Hypothesis Testing Data Analysis

46 Daud Yusuf, 2014 The Use of Scaffolding in Teaching Writing Universitas Pendidikan Indonesia | repository.upi.edu | perpustakaan.upi.edu

3.6.2 Hypothesis Testing

Hypothesis testing involved the analysis of data on pre-test and post-test. Pre-test was administered in the beginning session of teaching program, while post-test was administered in the end session of teaching program. Both pre-test and post- test were an essay composition assignment, in which students were asked to write a recount text based on their experience. Furthermore, the first step in analysing the pre-test and post-test data was analysing the normality of distribution. In order to conduct a parametric test, both pre-test and post-test data had to meet the assumption of normal distribution. One sample Kolsmogorov Smirnov of non-parametric test was used to test the normality of distribution. The calculation was done using IBM SPSS Statistics 20 see Appendix C for the result. The data was normal if the significant value of one sample Kolsmogorov Smirnov test were higher than the level of confidence, that is, 0.05. The second step in analysing the pre-test and post-test data was conducting the paired-sample t-test, in case the data met the assumption of normal distribution. Paired sample t- test was used to measure the difference of students’ writing performance. Moreover, paired-sample t-test was used because only one group was involved on the intervention. The calculation was done using IBM SPSS Statistics 20, and presented in chapter four. In case the difference was significant, effect size, thus, was calculated to measure the impact of the intervention to the treatment. Effect size used was r 2 coefficient Coolidge, 2000. SPSS did not calculate effect size automatically. Therefore, effect size was calculated using the equation as follows. Figure 3.1 r square equation Coolidge, 2000 47 Daud Yusuf, 2014 The Use of Scaffolding in Teaching Writing Universitas Pendidikan Indonesia | repository.upi.edu | perpustakaan.upi.edu where, r 2 = effect size = obtained t value = degree of freedom Likewise, the coefficient is interpreted as following table. Table 3.5 r square Coefficient Interpretation Coolidge, 2000 Effect Size Value Small .01 Medium .09 Large .25

3.6.3 Analysis of Student Texts