STRUCTURAL EQUATION MODELING (SEM) : CATEGORY DATA (Studies in Science Literacy and Ability Test for Seventh Grade Student Secondary School)
Ability is a construct that has value but can not be measured directly or latent
variables. Evaluation of ability to also be done on the measuring tool that is a
item. Quality of items can be known by item difficulty and item discrimination that
are parameters on item response theory (IRT). While relationship between test
result and ability expressed in a curve called item characteristic curve (ICC). If
ability is influenced other factors in this study are school accreditation and
gender, analysis of ability model can be use multiple indicators multiple causes
(MIMIC) model. In this research, latent variables were observed through the
indicator variables with category data so structural equation model (SEM) for
categorical data method can be used for analysis. While the aim of this study is
to determine the best model through estimation and comparison of parameters in
IRT and MIMIC models. The result of data analysis are significant difference
between IRT with MIMIC models. Evaluation of model results that IRT models is
the best model for the estimation of parameters showed better result and the value
of X2 and df is smaller than the MIMIC model. From the item characteristic
curve is known about the item number three has a good quality because it can
distinguish the ability of participants is better than any other items. From MIMIC
model can be seen that for each school accreditation, the ability of female and
male participants almost same although female participants was higher than
male participants and for the school with low accreditation then the ability is
lower.
No copy data
No other version available