Methods of Estimating Reliability of a Test

TEST RETEST OR STABILITY

The students received the test on Wednesday and were given the same test again on Tuesday.

If the correlation between the twosets of scores is high then the tests arereliable.

However memorization or experience may be involved in the second test, which means the test might be unreliable.

INTERNAL CONSISTENCY

Tests measure one idea. If the test items arerelated to each other then the student shoulddo well on the test. There are two ways to dothis.

Split half Method: each test item isassigned to a group, then the totalscore is taken.

Compare eachgroup to each other, is there acorrelation?

Kuder Richardson Method:

This methodcompares the test consistency witheach other.

ALTERNATIVE FORM

Two types of the same test are made andadministered to students, for example Form Aand Form B.

This is usually created bytextbook publishing companies.

The issuewith type of testing is how difficult it is tocome up with two reliable tests.