Table 3.

Equations of the relationships between covariates and prevalence that are fitted in the simulation study under the true model (${\mathcal{M}}_i$) and the misspecified model (${\mathcal{M}}_F$). To keep the mathematical notation simple, each covariate is denoted identically as d(x) and regression coefficients are omitted in each equation. Note that d(x) + max{d(x) − c, 0} denotes a linear spline with a single change point at d(x) = c.

covariate	${\mathcal{M}}_I$	${\mathcal{M}}_F$
temperature	d(x)	d(x)
EVI	d(x) + max{d(x) − 0.35, 0}	d(x) + max{d(x) − 0.35, 0}
NTL	d(x) + max{d(x) − 9, 0}	d(x)
log-population	d(x)	d(x)
log-precipitation	d(x) + max{d(x) − 6.85, 0}	d(x)