Widinda Normalia Arlianty, Farida Agus Setiawati
Data scaling plays an important role in analyzing research data. Some data analysis processes mostly require data processing using interval or metric scales. If the research data obtained still has an ordinal scale, nonparametric data analysis should be used. However, many studies want the data obtained using an interval scale because it is considered that the data is normally distributed. In addition, a good instrument is an instrument that has high validity and reliability. This study aims to evaluate the results of rescaling data using MSI and MSRS and compare their characteristics in terms of validity and reliability values. Data sourced from TIMSS 2015. The TIMSS data used is data obtained based on questionnaire instruments given to teachers. The questionnaire used a Likert scale with 4 answer options. The approach taken in this study is to use descriptive quantitative which in its implementation consists of two interrelated parts, namely scaling on secondary data, and comparison of scaled data characteristics using MSI and MSRS. Data analysis was conducted using Microsoft excel and R studio. Microsoft excel was used to scale the MSI and MSRS, while the R studio program was used to calculate validity through Confirmatory Factor Analysis (CFA) and reliability. The results obtained are that there is a change in the raw data score (RD) scaled using MSI with MSRS on the teacher questionnaire instrument. Furthermore, the characteristics of the data obtained based on Confirmatory Factor Analysis (CFA) show that the construct validity of the instrument based on the loading factor value is met and valid, both raw data (RD) and data scaled using MSI and MSRS. However, the scaling did not show any change in reliability values. The results generally showed that there was no significant difference in the CFA results of the Teacher Questionnaire Instrument data before and after transformation using the MSI and MSRS methods. This study contributes to the implementation of quantitative data scaling practices in measurement research and should the data transformation (scaling) process can improve accuracy in measurement. © 2026 Author(s).
Chemistry Education Study Program, Universitas Islam Indonesia, Yogyakarta, Indonesia; Universitas Negeri Yogyakarta, Yogyakarta, Indonesia
