Use and application of structural models in dental education research

The notion of prediction implies causation. Path and structural models explore the causal links rather than mere empirical relationships between variables. This technique involves a breakdown of correlations; it differs from correlation and regression methods in that it provides relevant information in the presence of important but unobserved (latent) explanatory variables and of measurement errors in the data. It also allows for more than one regression analysis simultaneously and affords inference through tests of the model. In this study, latent abilities of dental students were analyzed as causes and professional achievements as effects, with preadmission performances as indicators of latent abilities. A model with three constructs is consistent with the observed data. The results demonstrate that whereas correlation analysis presents limitations in interpretation, structural analysis focuses clearly on the direct impact of the quality of dental school education, rather than preadmission background, on clinical performance and board success as measures of future professional performance.