Table 5.

Determinants of treatment rates by community medicine distributors

Variablea	Coef.	Clustered robust SE	p value	95% confidence interval
In-group biasb	0.139	0.059	0.025	0.019	0.259
Selected by local councilc	− 0.085	0.053	0.121	− 0.194	0.024
Friends help with MDAd	0.084	0.040	0.041	0.004	0.165
Female	− 0.120	0.057	0.043	− 0.235	− 0.004
Fisherman/fishmongere	0.149	0.040	0.001	0.067	0.232
Household uses protected drinking water source	0.134	0.040	0.002	0.053	0.216
Network betweennessf	0.030	0.008	0.001	0.013	0.047
Constant	0.095	0.050	0.067	− 0.007	0.197

Obs. 59, R² = 0.331, F-stat. 13.69, F-stat. p value < 0.0001. Root mean squared error (RMSE) from 8-fold cross-validation = 0.183. Variables selected through Lasso with 8-fold cross-validation. Mean squared error (MSE) of Lasso cross-validation = 0.076

^aThe results shown are from an ordinary least squares regression with standard errors clustered by the village

^bIn-group bias is positive if the CMD was altruistic towards their friends and not altruistic towards strangers

^cThe base category includes CMD selection by community meeting or direct nomination from a village health team member

^dMDA = mass drug administration

^eThe base category for these occupations includes all other CMD occupations

^fNetwork betweenness was not significant (p value > 0.05) when one outlier was removed (Additional file 1: Table S4) 58/59 CMDs had network betweenness ≤ 3 whilst 1/59 CMDs had network betweenness = 11.58