Instrument Validity and Reliability
b.
Construct Validity
Construct validity is concerned with whether the test is line with the theory of
learning strategies toward reading what it means to know the language that is being measured. Construct Validity is concerned with whether the test is
actually in line with the theory of what it means to know the language Shohamy, 1985: 74.
In this research, the researcher formulated table of specification. So every test
items could be matched with the goal and the materials had been taught. Regarding the construct validity, it measures whether the construction had already
in line with the objective of the learning Hatch and Farhady, 1982: 251. To know whether the test is good reflection of the knowledge which the teacher
wants the students to know, the researcher compared the items of the test with table of specification. If the table represents the material that the researcher wants
to test, then it is valid from that point of view. A table of specification is an instrument that helps the test constructor planned the test see Appendix 2.
Table 2. Table of Specification of Data Collecting Instrument No.
Objectives No. of
Items Percentages
Question Number
1. Determining the Main
Idea 8
16 9, 16, 21, 33, 36, 41, 44, 47
2. Identifying Specific
Information 12
24 1, 2, 6, 10, 12, 15, 17, 20,
23, 24, 27, 30 3.
Inference 8
16 4, 22, 26, 31, 39, 45, 46, 48
4. Reference
11 22
3, 8, 13, 18, 25, 28, 34, 37, 38, 42, 50
5. Vocabulary
11 22
5, 7, 11, 14, 19, 29, 32, 35, 40, 43, 49
Total 50
100 50
Basically, the construct and content validity are overlap. It is a representative of the material from the subject. In line with Nuttal 1985 the relation validity of
instrument refers to construct validity in which the question represents five reading skills, they are determining the main idea, identifying specific
information, inference, reference, and vocabulary. Skills of reading in the test are a part of the construct validity and the item numbers are the part of the content
validity. Besides the construct validity, the researcher used inter-rater analysis. The
researcher used 4 inter-raters to make the reading test more valid. They are Jonatan Eko S. Budi, S.Pd., Drs. Muaddin Mubarok, Imelda Susan, S.Pd., and
Nasirwan, S.Pd. They are English teachers at SMAN 14 Bandar Lampung. They were at least 7 years teaching experience.
Table 3. Inter-rater Judgment
No. Item
Number Skill of Reading
Inter-rater Judgment Total
Percentage R1
R2 R3
R4
1. 9
16 21
33 36
41 44
47 Determining the Main Idea
25 25
25 25
25 25
25 25
25 25
25 25
25
25 25
25 25
25
25 25
25 25
25 25
25
25 25
100 75
75 100
75 75
75 100
2. 1
2 6
10 12
15 17
20 23
24 27
30 Identifying Specific
Information 25
25 25
25
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25
25 25
25 25
25 25
25 25
25
25 25
25 25
25 25
25
25 25
25 25
25 100
75 100
75 75
100 75
100 100
100 50
100
3. 4
22 26
31 39
45 46
48 Inference
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25
25 25
25 25
25 25
25 100
100 75
75 75
100 50
75 4.
3 8
13 18
25 28
34 37
38 42
50 Reference
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25
25 25
25 25
25 25
25 25
25 25
25 25
100 75
100 75
100 75
100 100
100 50
100 5.
5 7
11 14
19 29
32 35
40 43
49 Vocabulary
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25
25 25
25 25
25 25
25
25 25
25 25
25 25
25 25
25 25
25 25
25 25
25 75
100 75
100 75
100 75
100 75
100 100
3.5.2. Reliability
1. Reliability of Questionnaire
Since the LLSQ was a questionnaire for language learning strategies that had been developed using a Likert scale, a Cronbach
’s Alpha was used to measure the internal consistency of the items of the questionnaire. The reliability of the LLSQ
was determined for each individual category of language learning strategy. With 79 participants from an Indonesian university, the Cronbach
’s Alpha of sub-scales of the LLSQ was 0.89, 0.82, and 0.75 for metacognitive, deep-level, and surface-
level categories respectively. To find whether the test was reliable or not for the first grade students of SMAN 14 Bandar Lampung, the researcher used
Cronbach ’s Alpha. The alpha ranges between 0 and 1. The higher alpha, the more
reliable the items of the questionnaire will be Setiyadi, 2006:167.
Explanation:
r = reliability
n = the number of item
∑σi
2
= total variance of all items σi
2
= the total variance
To find out the variance, the researcher uses the formula as follow: σ =
Explanation: σ
= variance ∑X
2
= the total square of the number of data ∑X
2
= square of the total number of data N
= the number of data And for knowing the classification of reliability are as follows:
a. Between 0.800 to 1.00 = very high reliability
b. Between 0.600 to 0.800 = high reliability
c. Between 0.400 to 0.600 = moderate reliability
d. Between 0.200 to 0.400 = low reliability
e. Between 0.000 to 0.200 = very low reliability
∑X
2
- ∑X
2
N N
n n - 1
r = 1 -
∑σi
2
σi
2
2. Reliability of Reading Test
Reliability refers to whether the test is consistent in its score and gives us an indication of how accurate the test score are Shohamy, 1985: 70. A test is would
be reliable if the score gained by the examiners is constant whenever and by whomever the test is conducted. The reliability of the reading comprehension test
is measured based on Pearson Product Moment which examines the correlation coefficient of reliability between odd and even number reliability of the half test.
The formula can be seen as follows:
r
xy =
√ Explanation:
r
xy
= coefficient reliability between odd and even number x
= odd number y
= even number ∑x
2
= total score of odd number ∑y
2
= total score of even number ∑xy
= total score of odd and even number After the reliability of the half test is calculated, the researcher uses Spearman
Brown’s Prophecy formula to measure the reliability of the test as a whole as follows:
Explanation:
r
k
= the reliability of the whole test r
xy
=
the reliability of half test Hatch and Farhady, 1982: 247
r
k
=
2
r
xy
1
+ r
xy
The criteria of the reliability are: 0.90
– 1.00 = high 0.50
– 0.89 = moderate 0.00
– 0.49 = low 3.5.2.
Scoring System The researcher used
Arikunto’s formula in scoring the students’ result of the test. The higher score would be 100
S = x 100
Explanation: S: Score of the test
R: Right answer N: Total of the items