A Construction of a Pyramid Test with Multiple-Item in Each Level on the Set Topic, Mathematics Strand of Subjects for Matthayomsuksa 4 Students by Using a Computer Program

Abstract

A pyramid test with multiple items in each level by using a computer in doing
a test assisted the students for their convenience in choosing the next coming question
in each level of the pyramid to the tester to get a question which having a difficulty
value appropriately with the tester’s level of ability. This research aimed to construct a
pyramid test with multiple items in each level which using the computer in doing the
test, investigate the quality of the constructed pyramid test and evaluate the students’
satisfaction toward using the pyramid test which were conducted on computer. Samples
were 1,090 Matthayomsuksa 4 students attending the first semester of the academic
year 2010 at 7 schools under the Udonthani Office of Education Service Area Zone 1
whom were obtained by using the multi-stage random sampling technique. There were
two research instruments: three 50-item multiple-choice tests of Mathematic Subject
on Set Topic for Matthayomsuksa 4 students testing three times; the first test conducted
on 120 students whom were divided into 3 groups of 40 students using one test on
each group by conducting the normal test in order to find out the difficulty and
discriminating power before selecting the items with appropriate standard quality for
40 items in each test; the second test used 900 students by divided into 3 groups of
300 students using one test on each group by using a normal test to gain the
parameter value of the test by using the BILOG-MG3 6 Month Rental and select the
items with appropriate standard quality for 28 items on each test and 84 items in total
and construct the pyramid test with multiple-item in each level in 7 levels which
would be conducted on computer and the third test used 70 students from two
classes of 35 students each class who did the pyramid test with multiple-item in each
level on computer, the individual tester would do the test on 21 items by retrieving
questions from the computer differently based on personal ability autonomously for
two times with a two-week recession between each time and brought the results of
the test to find out its quality. The 25-item 5-rating scale questionnaire on the
students’ satisfaction toward using the pyramid test with multiple items at each level
by computer conducting on the same 70 students in a normal test.
The results of the study were as follows:
The first investigation on the 50-item test using the normal process to
indentify its quality in each item had a difficulty value ranged from 0.58 to 0.78 and a
discriminating power ranged from 0.20 to 0.70.
The second experiment on the 40-item test using the normal method was
the investigation of 3 parameters under the theory of test construction had a
discriminating power ranged from 0.373 to 6.332, a difficulty value ranged from -2.604
to 2.625 and the estimation coefficient ranged from 0.017 to 0.285.
The third investigation was on the test of 84 items in total using computer
in doing a test. The test had a reliability of 0.919 and the capability average scores
from the previous two tests combining with the learning achievement of the Basic
Mathematics Subject ค31101 of the sampling to find out the correlation coefficient
which was consistent with the criterion –related validity and revealed that class one
had the criterion –related validity at 0.737, class two had the criterion –related validity
at 0.800 and the overall criterion –related validity was 0.746. An analysis of the
reliability of the satisfaction survey was 0.92 and the sampling group had satisfaction
toward the pyramid test with multiple items at each level at a high level.