How do pro-social tendencies and provider biases affect service delivery? Evidence from the rollout of self-injection of DMPA-SC in Nigeria

Table 3 Association between pro-social tendencies and willingness to dispense DMPA-SC for SI

Regressor	Outcomes
	Stated intentions from in-depth interviews
	Offer discount	Stated intention to offer DMPA-SC for SI
		Older	Younger
Altruism	−0.024 (−0.134, 0.049)	−0.004 (−0.029, 0.021)	−0.065* (−0.104, −0.027)
Dictator game	0.537** (0.206, 0.868)	0.076 (−0.088, 0.240)	−0.081 (−0.362, 0.200)
N	52	81	81
Mean of dependent variable	0.385	0.914	0.778
	Dispensing behavior from mystery client interactions
	Availability of DMPA-SC		Offered DMPA-SC for SI		Perceived differential treatment
	Older	Younger	Older	Younger	Older	Younger
Altruism	−0.016 (−0.046, 0.013)	−0.019 (−0.057, 0.019)	−0.031 (−0.100, 0.039)	−0.038 (−0.105, 0.030)	0.001 (−0.047, 0.049)	0.016 (−0.056, 0.088)
Dictator game	0.204 (−0.003, 0.411)	0.104 (−0.089, 0.293)	0.167 (−0.122, 0.457)	0.380* (0.103, 0.658)	0.017 (−0.262, 0.296)	−0.215 (−0.537, 0.108)
N	81	81	0.296	0.272	0.173	0.333
Mean of dependent variable	0.877	0.877	81	81	81	81

Coefficients and 95% confidence intervals reported from linear regressions between pro-social tendency measures (regressors) and stated intentions/actual dispensing behavior outcomes. Each column refers to a different outcome. Altruism is measured from 0–10 (highest); dictator game is measured as percentage share of the initial monetary gain one is willing to donate (higher means more pro-social). A positive coefficient means that pro-social tendency measures are positively correlated with the outcome in question
^*p < 0.05
**p < 0.01
***p < 0.001

ISSN: 1472-6874