Elements of Difficulty Level in Mathematics* - Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed Computing - page 83

So, this method cannot use when examination designer must hide these ques-
tions. For example, problems used admission test in Japanese university must be
created originally and unreleased.
In addition, we have a data about IRC. Fig.1 and Fig.2 are the results of
employment mock exam. In Japan, many companies do paper test as employment
exam. Pieces of this exam are “Language area” and “Non-language area”. “Lan-
guage area” is including Japanese and English. “Non-language area” is including
Mathematics and Science. Fig.1 is “Language area”, and Fig.2 is “Non-language
area”. Vertical axis of Fig.1 and Fig.2 is ability of each learner, and Abscissa axis
is point of each learner of employment mock exam. Ability of each learner calcu-
lated used the way of IRC.
Fig. 1 Result of employment mock exam (Language area)
Fig. 2 Result of employment mock exam (Non-language area)
We calculated coefficient of correlation about Fig.1 and Fig.2. Coefficient of
correlation about Fig.1 is 0.96075, and Fig.2 is 0.86258. These points are very
high, but point of “Language area” is more than “Non-language area”. This means
we understand that this result shows that it is difficult for Mathematics to use IRC.