Psychometric properties of Mental Health Scale: An Item Response Theory Approach

Abstract

Indonesia Family Life Survey (IFLS) the fifth wave used the Centre for Epidemiological Studies Scale (CES-D10) to measure mental health state, which was depressive symptoms construct. It was primarily used in some studies regarding depressive symptoms. However, there was no specific validation research in the Indonesian context. Thus, this study aimed to (1) investigate the item quality of depressive symptom measurement from IFLS fifth wave, (2) measurement precision of this scale, and (3) the measurement invariance based on gender using the IRT approach. This study used data from ILFS-5 in the KP ("Keadaan Psikologis" or Mental Health) section, which was CES-D10 consisting of 10 items and conduct IRT analysis using a 2-PL model. The number of participants from the data was 31,447. CFA analysis resulted a unidimensional model fit for this scale. Moreover, finding showed that this scale had good psychometric properties, including item discrimination, item location, and item fit, except item 5 and item 8. Despite the poor quality for those items, the reliability coefficient, including the items, met a reliable measurement criterion. Also, this scale had much information for assessing medium and severe depressive symptoms. Moreover, the Differential Item Functioning (DIF) analysis indicated that there was no item exhibiting DIF.