Table 1.

Study design and data collection including purpose of each dataset.

Method	Areas (number of villages)	Sampling design	Data collection period	Interval between village-level campaign and coverage survey	Purpose
Post-vaccination transects	Serengeti (85)	1 transect in every sub-village (357 total) in all villages	May–October 2015	2–3 h	Coverage estimates at village and district level. Data used for simulations to explore how the number of transects/village affect precision of district-level estimates
	Southeast Tanzania and Pemba (2,070)	1 transect in 2 sub-villages (4,140 total) in every village/district	November 2014–January 2015	2–3 h	Setup and implementation costs
School-based surveys	Serengeti (4)	100 pupils/school in 4 schools/district (333 pupils)	July 2015	1 month	Coverage estimates at district level. Precision of estimates compared with census data and simulation experiments.
	Southeast Tanzania and Pemba (115)	100 pupils/school in 6 schools/district (8,254 pupils)	November 2014 and February 2015	1–2 months	Setup and implementation costs
Household survey	Southeast Tanzania and Pemba (160)	30 households/village in 6 villages/district (4,488 households)	July–August 2011	2–6 months	Setup and implementation costs. Data used to parameterize simulations for settings with high: human dog ratios to explore precision of household surveys
Complete human and dog census	Serengeti (88)	All households in district (35,867)	From 2008 to 2015	Vaccination campaigns ~May–July each year. Census at different times of year for each village	Census does not provide a point estimate of coverage relative to a specific campaign. Data used for simulation experiment to determine how sampling (e.g., household and school-based surveys) affects precision of coverage estimates