Test-Driven Development (TDD) is one of the cornerstone practices of the Extreme Programming agile methodology. Today, despite the large scale adoption of TDD in industry, including large software firms such as Microsoft and IBM, its usefulness with regard to the quality and productivity constructs it still under question. Empirical Research has failed to produce conclusive results; all possible results have been reported for both constructs. This research adopts non-empirical measures to gain a deeper understanding of TDD. A two-phased approach has been undertaken towards the goal. The first phase involves conducting a meta-analysis of past empirical research. The meta-analysis quantitatively combines the results of individual empirical studies and identifies moderator variables that could potentially govern the performance of TDD. The second phase of the approach involves the construction of a simulation model of a TDD-based development process. The presented model further analyzes the impact of changes in moderator variables.