Purpose: Stuttering is associated with clinically significant social anxiety, which emerges during early childhood for some, but not all, children who begin to stutter. The purpose of this review article is to develop a model of social anxiety development during early childhood stuttering and to present an empirical method by which it can be tested. Method: We propose a mediation model of how the exposure variable of stuttering may lead to an outcome of social anxiety. Our model includes confounder and mediator variables. We explain the concepts and procedures of mediation analysis and present a method to test our model. Results: We present the idea that negative peer responses to stuttering and negative self-perception of children are mediators of social anxiety development. We propose several confounder variables that involve children, their parents, and the home environment. We depict our model with a directed acyclic graph, and we present details of how it can be tested with a longitudinal research design. Discussion: This is the first attempt to model the development of social anxiety shortly after stuttering onset with an empirically testable method. The intended benefit of this innovation is to direct future clinical directions for the clinical management of stuttering arising shortly after childhood onset.