We use the generalized maximum entropy (GME) estimator to take into account the measurement error in the regression model with a composite indicator, Likert-type scales based, as explanatory variable. We show that, the reliability measure of the observed composite indicator can be used to define an estimator of the error variance and the supports required by the GME approach. As well as to obtain an estimate of the slope parameter of the model, that has statistical properties similar to the classical ordinary least squares adjusted for attenuation estimator, GME approach allows to estimate the measurement error that can be used to adjust the composite indicator of the latent explanatory variable. An extensive simulation and two case studies show the usefulness of this approach.
The GME estimator for the regression model with a composite indicator as explanatory variable
CARPITA, Maurizio
2015-01-01
Abstract
We use the generalized maximum entropy (GME) estimator to take into account the measurement error in the regression model with a composite indicator, Likert-type scales based, as explanatory variable. We show that, the reliability measure of the observed composite indicator can be used to define an estimator of the error variance and the supports required by the GME approach. As well as to obtain an estimate of the slope parameter of the model, that has statistical properties similar to the classical ordinary least squares adjusted for attenuation estimator, GME approach allows to estimate the measurement error that can be used to adjust the composite indicator of the latent explanatory variable. An extensive simulation and two case studies show the usefulness of this approach.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
