THE MAXIMUM LIKELIHOOD ESTIMATES BASED ON THE INCOMPLETE QUANTAL RESPONSE DATA

抄録

In this paper we present statistical techniques appropriate to one exceedingly important type of response, that known as all-or-nothing or quantal.
With incomplete quantal response data, the estimation of parameters has usually been solved approximately as the estimation problems based on grouped data, which is not always practicable or adequate [1]. We therefore propose a new method of maximum likelihood for estimating parameters based on ‘individual incomplete quantal response data’ without grouping. The existence theorems of its maximum likelihood estimates (its optimal solutions) are proved in the appendix.