CHAPTER 06
ESTIMATES ESTIMATES OF RELIA
THREE GENERAL METHODS OF ESTIMATING RELIABILITY,
EMPHASIZING THAT EACH METHOD REQUIRES TWO OR
MORE “TESTINGS”
ALTERNATE FORMS METHOD OF ESTIMATING
RELIABILITY
The alternate forms method (sometimes called parallel forms
reliability) is one method for estimating the reliability of test
scores. By obtaining scores from two different forms of a test,
test users can compute the correlation between the two forms
and may be able to interpret the correlation as an estimate of
reliability.
The ability to interpret the correlation between alternate forms
as an estimate of reliability is appropriate only if the two test
forms are parallel.
TEST–RETEST METHOD OF ESTIMATING
RELIABILITY
The test–retest method requires the same people
to take the same test on more than one occasion.
To the degree that observed scores from one
testing occasion are consistent with observed
scores from the second occasion, the test is
reliable.
INTERNAL CONSISTENCY METHOD OF ESTIMATING
RELIABILITY
The internal consistency method has the practical
advantage of
requiring respondents to complete only one test at only
one point in time.
The idea behind the internal consistency approach
is that the different “parts” of a test (i.e., items or groups
of items) can be treated as different forms of a test.
SPLIT-HALF ESTIMATES OF RELIABILITY
Split-half reliability is an approach to test
reliability where two parallel subtests of equal
size are created from a single test, allowing for
the computation of subtest scores to estimate
total test reliability.
RAW” COEFFICIENT ALPHA
“Raw” coefficient alpha (often called Cronbach’s
alpha), which is the most widely used method for
estimating reliability. The first step in computing
alpha is to obtain a set of test-level and item-level
statistics.
STANDARDIZED” COEFFICIENT ALPHA
Another method of estimating reliability is often
called the generalized Spearman - Brown formula
or the standardized alpha estimate. Standardized
alpha is used when a test score is created by
aggregating standardized responses to test items,
providing an estimate of reliability.
FACTORS AFFECTING THE RELIABILITY OF TEST
SCORES
The consistency among test parts directly impacts
reliability estimates. A test with greater internal
consistency, as indicated by a split-half correlation,
average interitem covariance, or average interitem
correlation, results in scores with higher estimated
reliability.
DEFINING DIFFERENCE SCORES
A difference score, defined as the difference between
a child’s scores on two tests, measures their level of
psychological attributes like reading skill
improvement, based on their scores on the x and y
tests.
• Change score:
Change scores, in which each person has two scores on
the same test or measure, with each person thus
having a difference score based on the difference
between the two test scores.
Discrepancy score:
Discrepancy score, in which each person again has
scores on two measures but in which the two measures
are from different tests. For example, an educational
psychologist might be interested in the discrepancy
THANK YOU