Merging And Splitting: Integrating Theory-Oriented Evaluation Approaches Barbara Befani University of Rome “La Sapienza”, barbara.befani@uniroma1.it This article presents an innovative evaluation approach jointly inspired by the two renowned, theory-oriented, Pawson and Tilley’ realistic evaluation and Carol Weiss’s theory-based evaluation. First, the two approaches’ main features are reviewed, with special regard to those which are preserved in the unified model. Secondly, their theoretical differences and similarities are tackled while a comprehensive framework that merges the most interesting features of the two is proposed. In particular, the program context is split in two to distinguish between its immutable and alterable (by the program) characteristics, in a way that will let the same concept of context be a part of both implementation and program theories. Finally, a practical example is presented when the model is applied to women’s employment policy. Keywords: Evaluation, Theory-Oriented Approaches, Scientific Realism, Implementation 1. Realistic Evaluation The realist approach to evaluation was introduced in the 1997 classic ‘Realistic Evaluation’ by Ray Pawson and Nick Tilley. The two authors were determined to bring “realism in the car park”, that is to use the principles of scientific realism for purposes different from academic discussions or philosophical debates. Being sociologists, they were attempting something remarkably innovative as only various hard and natural scientists had dared to apply realist principles in real-life, actual experiments. And, being social programs some kind of laboratories of social change, their evaluations probably looked as best-fit opportunities to try their hands at the job. In the following, I will outline the fundamental ideas behind this epistemology which has, since then, gained a certain popularity among evaluators. 1.1. The Basics Of Scientific Realism According to realist convictions, scientific progress occurs when experiments are able to incorporate and test theories – which are modified after experimental results – and not when connections are repeatedly discovered to exist among phenomena. In other words, in order for science to evolve, it is more important to construct experiments that embody a theory, to change that theory after evidence is obtained, and to build a new experiment incorporating the new theory, rather than repeating the same experiment over and over in