“…Firstly, ML estimators require us to make a number of simplifying assumptions (see Discussion). These assumptions may be acceptable in well-trained, relatively homogeneous groups of adults, but may be inappropriate for children, who, for example, often exhibit high levels of inattentiveness (Godwin et al, 2016;Jones, 2018b;Jones, Kalwarowsky, Braddick, Atkinson, & Nardini, 2015;Kaunhoven & Dorjee, 2017;Manning, Jones, Dekker, & Pellicano, 2018;Moore, Ferguson, Halliday, & Riley, 2008;Smallwood, Fishman, & Schooler, 2007;Wightman & Allen, 1992;Witton, Talcott, & Henning, 2017), response bias (Trehub, Schneider, Thorpe, & Judge, 1991;Werner, Marean, Halpin, Spetner, & Gillenwater, 1992), and other nonstationary behaviors. The concern is that such deviations from an ''ideal'' observer might at best degrade the efficiency of the test compared to standard Staircases, and at worse may cause the results to become excessively noisy or biased.…”