Table 1.

Average predicted absolute increase in healthcare access delay for phone users (in days).

	Sequence-insensitive measures						Sequence-sensitive measure		Difference between delay estimates based on sequence-insensitive and sequence-sensitive measures in percent
	Option a) Phone ownership		Option b) Any health- related use		Option c) Use during access		Option d) Relevant phone use
	Rajasthan	Gansu	Rajasthan	Gansu	Rajasthan	Gansu	Rajasthan	Gansu	Rajasthan			Gansu
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9) [(1) - (7)] (7)	(10) [(3) - (7)] (7)	(11) [(5) - (7)] (7)	(12) [(2) - (8)] (8)	(13) [(4) - (8)] (8)	(14) [(6) - (8)] (8)
Any Healthcare Provider (n=592)	0.0a	0.0a	+11.0	+3.0	+6.0	+1.9	+7.4	+2.1	–100%	+50%	–18%	–100%	+42%	–12%
Public Providers (n=380)	0.0a	0.0a	+34.2	+3.6	+39.2	+3.7	+41.8	+4.0	–100%	–18%	–6%	–100%	–9%	–5%
Private Providers (n=264)	0.0a	0.0a	+10.5	+1.1	0.0a	0.0a	+7.3	+0.7	–100%	+43%	–100%	–100%	+43%	–100%
Informal Providers (n=81)	+63.8	+0.9	+168.5	+1.3	0.0a	0.0a	0.0a	0.0a	N/A	N/A	N/A	N/A	N/A	N/A

Notes: Predicted values based on negative binomial regression model with village-cluster robust standard errors. Observations at the illness episode level, with up to one self-reported “mild” and “severe” illness per person.

^aCoefficient in negative binomial regression model was not statistically significant at the 10% level.

Source: Authors.