Theoretical Framework THEORETICAL REVIEW

28 applied to those in treatment group, while the control group got ordinary teaching- learning activities. The dependent variable is the variable that will be affected by the independent variable Fraenkel Wallen, 2006. The dependent variable in this research was students’ writing skill. The particular feature of quasi-experimental design is the lack of random assignment Gal, Joyce Borg, 2007. Actually, the function of random assignment is to ensure the balance in both groups. Eliminating it means the researcher cannot randomly put the participants into the control and treatment group. In the other words, the participants were already grouped. Quasi-experimental research needs two groups to compare: the control group and the experimental group Ary, Jacobs, Sorenson, Razavieh, 2010. Treatment group is the group that receives the special treatment, whereas control group is the group that receives either no or alternate treatment Gal, Joyce, Borg, 2007. Between these two groups, the one that was observed was actually the experimental group. It was because the experimental group received the special treatment. However, the control group was also important. This group provided the data for comparison. Later on, this data was calculated to determine the significance of the test result. To see the significant difference between two groups, the researcher compared the mean of the pre-test level of phenomenon before treatment and the post-test level of phenomenon after treatment Kothari, 2004. The pre-test was given to both groups before the students got the treatment. The purpose was to measure the students’ writing skill. After the students got the treatment, they had 29 post-test. From the result of the post-test, the researcher analyzed and compared the mean difference. In the end, the researcher was able to determine whether the mean difference found was significant. Figure 3.1. is the illustration of the research design quasi-experimental method: Figure 3.1. Pre-test and Post-test Control Group Design

3.2 Research Setting

To see the effect of GAIL in senior high school students’ writing skill, the quasi-experiment was conducted at SMA Kolese De Britto. Fraenkel and Wallen 2006 suggest that the sample represents the population. Therefore, this school was chosen. The students’ writing skill in this school is varied; there are some students who can write very well, some are good enough, and the others still need close guidance for their writing.

3.3 Research Population and Sample

In experimental research, there must be sample and population Ary, Jacobs, Sorenson, Razavieh, 2010. The population of this research was the students of SMA Kolese De Britto. For the sample, the researcher chose eleventh grader Science students. The sampling design applied here is cluster sampling. It requires small scale representation of the total population claimed as one sample which represents the universe Kothari, 2004. In determining the sample, the Experimental Group Control X 30 researcher chose two classes as the experimental group and another two classes as the control group. The experimental group consisted of 52 students and the control group consisted of 53 students. All of the group’s members had pre-test before they got the treatment and post-test after they got the treatment.

3.4 Instrument

The purpose of this study was to observe whether GAIL would improve SMA students’ writing skill. It could be seen from the different performance before and after the treatment Kothari, 2004. Therefore, the tests before pre-test and after they got the treatment post-test were necessary. In the other words, the researcher used writing tests APPENDIX 2 3 as the instruments to collect the data for this study. The students’ writings were then assessed by using a rubric APPENDIX 4 explained further in the next paragraph. This rubric is the second instrument used during the study. There are several things to consider in creating writing assessment. The first consideration is what scale that will be used. There are two kind of scales in language assessment: holistic scale, and analytical scale Hyland, 2003. Weigle 2002 points out two basic considerations for choosing the scales. The first one is whether the scale is made to assess single or general tasks and the second, whether each script will be given a single or multiple scores. In each scale, the score represents how well the students’ performance is. Holistic scale refers to comparing the students’ writing with the other students’ work in the class Sundem, 2007. Weigle 2002 explains the advantages and disadvantages of this scale. Holistic scale is preferred because it gives the most