Table 4.2 The Summary Table of Complexity in Controlled Class
In order to provide more detailed data of result in this study, the summary box plot below contained students’ pre-test and post test score in
experimental class and controlled class for complexity.
Table 4.3 Box Plot of Pre Test of Complexity
Number 1 was experimental class and number 2 was controlled class.
Controlled Class Pre-test
Post-test Mean
6,70 11,00
Median 6,50
10,50
Modus
6 17
Minimum
3 4
Maximum
11 20
Sum
135 228
Table 4.4 Box Plot of Post Test of Complexity
Number 1 was experimental class and number 2 was controlled class.
2.
Accuracy
The accuracy in writing will be scored by calculating the number of grammatical errors per total number of t-units.
2
Then the researcher counted the number of grammar errors in 100 words. The example below
how the writer counted the number of grammar error to get the accuracy score;
Now that the semester is over, I’m going to rest a few days and
then take a trip. This is counted as one grammar. Because I felt so sleepy, I went to bed earlier. This is counted
as one grammar. I went swimming although the weather is cold
.
This is counted as one grammar.
2
Rod Ellis, The Differential Effects of Three Types of Task Planning on the Fluency, Complexity, and Accuracy in L2 Oral Production: Applied Linguistic, 30, 2009, pp. 1-36
a. Accuracy in Experimental Class
The experimental class in this research was a class of students in XI IPA A at SMA Daar El-Qolam Tangerang. This class consisted of
20 students. Based on the test given to experimental class, the accuracy pre-test
mean score X was 4.80. In addition, the lowest score in pre-test was 0, while the highest score in pre-test was 9. After the treatment, which
was using blog in writing, mean score X of students was increased to 15.75. In addition, the lowest score in post-test was 2, while the
highest score in post- test was 26. The finding showed that students’
post test scores were higher than students’ pre-test score. The range
between pre-test mean score and post-test mean score was 10,95
points. It meant that the increasing score between pre-test and post-test for complexity was quiet significant. The pre-test and the post test
score of the students can be seen in table 4.5
Table 4.5 The Summary Table of Accuracy in Experimental Class Experimental Class
Pre-test Post test
Mean 4,80
15,75
Median 5,00
15,50
Modus
5 15
Minimum 2
Maximum 9
26
Sum
96 315
b. Accuracy in Controlled Class
The controlled class in this research was a class of students in XI IPA B at SMA Daar El-Qolam Tangerang. This class consisted of 20
students.
Based on the test given to controlled class, the accuracy of pre-test mean score X was 7,40. In addition, the lowest score in pre-test was
4, while the highest score in pre-test was 17. In controlled class, even though students learned to write without using blog, mean score X of
students was increased to 12.05. The lowest score in post-test was 5, while the highest score in post-test was 25. The finding showed that
students’ post test scores were higher than students’ pre-test score. The range between pre-test mean score and post-test mean score was 4.65
points. It meant that the increasing point between pre-test and post-test was not significant and the increasing point of controlled class is lower
than the increasing point of experimental class. The pre-test and the post test score of the students can be seen in table 4.6
Table 4.6 The Summary Table of Accuracy Pre-test and Post test
In order to provide more detailed data of result in this study, the summary box plot below contained students’ pre-test and post test score in
experimental class and controlled class for accuracy.
Controlled Class Pre-test
Post-test Mean
7,40 12,05
Median
7,50 11,00
Modus
5 9
Minimum
4 5
Maximum
17 25
Sum 148
241
Table 4.7 Box Plot of Pre Test of Accuracy
Number 1 was experimental class and number 2 was controlled class.
Table 4.8 Box Plot of Post Test of Accuracy
Number 1 was experimental class and number 2 was controlled class.
Table above showed that both class increased their score of complexity and accuracy. However, the score in experimental class increased significantly
compared to score in controlled class.
B. Analysis of Data
In analysis of data, there were two things that were needed to analyze before calculating statistical hypothesis; test of homogeneity and test of
normality.
1. Test of Normality
Test of normality was calculated in both of pre-test and post-test. These tests were conducted through SPSS 20. Test of normality was
calculated using Kolmogorov-Smirnov and Saphiro-Wilk. The following tables contained the result of test of normality of experimental class.
Table 4.9 The Result of Normality Test of Pre-Test of Complexity
Test of Normality
class Kolmogorov-Smirnov
a
Shapiro-Wilk Statistic
df Sig.
Statistic Df
Sig. score
1 .167
20 .146
.926 20
.130 2
.128 20
.200 .946
20 .312
Table 4.10 The Result of Normality Test of Pre-Test of Accuracy
Test of Normality
Class Kolmogorov-Smirnov
a
Shapiro-Wilk Statistic
df Sig.
Statistic df
Sig. score
1 .129
20 .200
.953 20
.410 2
.156 20
.200 .928
20 .139
In test of normality, data are stated as distributed normal when sig. score was above 0.05. In the table above, number 1 was a symbol of experimental class,
while number 2 was a symbol of controlled class. In Kolmogorov-Smirnov
column, it showed that both experimental class and controlled class had normal distribution data. The sig. score of experimental class for complexity was 0.146,
while the sig. score of controlled class was 0.200. Meanwhile the sig. score of experimental class for accuracy was 0.200, while the sig. score of controlled class
was 0.200. In Shapiro-Wilk column, it showed that both experimental class and
controlled class also had normal distribution data. The sig. score of experimental class for accuracy complexity was 0.130, while the sig. score of controlled class
was 0.312. Meanwhile, the sig. score of controlled class for accuracy was 0.410, while the sig. score of controlled class was 0.139
Thus, it could be concluded that pre-test score in both experimental class and controlled class had normal distribution data calculated by Kolmogorov-
Smirnov and Shapiro-Wilk.
Table 4.11 The Result of Normality Test of Post Test of Complexity
Tests of Normality
Class Kolmogorov-Smirnov
a
Shapiro-Wilk Statistic
df Sig.
Statistic df
Sig. Score
1 .088
20 .200
.959 20
.524 2
.111 20
.200 .968
20 .707
Table 4.12 The Result of Normality Test of Post Test of Accuracy
Tests of Normality
Class Kolmogorov-Smirnov
a
Shapiro-Wilk Statistic
df Sig.
Statistic df
Sig. Score
1 .105
20 .200
.969 20
.742 2
.177 20
.101 .920
20 .101
In test of normality, data are stated as distributed normal when sig. score was above 0.05. In the table above, number 1 was a symbol of experimental class,
while number 2 was a symbol of controlled class. In Kolmogorov-Smirnov column, it showed that both experimental class and controlled class had normal
distribution data. The sig. score of complexity of experimental class was 0.200,
while the sig. score of controlled class was 0.200. Meanwhile the sig. score of experimental class for accuracy was 0.200, while the sig. score of controlled class
was 0.101. In Shapiro-Wilk column, it showed that both experimental class and
controlled class also had normal distribution data. The sig. score of complexity of experimental class was 0.527, while the sig. score of controlled class was 0.707.
Meanwhile the sig. score of experimental class for accuracy was 0.742, while the sig. score of controlled class was 0.101. Thus, it could be concluded that post test
score in both experimental class and controlled class had normal distribution data calculated by Kolmogorov-Smirnov and Shapiro-Wilk.
2. Test of homogeneity
Test of homogeneity of complexity and accuracy was calculated in both of pre-test and post-test. These tests were conducted through SPSS
20. Test of homogeneity was calculated using Levine. The following tables contained the result of test of homogeneity of experimental class.
Table 4.13 The Result of Homogeneity Test of Pre-Test of Complexity
Test of Homogeneity of Variance
Levene Statistic df1
df2 Sig.
Score Based on Mean
3.566 1
38 .067
Based on Median 2.135
1 38
.152 Based on Median and with
adjusted df 2.135
1 35.879
.153 Based on trimmed mean
3.531 1
38 .068
Table 4.14 The Result of Homogeneity Test of Pre-Test of Accuracy
Levene Statistic df1
df2 Sig.
Score Based on Mean
.539 1
38 .467
Based on Median .398
1 38
.532 Based on Median and with
adjusted df .398
1 37.901
.532 Based on trimmed mean
.513 1
38 .478