Library
How are Total Effects on Target and Direct Effects on Target calculated in BayesiaLab?
Quote 0 0
Library
Both the Total and Direct Effects are the derivative of their corresponding Effect Function computed at the a-priori mean value (delta = 0).Total Effect FunctionThe Total Effect Function is estimated by using the Mean Value Analysis (based in the MinXEnt) to go through the variation domain of a variable in order to measure its impact on the Target mean.totalEffect.png Adding 1 to the mean value of Corresponds (green curve) increases the target node Pleasure_(4) by 0.78totalEffect1.png Total EffectsTotal Effects are the derivatives of the Total Effect functions, taken at the a-priori mean values of the variables,[latex:27t6gofi]Te=\frac{\delta_y}{\delta_x}[/latex:27t6gofi]the Standardized value normalizing the effect by taking into account the ration between the standard deviation of the variable and the one of the Target.[latex:27t6gofi]STe=\frac{\delta_y \times \sigma_x}{\delta_x \times \sigma_y}[/latex:27t6gofi]totalEffect2.png totalEffect3.png Direct Effect FunctionThe Direct Effect Function is estimated by using the Mean Value Analysis (based in the MinXEnt) to go through the variation domain of a variable in order to measure its impact on the Target mean, while holding fixed the probability distributions of all the variables, except:[*:27t6gofi]The Not-Observable variables belonging to the Class “Factor”[/*:m:27t6gofi][*:27t6gofi]The variables belonging to the Class “Non_Confounder” .[/*:m:27t6gofi]directEffect.png Adding 1 to the mean value of Corresponds (green curve) while holding fixed all the marginal probability distributions of all the other variables increases the target node Pleasure_(4) by 0.27directEffect1.png Direct EffectsDirect Effects are the derivatives of the Direct Effect functions, taken at the a-priori mean values of the variables,[latex:27t6gofi]De_x=\frac{\delta_y}{\delta_x}[/latex:27t6gofi]the Standardized value normalizing the effect by taking into account the ration between the standard deviation of the variable and the one of the Target.[latex:27t6gofi]SDe_x=\frac{\delta_y \times \sigma_x}{\delta_x \times \sigma_y}[/latex:27t6gofi]directEffect2.png directEffect3.png 
Quote 0 0