“…This is of course different from an empirical statement that students' submitted preferences always reflect their true preferences, and recently there has been a stream of work on this, much of it taking flight from Li (2017) on the distinction between strategy-proof and "obviously strategy-proof" mechanisms. Stable matching mechanisms are not obviously strategy-proof (Ashlagi and Gonczarowski 2016), and recent empirical work has focused on identifying cases in which some submitted preferences can be identified as departing from the likely underlying preferences, see, e.g., Rees-Jones (2017, Rees-Jones and Skowronek (2018), Hassidim et al (2017), and Shorrer and Sóvágó (2017). This work supports the conclusion that the instructions that accompany a mechanism (e.g., regarding strategy-proofness) are an important part of its design, and that effective instructions may be hard to design.…”