11,522 Publications · 65,312 Citations

In 16 participants from OpenNeuro ds000005, the additive quadratic likelihood-ratio test for equal block intercepts gave the same six Holm-adjusted rejections as the linear and full quadratic tests. Its six statistics ranged from 11.3052 to 36.9726. The additive quadratic score test gave five rejections; sub-14 changed decision across the model and statistic. Two full quadratic sub-10 fits were quasi separated and had undefined statistics. Simulation of the additive score statistic gave pointwise Monte Carlo limits that did not resolve sub-14 or sub-01. These asymptotic and plug-in simulation results concern a specified choice function in one session. They do not establish a causal block effect, a value function, or a loss-aversion trait. The source lists differ by block, missing choices remain, and block time order was not checked.