Introduction

This empirical study reviews the impact of differing levels of maternal education on child health outcomes. The study uses data on infant mortality, children's height-for-age, and immunisation status from the first round of Demographic and Health Surveys (DHS) in 22 developing countries. The authors argue that while a correlation exists between maternal education and these factors, there is not a direct causal relationship, and that the correlation is in part false because education levels are actually stand-ins for socioeconomic status and community of residence. The authors question the effectiveness of programmes that seek to improve child health by improving maternal education (this does not, however, suggest that improved female education is without merit on its own).

Key Points

The authors note that since Caldwell (1979), it has generally been accepted that a correlation exists between mothers' education and child health and, therefore, investments in women's education are essential to lowering infant and child mortality and improving health. The authors suggest that skeptics of this theory have been largely ignored, in part because of a lack of comprehensive statistical studies. They set out to disprove this theory by comparing maternal education against several key variables and subsequently introducing a series of control variables that weaken the relationships in several instances. When introducing the dependent variables, the authors also make an important distinction between actual health outcomes, in the form of infant mortality and height-for-age (anthropometric), and parental health-seeking behaviours, in the form of immunisation rates.

The authors note that most studies based on the full DHS or World Fertility Surveys (WFS) have shown relatively strong correlations between maternal education and child health and survival. It has previously been argued that this results from one of two potential paths: Either education enhances the use of modern health services, and/or education results in a range of favourable behaviours related to child care. Yet several studies have shown that child mortality has declined over time equally across all educational groups within certain countries, and, further, that some reductions have occurred despite stationary maternal educational levels.

The authors introduce several complex statistical models, all of which employ the three previously mentioned dependent variables. In all cases the mother's education is the main independent variable, with varying controls that include urban residency, father's education, whether or not the mother has a partner, and household variables such as the presence of piped water and toilets.

The authors used a regression model that divided residency into one of two groups, a low-education cluster and a high-education cluster (based on DHS community-level data). They found that in almost all countries mothers with secondary educations failed to benefit from their higher educations if they lived in low-education clusters; conversely, women (and their children) with limited education benefited from being in high education clusters.

The major regressions using the control variables also revealed that education was not as strong of an explanatory feature once other factors were controlled for. Assessments of infant mortality revealed that maternal educational differences were substantially reduced when other socioeconomic variables were introduced. Depending on the model employed, education was found to be significant in only 14 to 7 of the 22 countries studied, and the remaining statistically significant correlations were in almost all cases substantially weakened. Height-for-age data revealed a similar phenomenon: In the original non-controlled model it was shown to be correlated with maternal education, but after adding other socioeconomic factors the incidence of significance is also weakened. In the third model the strength of the relationships were only one-third as large as those in the original model.

Immunisation status has also been regularly correlated with maternal education, and the authors sought to evaluate this variable because it provides insight into health-seeking behaviour rather than child health outcomes. The authors used an index that included three doses of polio vaccine, three doses of DPT (diptheria-pertussis-tetanus), a measles vaccination, and Bacillus Calmette-Guerin (BCG). What the authors found was that maternal education remains statistically significant even after controlling for other factors in the 2nd and 3rd models. Declines in coefficients (the strength of the relationship) across the models are relatively small. The authors note that higher vaccination levels do not necessarily translate into improvements in mortality rates or anthropometric indices.

The authors conclude that maternal education, while important in effecting the health seeking behaviour of mothers, does not have a substantial effect on actual health outcomes. They do, however, note several limitations, one of which is that, while small, the effect of maternal education remained significant in a handful of countries (especially in Latin America) and coefficients consistently moved in the expected direction. Furthermore, they note that this study highlights two possible alternative routes through which maternal education contributes to child health: first, that education may effect access to health facilities at the community level (as higher educated populaces tend to come together and draw in better services) which also benefits those lower educated mothers who live in the area; and secondly, that the higher immunisation rates among children of better-educated mothers will reduce disease incidence.

The conclusion is that, while investment in women's education is vitally important to development, it cannot override various other socioeconomic and structural hamstrings that may prevent even educated mothers from affording their children the best levels of childcare. The authors note that a lack of critical attention to this fact can result in unrealistic expectations of mothers and increases in their already substantial burden levels. Thus, maternal education projects must be matched with comprehensive service provision and other forms of health development.