Table 3.

The effect of method of data collection on healthcare utilization findings of free care experiment

	Coefficient on interaction	95% CI	p-value	N	R2
	(1)	(2)	(3)	(4)	(5)
Any clinic visit past one month	−0.036	−0.076, 0.004	0.078	4,632	0.0118
Any clinic visit past six months/one year	−0.054	−0.101, -0.007	0.024	4,632	0.2110
Any informal care visit past one month	−0.00037	−0.041, 0.040	0.986	4,632	0.0150
Any informal care visit past six months/ one year	−0.0017	−0.051, 0.048	0.948	4,632	0.1805

Notes: Data are stacked such that each observation is a child corresponding to one of the two data collection methods. The coefficient reported is on the interaction between a dummy for free healthcare and a dummy for whether the data were collected through the household survey. It identifies the difference in treatment effect between the pictorial diary and household data. The 95% CI and p values are based on standard errors that are corrected for clustering at the household. All regressions control for mother’s education, number of children in household, age of the child, household wealth and dummies for male child, distance from the nearest health centre, religion, ethnicity, free healthcare and the method of data collection.