Simple mediation analysis and its use in A medical research

The main aim of the researchers in any discipline is to test whether there is a relationship among in the variables. If there is any relationship, they are trying to examine the structure of this relationship. But, in practice, the relationship among factors influencing the outcome variables usually not sufficiently considered. This is defined as the "black box model" by Kurt Lewis. Accordingly, the independent variables produce an outcome variable after many unknown processes in the black box. Even this model demonstrated that output is experimentally associated with the independent variable, this model does not give enough information about how and why this relationship occurs. Of course, this is a considerable restriction. Mediation analysis examines how happen the effect of independent variable on a dependent variable. According to the Hayes, in mediation analysis, some treatment variables affect the outcome variable through one or more intermediate variables. Thus, the effect of treatment variable indicated by "X" on outcome variable indicated by "Y" can be checked in two ways. The first of these, X variable makes an "indirect" effect on the Y through the mediating variable "M" (X --> M --> Y)and the second is treatment variable X is a "direct" effect on the outcome variable Y (X --> Y). For this study, a sample was examined consisting of rheumatology patients with chronic pain for six months. Between the variables in the data set, taking into account the clinical theories, it was created a variety of simple mediation models. During modeling, considering all the possible combinations among types of variables, various simple mediation models have been developed. Each model is individually analyzed, and the obtained results were evaluated separately. When considering the limitations of traditional methods, the simple mediation analysis, and more complex or advanced mediation models seems to get more and more important place in the coming years. Keywords: Direct effect, indirect effect, mediation analysis, R software, statistical models.