application of mediation model in health research: a simulation study

many health studies were interested to find the association between multiple predictors with a dependent variable. the popular method to analyze these studies was multiple regressions that assumed the predictors were independent. while in practice some predictors may depend on each other and they could not be presented in a regression model simultaneously, as a result the appropriate models such as mediation models must be used. the present simulation study was conducted to show the application of mediation model in health studies. the data were generated from multivariate normal distribution by statistical software r2.15.01. this simulation was according to a study that wished to find the association between knowledge, attitude and practice of oral and dental care among pregnant women. the mediation model was used to find the mediation role of attitude on practice. both knowledge and attitude were significantly related with practice among pregnant women. mediation model showed that the attitude was a mediator in the path way of association between knowledge and practice. according to the results of mediation analysis 64% of the effect of knowledge on practice was direct and the indirect effect was 36%. due to complex association between predictors the use of mediation model and path analysis, it is necessary to develop new health and medical studies.