d. Click AnalyzeScaleReliability Analysis e. Transfer the data from the left rectangular to the right rectangular
f. Click statistic g. Ceklist Scale if item deleted in Desciptive for option
h. Ceklist none in ANOVA Table i. Click Continue then O.K
After the steps have done, there was output. TheCorrected Item Total Correlation is represent r
hitung
for each questions. The validity and reability can be seen from r
table
and r
hitung.
If
.
r
hitung
r
table
, it means that the items mighht be used. The criteria of validity.
0.80r
xy
≤1.00 very high
0.60r
xy
≤0.80 high
0.40r
xy
≤0.60 enough
0.20r
xy
≤0.40 low
0.00r
xy
≤0.20 very low
r
xy
≤0.00 not valid
The writer also pay attention to reliability test. A reliable test is consitent, that is yields similar results over time with similar students in similar situation.
7
It also stated by Bartlett and Diana Burton that reliability describes the extent to
which a research instrument or method is repeatable.
8
It is clear that, the reliability test is necessary to be tested before the instruments will be given to the sample of
the study. The steps that must be done was same with analyze the validity. In output
of the analysis, it would be foung Reliability Statistics. In the left column, there was
Conbarch’s Alpha. It is an internal consistency score based on ekuavalen of each item that represent reliability. Both validity and reliability have the same
criteria.
7
JoAnne Schudt Caldwell, Comprehension Assessment, New York: The Guilford Press, 2008, p.30
8
Steve Bartlett and Diana Burton, Introduction to Education Studies, London: SAGE Published, 2007 p. 44
E. Preliminary Analysis
The writer conducted preliminary analysis that consist of normality and homogenity before analyzing te data with t-test.
1 Normality
The normality test is performed using Kolmogriv Smirnnov and Shapiro- Walk. It used to determine if the distribution of the data from sample is normal.
The test was done for both classess. The writer used IBM SPSS Statistics version 22 software. The formula used Kolmogrov
– Smirnov and Sapiro – Wilk. Here are the formulas:
1. Kolmogrov – Smirnov
D= sup {|F
n
z - Ф z |-∞≤z≤∞}
Where: F
n
= the empirical distribution function F
n
z = the sum up of z
k
≤zn Ф z = cumulatifdiatribution function
2. Shapiro – Wilk
Where: x
i
= the ordered random sample values a
i
= contants generated from the covariance, variance and means of the sample size n from a normally distributed sample.
9
The procedure to get normality analysis as follow: a. Open the IBM SPSS Statistics 22
b. Go to variable view and fill in the columns as follows. 1 Name Write
“class” in the first row. This is to indicate and differentiate between experimental and controlledlededed class.
write ”score” in the second row.
9
Stanislaus S. Uynto, P. hD, PedomanAnalisis Data dengan SPSS, Yogyakarta: GrahaIlmu, 2009, p.54
2 Type column was numeric 3 Width column is filled 8
4 Decimal change this row frow 2 to 0
5 Label column was left blank 6 Value column is none
7 Missing column is none 8 Column is 8
9 Align column is right
10 Measure coulumn is unknown 11 Role coulumn is filled with input
c. Click Data View, in score column, compute “1” as representing
experimental class and “2” representing controlledlededed class. d. In score column, compute score of each class.
e. Click AnalyzeDescriptive StatisticsExplore f. Fill in the Dependents List with score pret-test and then fill in the
Factor List with class. g. Click Plotschecklist Normality Polts with test, Histograms, Power
EstimationContinueO.K If p
– value α, the data is not distributed normal If p
– value ≥ α, the data is distributed normal Note: In SPSS, P
– value is same with Significance Sig.
10
2 Homogenity
Homogenity test is used to wether the data from two classes have same variant in order that the hypothesis can be tastes by t-test. The
procedure are same with normality’s steps. Furthermore, after testing both
normality and homogenity, the writer began test the hyphothesis.
F. Technique of Data Analysis
1. t-test
10
Ibid, p. 40