Abstract :
A reliable and valid test is essential to measure students learning outcomes. The
present study was designed to construct a valid Biology test and to analyze the
achievement of 9th
-grade science students. The sample of the present study was
two hundred and nine (209) students who were selected from nine Government
schools and private schools by using a simple random sampling technique. For
the validation process, two parallel forms of achievement tests were constructed
from the subject of Biology. Each form contains thirty-five MCQs. Items were
selected from the textbook of Biology of grade 9th and administered to two
hundred and nine students (Male and Female) in different private and government
secondary schools for boys and girls in Multan city. The validity and reliability of
tests were also ensured. Scoring of items was done in a dichotomous manner i.e
either correct or incorrect. “Z” test was applied to see the difference between the
mean performance of private schools and Government schools, and it was found
statistically significant. In the case of male and female performance, the “z”-test
was found to be insignificant. Content validity was achieved following a table of
specifications. The correlation coefficient between the two forms was found to be
0.78. Kuder Richardson-21 was also used to compute the reliability of the test.
Item analysis was done on four criteria i.e. facility index (FI) discrimination index
(D), phi coefficient, and point biserial correlation (rpbis). Based on all four criteria,
fourteen items were rejected from parallel form no.1 and fifteen items were
rejected on parallel form test no. two. It is recommended to use more than one
criteria to develop and a valid achievement test so that a good pool of items may
be generated.
Keyword :
achievement test, biology, reliability, traditional analysis, validity