ABSTRACT
In this chapter, we discuss some other important issues in the third-variable effect analysis. Section one includes the caveats in explaining third-variable effects. When associations among variables are causal or not causal, the third-variable effect can be called mediation or confounding. It is important to understand the research purposes to include variables in analysis and to explain the variables differently. Section one also discusses the explanation of negative indirect effects, very big relative effects, group effects and average effects. Section two discusses the power analysis and sample size calculation in the third-variable analysis. Two methods are discussed: one is only for generalized linear models on finding the distribution of the product of two random variables and the other is more general based on Monte Carlo method. Codes are provided for the Monte Carlo method to calculate computing powers. Finally, in Section three, we very briefly discuss the use of third-variable analysis for sequential and longitudinal data.
