Reliability
• Reliability = Indicator of how consistently (precisely) the test
measures the psychological construct.
– Consistency of test results from one administration of the
test to another administration (testretest reliability).
– Consistency of randomly chosen two subset of items in
the test (splithalf reliability).
– Consistency of each item to the other items in the test
(internal consistency).
• Virtually, all psychological measurements are not perfectly
reliable.
→
We want to know how reliable the test is.
• The measure of reliability is based on what we call “classical true
score test theory”.
The Classical True Score Model
X
=
T
+
E
where
X
is the observed score
T
is the unobserved true score (fixed)
E
is the error score (random)
(example)
• Assume your true score for a specific test is known to
be 108.
 True score is always unknown, though.
• You have obtained 112 on the test.
• Then,
X
= 112,
T
= 108, and
E
= 4.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentObserved Score as a Random Variable
This is the end of the preview.
Sign up
to
access the rest of the document.
 Spring '06
 Fenley
 Variance, Randomness, Classical True Score Model

Click to edit the document details