30
was that it is more practical than rewriting form because it would not take too much time for students to write their answers. Besides, the students’ authentic
answers would be clearer compared to multiple-choice test since there was less opportunity for them to predict or choose the answers randomly.
c. Face Validity
“Face validity means that the students perceive the test to be valid” Brown, 2004: 27. It means that face validity is validity that appeared from the
students’ perspective. Brown 2004 states that face validity would be fulfilled if the students found:
- the test is built well with familiar tasks - the total number and the time allocation in doing the test are balanced
- the test items are not complicated - the test provides clear and understandable directions
- the test really measures what should be measured The face validity would be known through students’ responses to the pilot test.
From the pilot test, the researcher would be able to know what should be improved from the test items.
2. Test Reliability
“A reliable test is consistent and dependable” Brown, 2004: 20. The test reliability can be estimated through reliability coefficient. According to Fraenkel
and Wallen 2008, reliability coefficient shows the “relationship between scores
of the same individual parts of the same instrum
Related to the t using the relationship be
same instrument. It wa conduct one pilot test
approach in calculati formulas. They are KR
KR21 is used when the Besides, it is used for
researcher chose KR21 according to Fraenkel a
From the formul M is the mean of the s
of test scores. Reliabi 1983, there is a criter
Table 3.2.: Crude Cr Coefficien
.00 to .20 .20 to .40
.40 to .60 .60 to .80
.80 to 1.00
viduals on the same instrument at two different tim instruments” p. 155.
the time efficiency, the researcher checked the test onship between scores on the same individuals on tw
It was considered more efficient since the resear ot test only. In this case, the researcher used Kude
lculating the reliability coefficient, which has are KR20 and KR21. According to Fraenkel and W
hen the test items are considered having the same di used for items that are scored right versus wrong.
KR21 formula in counting the reliability coef enkel and Wallen 2008: 156 is formulated as:
formula, it is known that K shows the number of of the set of test scores, and SD is the standard devi
eliability coefficient ranges from 0.00-1.00. Acc criterion of coefficient, which is shown in Table 3.2.
de Criterion for the Evaluation of a Coefficient Bes fficient r
Relationshi
.00 to .20 .20 to .40
.40 to .60 .60 to .80
.80 to 1.00 Negligible
Low Moderate
Substantial High to very hi
31
rent times, or on two
the test reliability by s on two parts of the
researcher needed to d Kuder-Richardson
has two kinds of l and Wallen 2008,
same difficulty level. rong. Therefore, the
y coefficient. KR21,
ber of the test items. d deviation of the set
According to Best able 3.2.
t Best, 1983: 255 ionship
ligible oderate
tantial very high