In many statistical and mathematical models, the idea of separating variables into independent variables and dependent variables is an important one. Independent variables are difficult for the researcher to control and are able to vary on their own. Dependent variables rely on the independent variables; that is, their values can be predicted from knowing the independent variables. For example, if you have a job that pays you based on how many hours you work, the time you put into your work is an independent variable whereas your monthly salary is a dependent variable. Its value depends on how much you work. In path analysis, endogenous variables act as the dependent variables. Path analysis is mainly interested in how the outcome of endogenous variables is related to the independent variables. Independent variables are called exogenous variables in path analysis.
Related to each endogenous is a disturbance variable. These disturbance variables are latent in that we cannot directly measure them. The disturbance represents the error in the prediction of an endogenous variable and is analogous to the idea of residual error in other statistical models.
One special thing about an endogenous variable is that it can arise from more than one exogenous variable. Take, for example, a path from A to C and another path from B to C in the same path analysis. If A and B are unrelated to each other but both related to C, a parallel relationship is formed. The implication is that both A and B must be present to cause C.
Unlike other models, path analysis allows dependent variables to have a sense of duality. In short, this duality is that endogenous variables can be exogenous at the same time. This allows endogenous variables to give rise to other endogenous variables. For example, in a linear path from nodes A to B to C, there are two arrows: one between A and B; the other between B and C. The first arrow shows that B is endogenous because it is caused by A. The second arrow shows that B is exogenous because it is the cause of C.