Abstract:
A test, like any other type of instrument used to measure should give the same results every time it
measures, if it is used under the same conditions. This is considered the Reliability of a test. But
human beings do not behave in exactly the same way on every occasion, even when the circumstances
seem identical. The performances of students on any test can vary for a variety of reasons.
The purpose of this research is to observe the reliability of two main tests. First I took a sample of
test scores of the Placement Test (a written test) for “English for Business Communication” course
(Test-1-A). This test was held in an examination hall in a fairly formal environment. After about one
month, I took the same group of students and held the same test inside my classroom (Test-1-B).
As the next step I held an oral test (presentation) again in a fairly formal environment (Test –2- A)
for this particular group. Next, I held the same oral test inside my classroom for the same group of
students. (Test-2-B)
This paper attempts to focus on the reliability of these two tests by calculating the Reliability
Coefficient - K-R 21. An analysis on how these reliability coefficients vary due to the sources of
Error Variance, as a result of environment (e.g. location, space, ventilation) and also as a result of
variance attributable to examinees (e.g. emotion, memory) will be discussed.
At the conclusion of this investigation, we would be able to find out how these sources of Error
Variance affect the reliability of the two tests. Moreover, we can minimize such Error Variance in
test-administration to some extent.