Data Description FINDINGS AND DISCUSSION

17 From the graphics 4.2, it shows that the pre-test score in control class is still varied after the treatment. Thus the treatment has affected the whole class not only several students. The students who get lower score in pre-test improved their score from previously 34 to 56. In this research, there were 32 participants in experiment class with the mean score of pre-test is 40.375, while the pre-test is 62.5625. Moreover, the gained score in experiment class was 22.1875. The wide range of the gained score increase in experimental class before and after giving the treatment can be concluded as significant. Meanwh ile, there were 32 participants in control class where the mean score of pre-test reach 37.875, while the post-test score was 56.125. the score gained from pre-test to post-test not as significant as experimental class, only 18.25 points. As previously mentioned in chapter III, in analyzing the data, the writer uses statistic calculation of the t-test formula with the degree of significance 5. Table 4.2 Standard Deviation Table Students X1 Gained Score X2 Gained Score X 1 X 2 X 1 2 X 2 2 1. 16 6 -6.2 -12.2 38.44 148.84 2. 26 8 3.8 -10.2 14.44 104.04 3. 20 -2.2 -18.2 4.84 331.24 4. 28 18 5.8 -0.2 33.64 0.04 5. 22 4 -0.2 -14.2 0.04 201.64 6. 20 14 -2.2 -4.2 4.84 17.64 7. 30 32 7.8 13.8 60.84 190.44 8. 36 6 13.8 -12.2 190.44 148.84 9. 10 38 -12.2 19.8 148.84 392.04 10. 26 16 3.8 -2.2 14.44 4.84 11. 32 4 9.8 -14.2 96.04 201.64 12. 30 22 7.8 3.8 60.84 14.44 18 13. 8 60 -14.2 41.8 201.64 1747.24 14. 14 14 -8.2 -4.2 67.24 17.64 15. 10 16 -12.2 -2.2 148.84 4.84 16. 36 24 13.8 5.8 190.44 33.64 17. 22 -0.2 -18.2 0.04 331.24 18. 2 4 -20.2 -14.2 408.04 201.64 19. 30 30 7.8 11.8 60.84 139.24 20. 12 2 -10.2 -16.2 104.04 262.44 21. 20 40 -2.2 21.8 4.84 475.24 22. 36 28 13.8 9.8 190.44 96.04 23. 24 32 1.8 13.8 3.24 190.44 24. 16 22 -6.2 3.8 38.44 14.44 25. 40 4 17.8 -14.2 316.84 201.64 26. 14 34 -8.2 15.8 67.24 249.64 27. 10 36 -12.2 17.8 148.84 316.84 28. 20 40 -2.2 21.8 4.84 475.24 29. 20 30 -2.2 11.8 4.84 139.24 30. 36 4 13.8 -14.2 190.44 201.64 31. 32 -4 9.8 -22.2 96.04 492.84 32. 12 -10.2 -18.2 104.04 331.24 N=32 ∑X1 = 710 ∑X2 = 584 ∑X1 = - 0.4 ∑X2 = 1.6 ∑ X12 = 3018.88 ∑ X22 = 7678.08 The writer used t-test formula to find out the effectiveness of the teaching writing by using peer feedback as follows: M1 = = = 22.1875 M2 = = = 18.25 19 t o = t o = t o = t o = t o = t o = t o = 2.857 Determining t-table in signific ance level 5 with dƒ: dƒ = N1 + N2 -2 = 32 + 32 – 2 = 62 The value of t table is │1.67│. From the statistic calculation result, it can be seen that the value of t o is 2.857 and the degree of freedom in the table of significance dƒ is 62.

B. Data Analysis

As mentioned in chapter one, this research is conducted in order to know whether teaching writing using peer feedback at SMA N 11 Kota Tangerang Selatan is more effective than using teacher feedback. 20 To answer the questions above, the writer made a null hypothesis and alternative hypothesis. According to Anas Sudjiono, if t o t t , the null hypothesis H o is rejected, on the contrary the alternative hypothesis H a is accepted. It means that between variable x and y is significance. Meanwhile, if t o t t , the null hypothesis is accepted and the alternative hypothesis is rejected. It means that there is no significance increase between variable x and y. By comparing the value of t o = 2,857 and t table on the degree of significance 5 = 1.67, the writer concludes that t o is higher than t table , 2,857 1,67 which means that the alternative hypothesis is accepted and the null hypothesis is rejected. In conclusion, teaching writing by using peer feedback is more effective to increase students’ writing achievement than teaching writing by using teacher feedback.

C. Discussion

As explained in the first chapter, the purpose of this study is to find out whether peer feedback can improve students’ writing achievement. In this section, the writer tries to discuss the findings of the research from the result of pre and posttest of experiment and control group. Before analyzing the hypothesis, the researcher calculated normality to find the data has normal distribution. Whenever the data has normal distribution, it can be calculated by using statistical parametric where the data is assumed as valid as it counted by refering to several parameters. After calculating the normality test, the writer counts the mean in each experiment and control group. The result shows that mean of experiment group was higher than control group. Table 4.3 Normality Test Result of Experiment Class Normality Test Result of Experiment Class Tests of Normality Kolmogorov-Smirnov a Shapiro-Wilk Statistic df Sig. Statistic df Sig. .123 32 .200 .973 32 .590 . This is a lower bound of the true significance. a. Lilliefors Significance Correction 21 Table 4.4 Normality Test Result of Control Class Normality Test Result of Control Class Tests of Normality Kolmogorov-Smirnov a Shapiro-Wilk Statistic df Sig. Statistic df Sig. preex .079 32 .200 .968 32 .438 . This is a lower bound of the true significance. a. Lilliefors Significance Correction The table 4.3 shows the result of normality test in experiment and control class. The result of normality test reach 0.590 or higher than the table of 0.05. Meanwhile, the table 4.4 shows the result of normality test in control class which reach 0.438 or higher than the table of 0.05. It can be concluded that the data used in experiment and control class are normal and assessed as valid to be used in this research. After variance of experiment and control group was found, the writer calculates homogenenity test to determined t-test formula. Based on the calculation of homogenity test, it could be seen that the test is homogeneous in both classes. The result of the homogenous test could be seen in the table below. Table 4.5 Homogenity Test Result of Experiment Class Test of Homogeneity of Variances preex Levene Statistic df1 df2 Sig. 1.804 7 14 .165 Table 4.6 Homogenity Test Result of Control Class Test of Homogeneity of Variances preex Levene Statistic df1 df2 Sig. 2.316 8 14 .081