Table 3. Impact on adherence of dispensing exact quantities required.

	Strict adherence criterion (no pills left)	One day tolerance criterion (less than four pills left)	Mixed adherence criterion (counting method combined with the Morisky scale)
Treated	0.23*	0.20*	0.21*
Pharmacies	[0.15; 0.31]	[0.12; 0.29]	[0.14; 0.28]
Excluding individuals who refused per-unit dispensing	Yes	Yes	Yes
Observations	1185	1185	1185
Censored observations	984	984	975

Notes: Probit estimations, with sample selection equation (at first step, using Heckman command). Marginal effects reported. Robust standard errors clustered at pharmacy level. IC at 95% reported in brackets.

* p< 0.01.

Control variables are age and type of treatments; they were introduced after a stepwise procedure which determines their significance at the critical p-value of 0.10.