1. Testing Normality of the Data
Testing normality was used to check whether the population had normal distribution or not. The formula was as follow
11
:
= p-
ɸ
= value of p = sum value of data probability
ɸ = value of Kolmogorov table
To get the ɸ value, the standard score of the data z had to be calculated first
with formula: z =
̅
z = standard score x = students‘ score
̅ = the mean score s = standard deviation
After the value of was gotten, the value of normality table with
significance 5 was sought. After the value of normality table was found, it had to be compared with the value of
to find whether the data had a normal distribution or not. If the data had a normal distribution, the value of
would be same as or lower than the value of normality table. Conversely, if the data did
not have a normal distribution, the value of would be higher than the value
of normality table.
2. Testing Homogeneity of the Data
11
Budi Susetyo, Statistika untuk analysis data penelitian, Bandung: Refika Aditama, 2010, p. 148-150
Further, it was testing homogeneity. It was used to check whether the two populations had an equivalent variance or not. The formula was as follow
12
:
F=
Note: = High variance
= Small variance
To determine whether the data was homogeny or not, the following criteria of homogeneity were used:
The test is homogeny is accepted if F
The test is not homogeny is accepted if F≥ With dk N-1 and significance 5
If the data had fulfilled the requirement of normality and homogeneity, the mean score of the two data could be examined using t-test calculation as follow:
3. Testing T-Test for Two Independent Sample
T-test was used to examine the truth or false of the study hypotheses by comparing the value of
to . To get such value, the following calculations had
to be done:
13
=
̅ ̅
√
= t observation ̅
= mean score of experimental class ̅
= mean score of control class = variance of experimental class
12
Ibid, p. 160-161
13
Ibid., p. 202-205
= variance of control class = total students in experimental class
= total students in control class
After the value of t observation had been gotten, the value of t table with significance
5, , and degree of freedom
+ -2 had to be sought.
The value of t observation and t table were then compared to know whether CALL was effective in teaching past tense or not.
4. Testing the Effect Size of CALLCohen’s d
After the value of t-test was gotten, the effect size of CALL was then measured. In this case,
Cohen’s d formula was used to measure whether the effect size of CALL was strong or weak.The formula was as follow:
14
d = Mean for experimental class —Mean for control class Pooled standard
deviation Where
Pooled standard deviation = Standard devation of experimental class + Standard devation of control class 2
Then, the criteria below acted as a guidance to determine the effect size of CALL:
0-0.20 = weak effect
0.21-0.50 = modest effect
0.51-1.00 = moderate effect
1.00 = strong effect
G. The Statistical Hypothesis of the Study
14
Daniel Muijs, Doing Quantitatve Research in Education, London: Sage Publications, 2004, pp. 136
—137.