Improving the Quality of Controlled Experiments in Software Engineering

Abstract

All experimental studies are prone to risk with regard to the validity of their results. After applying a determined action to control a threat, during planning or executing an experiment, new risks to the study validity can arise. Aim. To improve the quality of controlled experiments in Software Engineering (SE), setting out strategies that permit researchers giving priority to specific threats in causes virtue and consequences (trade-offs), which exist between threats to validity and possible measures for control. Method. We will employ a knowledge base acquired through a survey and a Systematic Literature Review to model an approach for prioritizing and controlling threats to validity. The proposed approach will also be evaluated through experiments. Contribution. To improve the control processes for threats to validity considered to be critical, reducing efforts required by researchers, thereby improving the quality of future controlled experiments in the Software Engineering area by increasing its results? validity.