INCOMPLETE QUANTAL RESPONSE DATA ANALYSIS －Maximum Likelihood Estimation of Parameters Based on Mixtures of Interval Data and Ordinary Ones－
In this paper, we attempt to classify the situations generating quantal response data into three main groups and develop a method of maximum likelihood estimation based on incomplete quantal response data. Incomplete quantal response data often arise in medical and biological examinations. In some cases the tolerance of each test subject can be known only to be above or below the value which is given or decided on the day of group examination. In such situations, the standard techniques of maximum likelihood estimation of parameters cannot be applied, because the values of observations are not specified. The following method of maximum likelihood estimation of parameters based on incomplete quantal response data will have a wide range of application in statistical estimation problems.