Methodological Advances in the Evaluation of Public Policy - An Experimental and Quasi-Experimental Approach

The dearth of social experiments, also known as randomised control trials (RCT), has led much evaluation research to rely on non-experimental and quasi-experimental methods which are more ethically and practically feasible, yet require bolder statistical assumptions. There is much debate concerning the validity of these assumptions and the conditions under which quasi-experimental methods most closely approximate the results of a well-designed RCT. This research programme will use the opportunity to design and utilise experimental data generated from a series of early childhood interventions in Ireland to test the validity and quantify the different sources of bias typically found in non-experimental studies. By combining high quality data from the field with rigorous econometric modelling this project will answer a central question of the evaluation literature - when randomised control trials are not feasible under which conditions do quasi-experimental designs most closely approximate experimental results? [References]