. Author manuscript; available in PMC: 2021 Sep 14.

Published in final edited form as: Am Econ Rev. 2020 Aug;110(8):2454–2484. doi: 10.1257/aer.20170861

Table 3—

Network Statistics Regressed on Treatment

	Percent supported (data) (1)	Percent supported (estimate) (2)	Graph-level proximity (estimate) (3)
Treatment neighborhood	−0.0655 (0.0319)	−0.0901 (0.0541)	−0.0500 (0.0164)
Constant	0.4427 (0.0628)	0.4364 (0.093)	0.3642 (0.0108)
Mean of the response variable	0.3880	0.3125	0.3375
Observations	3,514	3,598	62

Notes: Standard deviation of village-level block bootstrap in parentheses. Sample includes neighborhoods with estimated sampling rate ≥ 20 percent. For large number of excluded low sampling rate neighborhoods, the population count is top-coded at 500 households. For these very large neighborhoods, we calculate the sampling rate using a population of 500. The outcome variable of columns 1 and 2 is the share of links that are supported and in column 3 it is the average proximity in the graph.