Table 1.

Relationship between soil zinc availability and health outcomes among children and adult women.

	Stunting	Underweight	Wasting	Women’s height
	(1)	(2)	(3)	(4)
District-level soil Zn availability (% Satisfactory)	$-0.445\ast \ast \ast$	$-0.480\ast \ast \ast$	− 0.132	$0.012\ast \ast \ast$
	(0.125)	(0.147)	(0.105)	(0.003)
Individual controls	Yes	Yes	Yes	Yes
Region fixed effects	Yes	Yes	Yes	Yes
N	226,195	226,195	226,195	664,849
${\text{R}}^2$	0.02	0.03	0.01	0.02

Each cell in the table shows estimates from a separate OLS regression using Eq. (1) (see "Methods" Section). All specifications include controls for individual characteristics (mother’s religion, ethnicity, age at childbirth, age at start of first marriage, child’s gender, birth order, and month of birth, whether the child lives in an urban or rural location in columns (1–3); age, religion, ethnicity, urban/rural, and month of birth for column (4)). Outcome variables in columns (1–3) are children’s anthropometric measures defined as binary variables taking value 1000 if a given child was stunted, wasted, or underweight, and 0 otherwise. The outcome in Column (4) is women’s height measured in centimeters (cm). The coefficient on district-level soil Zn shows the change in prevalence of the outcome (per ’000 children) in columns (1–3) and the change in height (cm) (column 4) for a 1% increase in the number of satisfactory soil Zn tests in a district. Values in parentheses show the standard errors clustered at the district level. Coefficient significance at 1%, 5% and 10% are indicated by $\ast \ast \ast$, $\ast \ast$ and $\ast$, respectively.