimensions include physical dimensions like life, body health and body integrity, and non-physical dimensions such as senses, imagination, thought, emotions, and playful activities. In the case of children, the synergy between physical and non-physical dimensions implies, for example, that children who are unable to attend school due to being malnourished will have lower opportunities to be adequately educated. Prior work by Biggeri et al.^{(Reference Biggeri, Libanora, Mariani and Menchini44)} suggested 14 dimensions for child wellbeing based on the Convention of the Rights of the Child: life and physical health, bodily integrity and safety, love and care, leisure activities, respect, social relations, participation, mental wellbeing, education, freedom from economic and non-economic exploitation, environment (internal and external risks), religion and identity, time autonomy, and ease of mobility to education or health centres. These 14 dimensions were discussed and prioritised by children during the first ‘Children’s World Congress on Child Labor’ that was held in Florence (Italy) in May of 2004 (see Biggeri et al.^{(Reference Biggeri, Libanora, Mariani and Menchini44)}). Children from all continents participated in discussions and activities at this conference. The children were selected with a balance of girls and boys aged between 10 and 17 years, and the selection of the participants was made by the children themselves through a fair and democratic process during national and regional consultations intended to avoid any kind of discrimination. The child delegates were not expected to be considered representative of all the world’s children, but the sample is considered both selective and of high quality, because children acquired a high level of consciousness through their participation in NGO activities and their life experiences as former child labourers and activists, which helped them to better understand children’s wishes concerning how their life should progress. For the operationalisation of MICG as an index, we included the 14 dimensions of child wellbeing. Data on relevant indicators for multidimensional child growth on each of those dimensions need to be either collected through primary data collection or selected from a suitable database. As children’s opportunities and achievements depend on parental attitudes and the children’s environment, these are included as part of the MICG indicators.

During the proof-of-concept prototype, the next step in the operationalisation of MICG was the identification of a suitable database to apply our approach at population level. In order to have variables that properly capture all the multiple dimensions of child growth, we identified the YLS as a database that is sufficiently rich to capture children’s opportunities in the 14 dimensions. The participants of the YLS were a younger cohort of 2000 children per country, aged 6–18 months in 2002, and an older cohort of 1000 children per country, aged 7.5–8.5 years in 2002. The YLS (Round 2) used in this study has cross-sectional information on a cohort of children aged 5 years old living in regions of Ethiopia, Peru, Vietnam, and Andhra Pradesh and Telangana in India, including contextual information of the children’s environment and parental attitudes that affect child growth in physical and non-physical dimensions. The Round 2 survey applied several questionnaires tailored for different age groups and contexts: a questionnaire applied to young children (aged 5 years old), a household questionnaire applied to parents and caregivers, and a community questionnaire applied to stakeholders. The questionnaire applied to young children included sections on parental background, household education, livelihoods, food and non-food consumption, social capital, socio--economic status, childcare, education, health, anthropometry, and caregiver attitudes, school activities, social networks, feelings and attitudes, and future perceptions. The household and community questionnaire captured general characteristics of the households where children live, the localities, social environment, access to services, and specific child-related modules on education and health services. The setting of the YLS was based on locating sentinel sites where surveillance systems were implemented to collect data about the development of children in Ethiopia, India (Andhra Pradesh), Peru, and Vietnam. The sentinel system was designed to gather high-quality, representative data over time, ensuring the sustainability and accuracy of the longitudinal study. Sentinel sites were chosen to ensure a balanced representation of regional diversity and rural/urban differences in heterogeneous agroclimatic areas, with a pro-poor bias aimed to reflect the plurality of socio-economic status. The aim was to analyse how the living standards and characteristics of the Young Lives children compare with other children, ensuring that different levels of living standards are represented within the dataset. The sampling methodology was based on a multi-stage, purposive sampling to select cohorts of children. Purposive sampling is a selective, subjective, non-probability sampling technique where locations are selected based on judgements about which elements will be the most useful for the goals of the study. This method is useful to gain specific insights and particular characteristics of a phenomenon that is not evenly distributed across the population. This sampling method is more useful when in-depth understanding is more considered valuable than generalisability.

Table 1 below shows the 29 indicators used to measure the 14 dimensions of the MICG and provides the complete list of cut-offs that define a child as deprived for each indicator. To build these indicators, relevant questions from the Young Lives Survey (YLS) were first identified and matched to each dimension of the MICG. For instance, for the dimension of life and physical health, questions on anthropometry (height, weight, and age of the child) were used to derive stunting, wasting, and overweight indicators, while questions on dietary recall and food group consumption informed the indicators of nutrition frequency and diversity. Similarly, household survey items on shelter materials, access to education, time spent on domestic chores, and parental attitudes towards children were mapped to the respective dimensions of bodily integrity, access to education, time autonomy, and love and care. The non-physical indicators of MICG account for the attitude of parents towards caregiving, cognitive and verbal development of children, domestic tasks in the household, and the internal and external risks that affect a child’s safety, among other indicators.

Table 1.

Indicators included in the Multidimensional Index of Child Growth (MICG) based on the availability of information from the Young Lives Survey (Round 2)

Table 1 Long description

Panel A: A table listing indicators and deprivation cutoffs for measuring child well-being across various dimensions. The table includes columns for Capability, Indicator, and Deprivations cutoff for the indicators. Each row details specific indicators such as Overweight, Stunting, Wasting, Nutrition (frequency and diversity), Vaccination, Health, Safety in the street, Shelter, Danger, Love of parents, Proud of children, Parents’ responsibility, Parents’ fulfillment, Leisure, Respect, Friendship, Cooperation, Cognitive, Verbal, Access to education, Imagination, Paid/unpaid work, External risks, Internal risks, Children’s religion, Children’s ethnicity, Hours of domestic tasks, and Time to get to school. Each indicator has a corresponding deprivation cutoff that defines a child as deprived.

(*) The cut-offs were chosen based on international standards (e.g. 2-standard deviations) and informative discussions with experts in nutrition and non-physical child growth.

Cut-offs used to identify deprived children were calculated based on international standards⁽²⁵⁾ and informative discussions with experts in nutrition and non-physical child growth. For example, a child is considered stunted if his or her height-for-age z-score is less than –2 SD, wasted if the weight-for-length/height z-score is less than –2 SD, and deprived of education if the household does not have access to schooling. Likewise, a child is classified as deprived of safety if parents report that it is not safe for the child to go to school or play on the street. These cut-offs combine international norms with context-specific questions from the YLS, ensuring that both physical and non-physical aspects of child growth are captured in a multidimensional framework.

After identifying and coding the 29 indicators, these were aggregated into the MICG following a counting approach inspired by the Alkire-Foster methodology.^{(Reference Alkire and Foster45)} Each indicator was first coded as deprived (1) or non-deprived (0) using the cut-offs described in Table 1. Indicators were then grouped under their corresponding 14 dimensions, and within each dimension, indicators were equally weighted so that the dimension carried the same importance regardless of the number of indicators it contained. Subsequently, each dimension was assigned an equal weight in the overall MICG, reflecting the multidimensional nature of child growth and ensuring that non-physical aspects such as love and care or social participation contributed on par with traditional anthropometric measures. A child was classified as multidimensionally deprived if their weighted deprivation score crossed a specified cut-off (e.g. one-third of the total weighted deprivations), consistent with international practice in multidimensional poverty and child wellbeing research. This procedure allowed us to move beyond a narrow reliance on physical growth indicators and instead capture the multiple, intersecting deprivations that shape children’s development.

After identifying all deprivations, a deprivation score is calculated as the weighted sum of indicators. The MICG is then expressed as one minus this weighted deprivation score, representing the child’s overall achievements. The weights for each indicator can be determined either endogenously or exogenously. Endogenous weights are derived statistically from the covariance structure of the data, while exogenous weights are based on normative judgements regarding the relative importance of each dimension.

A comparative analysis was performed to evaluate how sensitive the results are to the use of endogenous weights versus exogenous weights. Endogenous weights were obtained with binary principal component analysis and exogenous weights were chosen by using two weighting schemes: equal weights, and weights based on the preferences of children, since when children were asked what the most important opportunities for a child during her/his life are, they prioritised education and love and care over physical health. Figure 2a shows the kernel density of the MICG of Ethiopia obtained with the 3 different types of weights. Endogenous weights calculated with data produced a more dispersed index of child growth compared to exogenous weights.

Figure 2.

(a) Kernel density of the MICG for Ethiopia with different weighting schemes. (b) Correlation matrix between the MICG estimates for Ethiopia with different weights.

Figure 2b shows the correlation between the MICG estimates. The MICG calculated with endogenous weights is strongly correlated with the MICG calculated with exogenous weights chosen through a children’s participatory process (Spearman’s rank correlation = 0.91, P < 0.0001), but the highest correlation (Spearman’s rank correlation = 0.97, P < 0.0001) is obtained for a MICG calculated with equal weights and a MICG calculated with exogenous weights chosen by children. In consequence, applying equal weights is a robust strategy without a strong impact in the calculation of the aggregated index of child growth.

Additionally, since weights derived from statistical methods have been criticised for violating key properties of wellbeing indices such as monotonicity and subgroup consistency,^{(Reference Dutta, Nogales and Yalonetzky46)} we assign equal (non-statistical) weights to each dimension in the results section. This choice is further justified by the fact that similar results were obtained when comparing endogenous weights derived from the data and exogenous weights informed by children’s perspectives. Because the 14 dimensions of the MICG are aligned with the United Nations Convention on the Rights of the Child, equal weighting reflects the principle that all children’s capabilities and rights are equally important. Consequently, dimensions with more indicators distribute their weight across several measures, while dimensions with fewer indicators allocate a larger share of weight to each indicator.

Figure 3 shows the spiderweb chart we designed to visualise the 14 dimensions of Multidimensional Child Growth (MICG). This framework, grounded in the work of Biggeri et al.,^{(Reference Biggeri, Libanora, Mariani and Menchini44)} is informed by the Capability Approach and aligns closely with the Convention on the Rights of the Child (CRC). Additionally, the MICG framework aligns with international policy agendas such as the UN Sustainable Development Goals (SDGs), particularly SDG 3 (Good Health and Wellbeing), SDG 4 (Quality Education), SDG 10 (Reduced Inequalities), and SDG 16 (Peace, Justice, and Strong Institutions). It provides an actionable bridge between high-level human rights commitments and concrete, measurable outcomes that reflect child growth and the context in which a child thrives. We argue that this visualisation diagram can be used for monitoring multidimensional child growth at national level and individual level, in a similar way that WHO growth charts are used for monitoring the weight and length/height growth of children. The axes of the chart of multidimensional growth measure the percentage of achievements in each of the 14 dimensions of child growth, based on the indicators described in Table 1.

Figure 3.

(a) Chart of anthropometric child growth from the WHO standards for monitoring physical child growth*. (*) The source of Figure 3a is the WHO child growth standards (see WHO, 2006, Figure 57, p. 124, weight-for-age percentiles for girls from birth to 60 months). (b) Multidimensional child-growth chart based on physical and non-physical dimensions.

Figure 3 Long description

Panel A: A line graph displays weight in kilograms on the vertical axis and age in months on the horizontal axis. It includes multiple percentiles: 3rd, 15th, 50th, 85th, and 97th, each represented by differently colored lines. The lines show an upward trend, indicating that weight increases with age. Panel B: A radar chart illustrates multidimensional child growth, with axes representing various dimensions such as life and physical health, mobility, time autonomy, religion and identity, environment, economic freedom, education, mental wellbeing, participation, social relations, respect, leisure activities, love and care, bodily integrity and safety. Each axis is marked with percentage values ranging from 0 to 100 percent. The chart shows a balanced distribution across these dimensions.

Each axis in the spiderweb corresponds to a core dimension of child growth: Life and Physical Health, Mobility, Time Autonomy, Religion and Identity, Environment, Economic Freedom, Education, Mental Wellbeing, Participation, Social Relations, Respect, Leisure Activities, Love and Care, and Bodily Integrity and Safety. These dimensions reflect not only the instrumental conditions for physical and non-physical child growth but also intrinsic context-dependent environmental determinants of child growth. The values plotted along each axis are scaled from 0 to 100%, representing the degree of achievement or access in each domain, thereby enabling a comparative, profile-based interpretation.

From an academic standpoint, this visualisation addresses long-standing challenges in child development research, particularly the difficulty of integrating qualitative, experiential dimensions with measurable, policy-relevant indicators. The MICG framework moves beyond reductionist monitoring and evaluation of child growth based on anthropometrics and foregrounds children’s agency, subjectivity, and contextual embeddedness. It reflects an ontological shift from viewing children as passive recipients of goods and services to recognising them as active social agents with their own evolving capabilities.

The spiderweb visualisation also serves as a heuristic for identifying synergistic or compensatory relationships among dimensions. For instance, limitations in ‘Economic Freedom’ may co-exist with high ‘Love and Care’ or ‘Social Relations’, which may buffer some of the adverse effects. Conversely, deficiencies in ‘Respect’ or ‘Bodily Integrity and Safety’ may have cross-cutting implications for other dimensions, such as mental health or participation. Thus, the framework invites complex systems thinking, highlighting feedback loops and threshold effects that traditional linear models may overlook. Moreover, the inclusion of dimensions such as ‘Religion and Identity’ and ‘Time Autonomy’ expands the epistemic boundaries of dominant development paradigms by incorporating cultural, emotional, and temporal freedoms, which are often marginalised in both academic metrics and global indices. Contextualisation is needed to properly create questions and instruments that reflect and capture in a similar way the different concepts (such as love and care) in different contexts. Contextualisation is also crucial as the structural flaws could seriously limit parents and caregivers’ ability to fulfil children’s safety or dignity. For example, human-made famine or war are struggles that are often beyond parents’ ability to address and protect their children’s safety, dignity or nutritional needs.

For policymakers working on health and nutrition interventions, the spiderweb offers a strategic tool for diagnostic assessment, resource allocation, and monitoring. It enables stakeholders to visualise areas of underperformance and strength across a range of wellbeing domains and to identify where integrated interventions might yield the greatest impact, across multiple micro-, meso- and macro-levels. For example, persistent deficits in ‘Mobility’ or ‘Time Autonomy’ may suggest the need for child-friendly urban planning or more flexible schooling schedules, whereas gaps in ‘Mental Wellbeing’ might point to the urgency of scaling up psychosocial support services in schools and communities. It provides an integrative lens that bridges academic inquiry and policy implementation, empowering stakeholders to pursue holistic, context-sensitive, and justice-oriented interventions in support of children’s flourishing.

Finally, a methodological concern when assessing multidimensional child growth is the focus on realised and observable outcomes – such as being vaccinated – rather than on the underlying potentialities that are central to the concept of capability. This limitation makes it difficult to capture the full range of freedoms and opportunities that children may or may not have access to, since the MICG measures the achievements of children with observed indicators, with achievements being equal to the observed situation of children.

To address the previous limitation, and with the aim of properly measuring the opportunities available to children – that is, their potential child growth above or below their achievements –, Bayesian stochastic frontier analysis was applied to the MICG based on the approach of Henderson and Follett.^{(Reference Henderson and Follett47)} The Bayesian model estimates potential alternative values of multidimensional child growth that could be negatively or positively affected by children’s observed characteristics. We chose to use geographical location and sex as discriminating characteristics, besides random circumstances that affect the achievements measured by the MICG. Specific children that could be left behind are those with the lowest opportunities for physical and non-physical growth; that is, those children that have the lowest estimated values of the potential MICG.

Figure 4 below illustrates the Bayesian extension of the Multidimensional Index of Child Growth (MICG), transitioning the focus from achievements (outcomes) to opportunities. The horizontal axis (MICG) represents the achievements of children based on their observed physical and non-physical conditions. These are direct measurements (e.g. health, cognitive scores, safety perception). Each curve in the vertical axes represents an individual child (in this case 10 children), and the probability density functions reflect their estimated opportunity of child growth, that is, the probability of achieving a specific outcome. Densities closer to a MICG equal to one reflects higher estimated opportunities for multidimensional growth, while density curves close to a MICG equal to zero indicate lower opportunities to achieve desired outcomes. These opportunities (i.e. capabilities) are derived using Bayesian stochastic frontier analysis.

Figure 4.

Bayesian extension of the Multidimensional Index of Child Growth.

The grey dot on each curve marks the actual multidimensional achievement of each child. It’s where they currently stand in terms of growth. Compared to observed achievements, the Bayesian estimation of opportunities provides a counterfactual frontier that reflects the level of multidimensional growth each child could theoretically achieve under different circumstances. Based on the capability framework, the core idea of applying a Bayesian extension is that while two children might have the same achievements, their latent opportunities – their potential to grow physically, emotionally, and cognitively – may differ drastically. The Bayesian extension of the MICG thus allows identifying children that could be “left behind” not just in actual outcomes, but in unrealised potential.