Child Development in Low- and Middle-Income Countries

The data reported here come from the United Nations Children’s Fund (UNICEF) MICS4 and MICS5, conducted between 2009 and 2017. Participant selection and data collection were conducted via standardized procedures established by UNICEF, codified in MICS manuals,^8,9  and conveyed to enumerators in each nation during regional workshops. Each country selected a nationally representative probability sample by dividing itself into proportionately populated regions and randomly selecting participants within each region, as detailed in MICS manuals.^8,9  In each nation, pairs of UNICEF-trained field-workers interviewed household members at their homes, using standardized MICS4 and MICS5 questionnaires and recording answers on portable computer devices. Data were then double entered, with built-in internal consistency checks to ensure accuracy, and subsequently archived centrally. A UNICEF MICS team oversaw data collection and aggregation and was available to consult with national governments at any time. Global MICS4 and MICS5 evaluations confirmed that the entire MICS procedure was of high quality and captured accurate data.^8,9 

Data were collected in 51 LMICs on 159 959 children between 36 and 59 months of age (mean = 47.3 months, SD = 6.9). In Table 1, we name the countries and gives their sample sizes and locations. The total sample was evenly divided, with 50.7% of children being boys and 49.3% girls. Respondents to the MICS4 and MICS5 identified the person who served in the role of the child’s primary caregiver, called here “mother.” Mothers were almost always the child’s biological mother but might have included some adoptive mothers, stepmothers, aunts, grandmothers, and foster mothers. Mothers were on average 32.0 years old (SD = 8.9). Mothers’ educational achievements varied: 27.9% reported receiving no formal education, 28.4% reported receiving primary education, 32.4% reported receiving secondary education, and 11.3% reported receiving higher education.

LMICs¹⁰  were defined with reference to the World Bank¹¹  system of classification of economies, which is based on gross national incomes per capita, quality of life (life expectancy, literacy rates), and economic diversification (labor force, consumption). Although 70 LMICs conducted the MICS4 and MICS5, data from 51 LMICs were used here. Nineteen countries were excluded because they did not include questions that contained data pertinent to these analyses. HDI scores for excluded countries did not significantly differ from those for the 51 included countries. The 51 LMICs represent 17 countries from Africa, 12 countries from Asia, 9 countries from North America, 8 countries from Europe, and 5 countries from South America.

The 51 LMICs represent a range of human development conditions, as measured by the UN HDI,¹²  a 0 (low) to 1 (high) composite indicator of a country’s status on 3 dimensions. The composite includes indicators of health (captured by the Life Expectancy Index [LEI]), education (captured by the Education Index [EDI]), and income (captured by the Income Index [INI]). Table 1 gives each country’s HDI score. HDI scores ≤0.550 indicate low human development, scores 0.550 to 0.699 indicate medium human development, scores 0.700 to 0.799 indicate high human development, and scores ≥0.800 indicate very high human development. Of the 51 LMICs, 14 fell in the low category, 14 in the medium category, 20 in the high category, and 3 in the very high category on the 2013 HDI.¹² 

Both MICS4 and MICS5 contained several different questionnaires.^8,9  The Questionnaire for Children Under Five is composed of sets of standardized questions about all children <5 years of age in a household.¹³  Ten questions are on whether target milestones of child development have been observed. These 10 questions, collectively called the ECDI,¹⁴  were developed by UNICEF specifically to assess development in children aged 36 to 59 months in 4 domains plus a total score.¹⁴  The 4 domains of child development include literacy and numeracy development (to identify, know, and use numbers, letters, and words), socioemotional development (to recognize and appropriately express thoughts and feelings and interact well with others), physical health development (to physically grow, avoid rampant sickness or disease, and develop gross and fine motor abilities), and approaches to learning (to follow directions to achieve goals and complete tasks independently). The overall child development score is the combination of child literacy and numeracy development, socioemotional development, physical health development, and approaches to learning.¹⁴  These 5 measures of child development have been identified as critical to evaluating child health and well-being and ensuring that the world achieves the UN Millennium Development Goals, World Fit for Children Declaration and Plan of Action goals, and Sustainable Development Goals.^9,10,14  Furthermore, UNICEF also designates a child as being “on track” when the child passes 2 of 3 items in literacy and numeracy development and in socioemotional development and 1 of 2 items in physical health development and in approaches to learning; children are on track in their overall development if they are on track in at least 3 of the 4 developmental domains (for details see the Supplemental Information).

Three analyses are reported below. First, we examined correlations of the HDI and subindices with the ECDI and subindices. Second, we examine partial correlations that capture the unique association between an HDI subindex and child development scores, while controlling for the effects of other HDI subindices. Third, we examine regression models of the HDI score as a predictor of ECDI scores, after controlling for numerous child, maternal, household, and country demographic covariates.

The HDI and its constituent indices were strongly and positively associated with the ECDI and its constituent indices (Table 2). All associations exceed a statistical cutoff, indicating effects of practicable significance (r = 0.10), and many associations fall in the medium effect-size (r = 0.30) range.¹⁵  Notably, the overall HDI score revealed a strong positive association (r = 0.40) with the ECDI score: the HDI score explained 16% of variance in children’s ECDI scores and was the most influential predictor of ECDI scores examined. In other words, countries that create more optimal conditions for human development in terms of increasing life expectancy, access to education, and income have children who are more optimally developed in terms of their literacy and numeracy, socioemotional development, physical health, and approaches to learning. The linear association between the HDI and ECDI is graphed in Fig 1: countries with higher HDI scores have children who score higher on the ECDI. Figure 1 also reveals that this trend holds in the different UN-designated groupings of countries on the HDI. Countries that the UN categorizes in the low human development group are clustered in the lower left of the graph (as circles), where both human development scores and child development scores are low. Countries categorized by the UN as being in the high human development group are in the upper right of the graph (as squares), where both human development scores and child development scores are high. Furthermore, only 5 in 10 (54.6%) children in countries from the low human development group met UN criteria for being on track in their development,¹⁴  whereas 7 in 10 (73.0%) children in countries with medium human development did, >8 in 10 (84.5%) children in countries with high human development did, and nearly 9 in 10 (89.8%) children in countries with very high human development did.

Higher human development scores are related to higher child development scores. Moreover, each component of the HDI supports child development beyond what other HDI components afford. Table 2 reveals associations between the HDI LEI, HDI EDI, and HDI INI and child development indices after we controlled for associations between the other 2 HDI subindices and child development. So, for instance, the partial correlation in the upper left (r_partial = 0.11) indicates the unique correlation between the LEI and the ECDI after the correlations between the EDI and INI with the ECDI are taken into account. As can be seen, all 3 indices are significantly and uniquely associated with both overall child development and the 4 subindices of child development. Nations creating better conditions for their populations’ longer life expectancy, higher education, and greater wealth are each uniquely associated with overall child development in the population as well as the nation’s children’s development of literacy and numeracy, socioemotional health, physical health, and approaches to learning.

Each component of the HDI uniquely supports children’s development, but a country’s educational index score matters the most for overall child development, literacy and numeracy development, and physical health development. The unique EDI associations with these 3 child development scales are highest: higher scores on the EDI are associated with greater overall child development, literacy and numeracy development, and physical health development. For socioemotional development and approaches to learning, a nation’s LEI matters the most: countries that create conditions for long life have children with greater socioemotional development and who develop better approaches to learning.

These analyses were focused on associations between country-level indices, namely, the HDI and its constituents and the ECDI and its constituents. Additional analyses reveal that child and maternal sociodemographic characteristics and household and country demographic characteristics also correlate with child development indices. In Supplemental Table 3, we give results in which all sociodemographic and demographic characteristics are simultaneously used in regression models to predict child development scores. Even after accounting for these several covariates, the HDI still emerges as the most powerful predictor of the ECDI. These results are described in greater detail in Simultaneous Associations Between All Demographic Characteristics and ECDIs in a Linear Regression Framework in the Supplemental Information.

In summary, the higher the national status of a country on the HDI (and especially the EDI) the greater are children’s literacy and numeracy development, socioemotional development, physical health development, and approaches to learning. These relations were strong and linear and held even when we considered multiple aspects of children, mothers, and households as covariates. No family demographic variable was a stronger predictor of child development than national HDI scores.

In this investigation, we examined human development at the county level and child development scores in almost 160 000 families in 51 LMICs. We found that country HDI scores are strongly and consistently associated with many ECDIs. In both the correlation analyses and the regression analyses (in the Supplemental Information), HDI scores were the strongest and most robust predictors of overall ECDI scores. Nations with higher human development scores had higher levels of child development (Fig 1). All 3 subindices that compose a country’s HDI score (life expectancy, education, and wealth) uniquely contributed to all 4 domains of early child development; however, of the 3 subindices of the HDI, education emerged as the strongest unique predictor of overall early child development scores, child literacy and numeracy scores, and child physical health scores. Additionally, life expectancy was the strongest predictor of child socioemotional development and approaches to learning scores.

Similar analyses using MICS4 data from 97 731 children from 35 LMICs found that mean child development scale scores were higher in countries with high HDI scores compared with countries with low HDI scores.¹⁶  In the present analyses, we examine almost 160 000 children from 51 LMICs with MICS4 and MICS5 data and so replicate and build on findings from previous research by demonstrating that HDI scores uniquely predict child development and do so even after we control for numerous other child and maternal sociodemographic covariates and household demographic covariates (see the Supplemental Information). In previous work,¹⁶  researchers investigated associations between the overall HDI score and overall child development. In the present analyses, we extend that work by reporting associations of specific HDI indices (ie, life expectancy, education, and income) with specific categories of the ECDI (ie, literacy and numeracy, socioemotional development, physical health, and approaches to learning).

Countries with higher HDI scores tend to invest more in sanitation, welfare, housing, and educational and nutritional systems that promote healthy child development. Such nations are also likely to have more highly developed policy infrastructures that promote sensitive caregiving through parental access to educational materials, greater dissemination of parent knowledge about adaptive parenting practices, and promotion of parents' time and effort investment in their children. In subsequent work, researchers can further explore how high HDI scores may manifest their effects in child development through parental caregiving. For now, these findings indicate that no demographic variable appears to be a better or stronger predictor of early child development than national development index scores.

As powerful as HDI scores (and child and maternal factors; see the Supplemental Information) are in predicting child development scores, heterogeneity within countries and factors not captured with precision by the HDI may also play roles in child development. In essence, important factors that promote child development worldwide remain unstudied. Specifically, the factors identified in Supplemental Table 3 account for an impressive, nearly large effect size but still explain only one-quarter of the variance in child development scores. More than three-quarters of the variance in overall child development scores was not explained. It is likely that direct proximal caregiving explains more variance in child development scores across the majority of the world.^16–19 

The MICS provides the fields of parenting and developmental science as well as policy and intervention with unique and bountiful data. However, its limitations need to be acknowledged vis-à-vis its strengths. A distinct advantage of MICS data is that an impressive number of inhabitants in a substantial number of underresearched LMICs are represented. Therefore, MICS data shine much-needed light on early child development in areas of the world underrepresented in current theory, research, and policy.^13,20,21  Yet MICS sample sizes vary considerably across countries, not all countries provide all data (even with 51 LMICs represented, the data set falls far short of worldwide representation), and comparable data from high-income countries are missing. Because of the large sample size, appropriately conservative effect-size approaches were adopted here to interpret the statistical results. An additional strength of the MICS is its nationally representative sampling and data collection by well-trained national civil workers, demographers, and enumerators.^8,21  The unit of analysis used across MICS surveys is country,^8,9  and the sheer number of countries in this rich data set precludes reference to individual national and cultural literatures. For this reason, we appealed to the HDI as a cross-national organizing and comparative construct. Consequently, these analyses are focused on between-country, not within-country, patterns, which of course can be expected to vary considerably both within and across samples.^22,23 

Mothers reported on MICS measures used in these analyses, so perspectives from other family members, including fathers, are not included. Given that mothers, fathers, and other caregivers, as well as children, each may have different perspectives on child development that likely generate disparate reports on children,^24–30  single-reporter bias is a limitation of the current approach and analyses. MICS indicators are also cross-sectional (precluding causal interpretations) and subject to historical time as well as seasonality effects. These limitations on MICS data moderate their explanatory potential.¹³ 

Despite these limitations, MICS data provide an unprecedented opportunity to examine how the development of a country in terms of health, education, and wealth shapes multiple different aspects of child development on a global scale. Following UNICEF’s aspirations, policy makers may use MICS data to create a better world for children by reducing vulnerabilities and building on strengths in national income, life expectancy, and, especially, education.^12,21 

ECDI

Early Childhood Development Index

EDI

Education Index

HDI

Human Development Index

INI

Income Index

LEI

Life Expectancy Index

LMICs

low- and middle-income countries

MICS

Multiple Indicator Cluster Surveys

UN

United Nations

UNICEF

United Nations Children’s Fund