Simulation of Computer Adaptive Learning and Improved Algorithm of Pyramidal Testing

1
Simulation of Computer Adaptive Learning and
Improved Algorithm of Pyramidal Testing
FP1
Taras Lendyuk1, Sergey Rippa2, Svitlana
Sachenko1 1 Ternopil National Economic
University, 3 Peremoga Square, Ternopil, 46020,
Ukraine, E-mail tl_at_tneu.edu.ua,
ssachenko_at_yahoo.com 2 National University of
State Tax Service of Ukraine, 31, Karl Marx
Street, Irpin, 08201, Kiev region, Ukraine
E-mail rippa_serg_at_ukr.net
Abstract In this paper the implementation of
computer adaptive testing using
mathematical-statistical tools of Item Response
Theory is considered. It is aimed for knowledge
level quick determination of full-time students
and distance learning students to identify gaps
in their knowledge. The goal of computer adaptive
testing is the student knowledge assessment.
There are made commendations for adaptive
learning system to individual learning path
construction. Such system can be used in within
LMS as well as for students self-training.
When the student passes test with minimally
required grade or higher one, he/she is
encouraged to study at a higher level of
difficulty. In a case a student agrees to
continue, the educational material is expanded
with additional learning objects of higher
difficulty level. If the student does not agree
to continue then the learning process terminates.
Therefore learning system controls the receiving
of appropriate education material by student.
Also the system prepares the tests with desired
minimal grade. The higher grade obtaining depends
on the student desire to continue learning.
Existing Methods of Computer Adaptive Testing
The main disadvantage of the classical tests is a
fixed number of questions where a student must
answer all questions and spend a significant
time. For overcoming of this one the computer
adaptive testing (CAT) is used, which has a
possibility of adaptation to students needs and
to their knowledge level. During test passing the
students model is filled.
The Proposed Model of Adaptive Testing
Authors proposed to combine the question in
thematic blocks with an odd items number, for
example three or five to determine the majority
of correct or incorrect answers. At the majority
of wrong answers student moves to a lower level
of questions difficulty and in case of all wrong
answers two levels down. With the majority of
correct answers student is moving to higher
difficulty level and at all correct answers he
moves two levels up. This will improve the
students motivation with a high knowledge level,
because they will see that adaptive system
properly assess their high knowledge level and
gives them the difficult questions. For bad
students the motivation is increased in the case
of correct answers to the question, although the
question is easy.
Fig. 2 Cycle diagram of adaptive testing
Experiments
There were 24 students tested on 40 test
questions of subject Algorithmic and
Programming within Computer Science Curricula,
which were used to determine the difficulty of
test questions. For the procession convenience
answers during processing were sorted by
difficulty, and results was ordered by grades.
The experiment used a pyramidal testing,
classical testing and improved algorithm of
pyramidal testing. The duration of classical test
was 1052 s, pyramidal test 438 s, and the
improved algorithm pyramidal test 332 s.
Designed algorithm of adaptive learning system is
implemented as a plug-in for LMS Moodle. This
plugin provides student registration before
learning of educational material, then the system
checks students enrollment on the selected
course, and students data are recorded into the
model. During adaptive testing students model is
filled with test results, items list and learning
objects difficulties, and the time spent on each
response.
Conclusion Improved algorithm of pyramidal
testing is proposed. and its implemented as a
plug-in for LMS Moodle. There are taken into
account the items of difficulty as well as tests
are adapted to the students needs. As a result
the students motivation is increased, and the
duration of adaptive tests is reduces in about
25.
Fig . 1 An improved algorithm of pyramidal
testing
As it is shown in Fig.1, for example, the student
answering questions of test unit 3, can be moved
to blocks 811, depending on the answers. On the
arrows there are indicated numbers of correct
answers. If a student gives the three correct
answers he/she will be moved to the block 11,and
a student who answered all the questions wrongly
will be moved to the block 8. Student, who gives
one or two correct answers, will be moved to
blocks 9 or 10 respectively.