Dampak Kesalahan Penetapan Indikator Efek dan Kausal terhadap Indeks Ketepatan dalam Pemodelan Persamaan Struktural

Abstract

Psychometrics theory has divided two types characteristics of indicators: effect indicator (reflective) and indicator causal (formative). In the structural equation modeling (SEM) approach, these differences were operated on different figure of analysis. Effect indicator is indicated by the direction of arrows from latent construct to its indicators, whereas the causal is vice versa. this study aimed to compare fit indices between models that accommodate these differences, and did not. Data analysis was conducted on secondary data from three studies that measured individual level of stress. One study involving causal indicators and three studies involving effect indicators. Two models, model with the causal and effect indicator were fitted to the data. The analysis suggests that both models reveal high fit indices and match with expected criteria. However, the model which accommodated the indicator characteristics reveals higher fit indices. This study recommends that researcher should accommodate the characteristics of construct indicators when into model in order to obtain higher fit indices.