Abstract
The aim of this paper is to pursue predictors contributing to English writing ability in the high school setting, which is expected to be contributory to further research of English language teaching. In this paper, the data obtained were computed and multiple regression analysis was conducted with the data results of two types of writing task (dependent variables), and the linguistic tests plus metacognitive and affective factors (independent variables). Furthermore, using each factor as a dependent variable, multiple regression analysis was conducted repeatedly in order to identify the relationships among all the writing factors. Based on the data results, the path models were produced using Structural Equation Modeling (SEM) and the best model for writing ability was selected according to fit statistics. As a result, the primary common writing variables, such as 'Cloze', 'Motivation', and diverse metacognitive factors turned out to emerge; furthermore, various factors seem to be intertwined directly or indirectly in the Japanese high school setting.
