Page 92 - DEDU504_EDUCATIONAL_MEASUREMENT_AND_EVALUATION_ENGLISH

Educational Measurement and Evaluation

Notes (c) A test is split half using several methods. Therefore, reliability coefficient can be
different for splitting half a test using different methods. This type of coefficient has
no unique value.
(d) Sometimes, the two parts of a test may not be homogeneous.
(e) In this test, the basis of reliability is half-test, though its reliability is counted for the
full-length test, so from this view, splitting half a test may reduce its reliability.
Merits
(a) The chief characteristic of this method is that it does not consume much time and
energy. A group is administered the test only once. Therefore, in such a situation
where a retest is not possible, nor is it possible to prepare the two parallel or
homogeneous tests, then this method becomes very important. Such a situation arises
when reliability of a performance test or personality test has to be calculated, because
these two types of tests cannot be constructed in parallel or homogeneous form, nor
can they be given for retest; for example, Rorschach test or TAT test. Besides, this
method is also suitable for calculating aptitude and interest.
(b) As the two parts are administered at the same time, there is no influence of time
interval, practice, memory etc. on scores.

Hoyt’s Reliability
Like Kuder-Richardson, Hoyt (1941) too laid down a method to find out reliability coefficient.
According to this method, differences or deviation occurring in achievement of an individual
from one item to another are not errors, but they are actual differences which we call intra-
individual difference.
Hoyt has defined reliability as follows :
“... total variation observed is conceived to be made up of three components, true inter-individual
differences, intra-individual differences (measured by item variance) and error inter-individual
differences.”
The above definition can be described in the following equation :

2
σ = σ + t 2 σ 2 i σ + e 2
x
2
Where, σ = Variation in scores
x
2
σ = Item variation
i
2
σ = Error variation
e
2
σ = Total variation
t
Therefore, according to Hoyt, the suitable definition of reliability is as follows :

σi
σ t 2 ( 2 – σ 2 ) σx – e 2
i
VX = σ x 2 – i 2 = σ 2 – i 2
σ
σx
On applying variance method, the calculation of reliability can be found out by the following
equation :
MS – MS
VXX = ind resi
i
MS ind

86 LOVELY PROFESSIONAL UNIVERSITY