DEVELOPING A RUBRIC FOR PORTFOLIO ASSESSMENT IN WRITING OF GRADE VIII STUDENTS AT SMP NEGERI 15 YOGYAKARTA

The objective of this research is to develop a rubric for portfolio assessment in writing of grade VIII students at SMP Negeri 15 Yogyakarta. This research applied a Research and Development study which adopts the model of R & D proposed by Dick and Carey in Gall, Gall and Borg (2003). The data collected were qualitative and quantitative. The qualitative data were in the form of interview transcript and the quantitative data were in the form of students’ writing scores. These data were used to evaluate and then to revise the rubric. On the basis of some theories related to the development of effective scoring rubrics and the result of the needs analysis, the first draft of the rubric was then developed. The researcher consulted the rubric to a writing expert to know whether the rubric meets criteria which are appropriate to measure the students’ writing. After the rubric was implemented, an evaluation was then conducted. The researcher conducted interviews with the English teachers who participated in the implementation to find out the strengths and weaknesses of the rubric and to ask the teachers’ suggestion concerning the rubric. The data of the interview were used to revise the rubric. The research was conducted through six steps, they are: researching and collecting the information, planning, developing the rubric, obtaining expert judgment, field testing and developing the final product of the rubric. The instrument used in collecting the data was interviews and questionnaires to cross check the result of the interviews. The finding shows that there are four dimensions of portfolio assessment in the rubric: (1) characteristics of the writer, (2) characteristics of the portfolio as a whole, (3) characteristics of individual texts, and (4) intratextual features. Each dimension is written in three properties of the rubric: criteria, levels and descriptors. Each dimension has criteria which indicate good performance on a task. Each criterion is graded in five levels of performances, and the descriptors tell precisely what the performance looks like at each level