CHARACTERISTICS OF A GOOD TEST: RELIABILITY (GROUP 5)
Lies Nureni & Lukman Chamdani
Reliable = Stable = Consistent
A reliable test is the test that can produce stable scores or consistent scores.
A reliable test should be demonstrated by the score consistency within raters or intra raters, between raters or inter raters, and across time and place.
We can say that a test has a high reliability if the scores demonstrate consistency no matter who administer the test, when the test is administered, and where the test is administered.
How do we formulate reliability?
X: obtained score/observed score
T: true score
A high reliability test will contain no error or the error is 0. Thus, the obtained score will be the same with the true score.
Reliability can be estimated by using several approaches:
- 1. Test Retest
The same test is administered twice to the same group of subjects on the different testing occasion.
|ü It has only one set of test to be constructed.||ü Time and energy consuming.
ü It is not easy to create similar condition on different time.
ü It is not known exactly the best time interval to conduct the second test. It can be too far or too close.
- 2. Equivalent Forms
It requires the construction of two or more sets of tests which are equal in every aspect of the tests. The two tests should be administered at the same time or at least in a very minimum time interval.
|ü The two forms can be used independently.||ü Time and energy consuming.
ü It is not easy to generate lots of items that reflect the same construct.
ü It is not easy to keep the test takers’ mental condition the same.
ü It is not easy to determine the interval of the testing time.
- 3. Internal Consistency
The subjects complete one instrument one time. There are two types of internal consistency:
- a. Split-half
The reliability is obtained by correlating the result of the halves of the test. A total score for the odd number questions is correlated with a total score for the even number (although it may be the first half with the second half). The split-half reliability is simply the correlation between those total score. However, different results of reliability values due to making different splits or grouping are obvious.
- b. Inter-item Estimation
This is basically a kind of estimation where the test scores are correlated with themselves within the test.