ABSTRACT
S. M. Boker et al. introduced the multivariate latent differential equation approach, which is a model as well as an estimation procedure. The estimation procedure uses structural equation modeling, in particular the program Mx, to simultaneously estimate the parameters of the differential equation model and the measurement model. The contribution starts with a precise description of the oscillator in deterministic and stochastic form. To evaluate the effectiveness of the procedures in recovering parameter values of the oscillator, two Monte Carlo simulations were performed. The condition of all eigenvalues of A having nonpositive real part implies a stable movement. The program Linear Stochastic Differential Equations, written by Herman Singer, performs the maximum likelihood estimation of the model for panel data on the basis of the so-called exact discrete model (EDM). The EDM makes sure that the parameters estimated are exactly equal to the parameters of the underlying differential equation model.
