Page 126 - BUKU SYNOPSIS

Table 5. Criteria for Item and Person Reliability of Rasch Model

Category
< 0.94 Excellent
0.91 - 94 Very Good
0.81 – 0.90 Good
0.67 – 0.80 Enough
< 0.67 Low

Lastly, Andrich Threshold was applied to analyze whether the

description of each scale was clear and understood by the respondents.

2) Item Fit of Analytical Scoring Rubrics

Construct validity refers to the degree to which an instrument can

accurately measure the intended theoretical construct defined by the

examiner. The rating scale is considered the most definitive

expression of the evaluated construct when measuring students'

writing abilities (Knoch, 2009; McNamara, 1996; Weigle, 2002). This

study established the item fit of analytical scoring rubrics by utilizing

the Rasch Model, specifically the Many-Facet Rasch Measurement

(MFMR). The MFMR is advantageous for researchers since it

presents the students’ writing achievement clearly and it enables the

examination of instrument constructs using a substantial amount of

data, ranging from 27 to 61 respondents (Yudiarso, 2023). The

evaluation of the item fit of the instrument can be observed from outfit

MNSQ and ZSTD outfits and point measure correlation with the

121