This study extends a previous experiment, showing how a "self-explanation" training intervention promoted learning in an auditing context. A causal model is proposed to examine the effect of providing outcome feedback prior to or after self-explaining on the level of reasoning exhibited in self-explanations, and then the effect of level of reasoning on performance. The model was tested through an experiment in which auditors were trained to assess the reasonableness of client-provided discount rates used in real estate valuation. The following relationships were tested: 1. the effect of outcome feedback timing on the level of reasoning exhibited, 2. the effect of outcome feedback timing on performance, and 3. the diminishing effect of outcome feedback timing on performance when level of reasoning is introduced as a covariate, with level of reasoning then having a significant effect on performance. Results support the causal model, indicating that in the absence of outcome feedback, auditors exhibit a lower level of reasoning in their self-explanations, which then adversely affects their performance.