Table 2.

Marginal effects indicating the probability of patients selecting provider types due to changes in provider attributes and patient characteristics.

	PHC	Private Clinic	Private Hospital
Reduce household distance to PHC by 1 Km	0.0004 [0.0002, 0.0010]	−0.0004 [-0.0009, −0.0002]	−0.0001 [-0.0001, −0.00002]
Increase PHC clinician competence (76/100)	−0.0011 [-0.0039, 0.0018]	0.0010 [-0.0016, 0.0036]	0.0002 [-0.0025, 0.0005]
Reduce patient costs at PHC by 50%	0.0076 [0.0065, 0.0087]	−0.0070 [-0.0079, −0.0060]	−0.0009 [-0.0011, −0.0008]
Severe patient (ref: not severe)	−0.0019 [-0.0027, −0.0011]	−0.0326 [-0.0476, −0.0176]	0.0519 [0.0295, 0.0474]
Wealth quintile
Quintile 2 (ref: Poorest 20%)	0.0029 [0.0019, 0.0038]	−0.0329 [-0.0485,- 0.0172]	0.0451 [0.0217, 0.0686]
Quintile 3 (ref: Quintile 2)	0.0012 [-0.0001, 0.0026]	−0.0267 [-0.0457, −0.0077]	0.0383 [0.0099, 0.0667]
Quintile 4 (ref: Quintile 3)	0.0050 [0.0037, 0.0062]	−0.0364 [-0.0529, −0.0199]	0.0472 [0.0225, 0.0719]
Richest 20% (ref: quintile 4)	0.0053 [0.0039, 0.0067]	−0.0563 [-0.0742, −0.0385]	0.0768 [0.0501, 0.1034]
Observations	19398	19398	19398

Note: Figures are marginal effects. This is the change in probability of a patient selecting a particular provider associated with a unit change in a continuous independent variable or a unit change from the reference group in a binary variable, with all other variables at their reference value. Multiplying these figures by 100 gives the percentage point change in a patient selecting a particular provider associated with a unit change in the independent variable. Marginal effects are based on regression results in Appendix A, Model 4.