Calculating racial disparities in child poverty
This week, the American Enterprise Institute published a blog that raises an important question regarding the methods used to estimate gaps by race in child poverty—including the method Child Trends used in our 2019 analysis of early childhood poverty rates in 2017 and 2018. Our analysis found that the gaps between non-Hispanic White, Black, and Hispanic children under age 6 widened from 2017 to 2018, despite overall declines in the poverty rate for each racial/ethnic group. Conversely, a scholar at AEI found that the gap between Black and White children had narrowed.
Interestingly, different analytic methods can tell different stories about racial and ethnic disparities. While the AEI blog post reports that the gap between the rate of Black and White children in poverty decreased, our analysis found that the gap between the ratio of Black and White children in poverty increased.
What explains the difference between these analyses? It’s all about whether you subtract or divide.
You can calculate gaps between two percentages in two equally valid ways. The first method is to simply subtract one percentage from the other, resulting in the percentage point difference.
Alternatively, you can divide one percentage by the other, resulting in a ratio of the percent of children living in poverty. This is the method we used to determine that the gap between non-Hispanic White and Black children in poverty has widened: Compared to non-Hispanic White children, there were relatively more Black children living in poverty in 2018 than in 2017.
In our analysis, we presented figures that displayed the poverty rates among young children from birth through age 5 in 2017 and 2018 by race and ethnic group. The figures indicate that, in 2017, there was a percentage point difference of 23.6 between non-Hispanic White and Black children, and that, in 2018, there was a percentage point difference of 23.3. If one were to characterize the poverty gap as the percentage point difference, then one would conclude that the poverty gap decreased by 0.3 percentage points. However, we also examined the ratio of non-Hispanic White children in poverty to Black children in poverty. Using this method, we concluded that, for every non-Hispanic White child in poverty in 2017, 3.21 Black children were in poverty; in 2018, for every non-Hispanic White child in poverty, 3.56 Black children were in poverty.
We interpret ‘the poverty gap’ by this latter metric because it is a meaningful and interpretable metric of the magnitude of proportional difference between the two groups. We acknowledge that as poverty rates decrease (and approach 0), the magnitude of proportional differences between racial and ethnic groups, expressed as a ratio, can grow larger; this is a feature of using ratios versus percentage point differences. But if the goal is to reduce child poverty to zero, or to approach equality in early childhood experiences, then both the overall poverty rates and the proportional differences in the rates at which young children are exposed to poverty matter. When one group of children has a substantively greater likelihood of being exposed to poverty, policymakers should be concerned, even when findings suggest overall reductions in numbers and percentages of child poverty.
Whether one looks at ratios and concludes that the Black-White gap has increased slightly or looks instead at rates—and concludes that it narrowed slightly—substantial gaps nevertheless remain between Black and Hispanic children in poverty and White children in poverty. We must close these gaps and work to end poverty for all children.