Reza Taufik Maulana, 2014 NUMBERED HEADS TOGETHER NHT: AN ENDEAVOUR TO IMPROVE STUDE
NT’S SCIENTIFIC CREATIVITY AND MASTERY CONCEPT IN LEARNING GLOBAL WARMING
Universitas Pendidikan Indonesia
| repository.upi.edu
| perpustakaan.upi.edu
The conversion of raw score into percentage is conducted to analyze the rubrics. Further, the result of percentage can be classified into several categories. The
technique of converting score into percentage is using formula written Sejati, 2013:
Score =
� � � �� � � � �
x 100
E. Instrument Development
Thedevelopment process of instrument begins with of the curriculumanalysis that is applied inschool. The researcher then formulates the questions to be used as an
instrument of pretest and post-test. The tests used in this research are writing tests consist of 20 multiple choice questions and four essay questions.
The instrument needs to be consulted judgment by the concerned lecturer and some experts in related fields. After being judged, the instrument which is not
appropriate enough should be revised. After the instrument revised, it should be tried out on another class which had learned the topic before. Based on the test results, the
instrument questions will be analyzed with the following requirements:
1. Instrument Test Requirements
a. Validity
Validity can be describedas the ability of an Instrument to measure what it is designed to measure. Kumar, 2005 quoted by Sejati 2013 Anderson
as quoted in Arikunto, 2010 quoted by Sejati 2013 revealed that “A test
is valid if it measure what it purpose to measure”. The results of validity
test shows that the data collected from the instrument is not deviating from the idea. Hence, by having this validity test it can measure whether the data
that is produced from the test is valid with the variable that want to be measured and interpreted
Reza Taufik Maulana, 2014 NUMBERED HEADS TOGETHER NHT: AN ENDEAVOUR TO IMPROVE STUDE
NT’S SCIENTIFIC CREATIVITY AND MASTERY CONCEPT IN LEARNING GLOBAL WARMING
Universitas Pendidikan Indonesia
| repository.upi.edu
| perpustakaan.upi.edu
The researcher use the Coefficient of Product Moment Karl Pearson to measure the validity of each test item, there is:
= �
− �
2
−
2
�
2
−
2
Sejati, 2013 With,
: correlation coefficient between x and y variable n : amount of student
x : total score in test item y : total score of student
Interpretation about will be divided into different categories based
on Guilford quoted by Arikunto 2010 cited by Sejati 2013. Table 3.11 Classification validity coefficient
Value �
Interpretation
0,90 ≤� ≤ 1,00 Very high validity
0,70 ≤� 0,90 High validity
0,40 ≤� 0,70 Medium validity
0,20 ≤� 0,40 Low validity
0,00 ≤� 0,20 Very low validity
� 0,00 Invalid
b. Reliability
Anderson as quoted by Arikunto, 2010 quoted in Sejati 2013 revealed that validity and reliability are two important factors in developing
test instrument, “A reliable measure in one that provides consistent and
stable indication of the characteristic being investigated”. To be said as a reliable, the research instrument has to be stable and consistent, which
Reza Taufik Maulana, 2014 NUMBERED HEADS TOGETHER NHT: AN ENDEAVOUR TO IMPROVE STUDE
NT’S SCIENTIFIC CREATIVITY AND MASTERY CONCEPT IN LEARNING GLOBAL WARMING
Universitas Pendidikan Indonesia
| repository.upi.edu
| perpustakaan.upi.edu
eventually leads to other characteristics such as predictable and accurate and eventually conclude to be reliable. In another word reliability can be
described that the test will resulted the similar results even it is done by different people, times and places. Hence, it can be concluded that the more
reliable the research instrument are influenced by greater degree of consistency and stability. Arikunto, 2006 quoted by Sejati 2013.
The value of reliability is determined based on coefficient value which is gained by Alpha formula, as follows:
11
r
=
2 2
1 1
t i
s s
n n
Arikunto2006 quoted by Sejati 2013
Table 3.12 Classification of Reliability Coefficient
Value
11
r
Interpretation
0,90 ≤� ≤ 1,00 Very high reliability degree
0,70 ≤� 0,90 High reliability degree
0,40 ≤� 0,70 Medium reliability degree
0,20 ≤� 0,40 low reliability degree
� 0,20 Very low reliability degree
Explanation:
11
r
: reliability coefficient
2 i
s
: score variant each test item n : amount of test item
2
∶ total score variant
Reza Taufik Maulana, 2014 NUMBERED HEADS TOGETHER NHT: AN ENDEAVOUR TO IMPROVE STUDE
NT’S SCIENTIFIC CREATIVITY AND MASTERY CONCEPT IN LEARNING GLOBAL WARMING
Universitas Pendidikan Indonesia
| repository.upi.edu
| perpustakaan.upi.edu
c. Discriminating Power
