“…While these considerations imply that our sample allows identifying behavioral regularities with regard to medical service provision, we are careful with drawing conclusions with regard to specific effect sizes.17 However, as pointed out by a reviewer, medical students may care more about patients' outcome than students in other disciplines. The ranking would then be more open.18 Starting with Hennig-Schmidt et al (2011), similar or equivalent mechanisms have been employed in several behavioral experiments in health analyzing medical service provision.19 Physician altruism is a common assumption in health economics models (see, e.g.,Allard et al, 2011;Ellis & McGuire, 1986;McGuire, 2000).This assumption has been supported by previous laboratory research (see, e.g.,Attema et al, 2023;Brosig-Koch, Hennig-Schmidt, et al, 2017;Godager & Wiesen, 2013;Li et al, 2017Li et al, , 2022.20 More specifically, we obtain that q H (α) = q H maximizes utility U H (q; α) for α ≥ 1/3, while q L (α) = q L maximizes U L (q; α) for α ≥ 13/23.21 The assumption of separability represents a standard assumption in the literature when the choice of treatment quality is examined (see, e.g.,Ellis, 1998).22 Notice that, for the case of profit-maximizing physicians, (α 1 , α 2 ) = (0, 0), our duopoly model of quality competition is similar to the Bertrand model of price competition. However, unlike the Bertrand model, our model has a discrete strategy set Q and allows for customer-oriented firm objectives.23 As opposed to explicit collusion, tacit collusion refers to cooperative behavior between competitors that do not explicitly communicate with each other or exchange information otherwise.24 For between-subject analyses, we employ Mann-Whitney U tests.…”