Table 3.

Difference-in-difference analysis of knowledge changes by cadre

	(1)		(2)		(3)		(4)		(5)		(6)
	History: Previous pregnancies		History: Current pregnancy		History: Medical history		Examinations		Investigations		Drugs/ supplies
	Coeff.	CI	Coeff.	CI	Coeff.	CI	Coeff.	CI	Coeff.	CI	Coeff.	CI
Panel A: Midlevel care providers
Difference-in-difference coefficient (QI*Endline)	1.03**	(0.4–2.01)	0.91	(−0.38–2.2)	0.13	(−0.71–0.97)	1.15*	(−0.05–2.34)	0.47	(0–0.52-1.47)	0.22	(−0.15–0.60)
Observations	307		307		307		307		308		307
R-squared	0.066		0.031		0.091		0.032		0.044		0.244
Panel B: HEWs
Difference-in-difference coefficient (QI*Endline)	0.57*	(−0.05–1.19)	1.66***	(0.8–2.51)	0.35	(−1.67–0.87)	0.68*	(−0.11–1.46)	0.64	(−0.14–1.42)	0.50***	(0.28–0.73)
Observations	401		401		396		399		375		399
R-squared	0.045		0.038		0.067		0.016		0.026		0.305

Notes: Coefficients are interpreted as the change in the number of correct items listed by respondents in QI compared with comparison areas. Each column shows the results of an ordinary least squares regression model where the dependent variable is the number of relevant items that a patient-facing health worker mentioned when asked an open question regarding the care of Mrs X, a married woman of 26 years who has moved into the area and comes to see you for the first time. The dependent variable of Column 1 is the number of items respondents report from a list when asked ‘what questions would you ask Mrs X about her previous pregnancies’. In Column 2, the question asks about her current pregnancy. In Column 3, the question asks about her medical history. In Column 4, the question asks what physical examinations would be performed. In Column 5, the question asks what investigations would be performed. In Column 6, the question asks what drugs or supplies the respondent would provide to Mrs X. Standard errors are clustered at woreda level, regressions include facility fixed effects.