Page 80 - BUKU SYNOPSIS

using the Many-Facet Rasch Measurement (MFRM) to anticipate such

pitfall. In addition, McNamara (1996) suggested the MFRM is a

suitable tool for analyzing the results of performance tests such as the

writing skill test.

The MFRM is an expansion of the original Rasch model based on

two considerations. The former is that there is no restriction on the

analysis of only two aspects (such as examinees and items), and the

latter is that the analyzed data need not be dichotomous (Eckes, 2019).

In addition, this model provides a framework for calibrating writing

skill assessments (Engelhard, 1992). Linacre (1989) asserts that

although the recent extension of the Rasch Model can calibrate multiple

dimensions simultaneously, it must be examined separately. As the

MFRM is derived from the Rasch Model, it can predict the subjective

measurement framework that results from statistical invariance across

raters, writing tasks, and other aspects of the writing assessment

procedure. To achieve validity and reliability in score measurement, the

MFRM accommodates research in the LLA context, focusing on

assessing writing essays comprised of relevant factors, such as

examinees, raters, and criteria.

In this study, examinees will be evaluated by two raters on the

following domains: organization, content, language employed, and

mechanics, using a sequential scale category/rating scale ranging from

1 to 4 (lowest to highest score). Therefore, the data are polytomous

75