Structure equation modeling (SEM) is widely used to model complex data structures in the social sciences. Conventionally, aximum likelihood (ML) based iterative procedures are used to obtain SEM model parameter estimates. When sample size is small, such iterative procedures are either hard to converge or standard errors of the parameter estimates are inflated for converged solutions. To solve these problems, the ridge SEM has been proposed to improve both convergence and efficiency of parameter estimates.This study investigates a Bayesian method to deal with the same problem. We show the connection between the ridge SEM and our Bayesian procedure and conduct a simulation to compare their performance.