next up previous
Next: Results Up: Structural Equation Modeling Using Previous: Statistical Measures

Structural Equations

Structural equation modeling requires independent observations. Moreover, the mean as well as the variance contain sampling error. Therefore, two independent estimates for the mean were obtained, as well as two independent estimates for the variance. The assumption was made that the sampling error of the minimum reaction time could be neglected. The following observed variables were used for EQS:
\begin{align*}V_1 & = \mbox{min(RT01 to RT06) (estimator of the parameter} A)\\ ...
...f} \ln \mbox{Var}(T))\\
V_6 & = \mbox{age in years (range: 6-18})
\end{align*}
The following latent variables were used
\begin{align*}F_1 & = \ln \lambda\\
F_2 & = \ln \delta\\
E_2 & = \mbox{error...
...V_3\\
E_4 & = \mbox{error in} V_4\\
E_5 & = \mbox{error in} V_5
\end{align*}
The following equations were included in the model:
\begin{align*}V_1 & = E_1\\
V_2 & = V_1 + F_1 - F_2 + E_2\\
V_3 & = V_1 + F_...
...F_1 - 2F_2 + E_4\\
V_5 & = V_1 + F_1 - 2F_2 + E_5\\
V_6 & = E_6
\end{align*}
with the following constraints: all the variances of the error components equal to each other and all covariances between the error components equal to null.



AHGS VAN DER VEN
2002-01-14