An approach for quantitative aggregation of evidence from controlled experiments in software engineering

Empirical studies are necessary to gain reliable insights into the effects of software engineering technologies and to allow controlling risks associated with their usage. Recently, many empirical studies have been run in many software engineering areas (e.g., inspections). However, in order to be useful for decision-making, synthesis is required. Synthesis means to analyze, combine, summarize, and generalize the results of empirical studies. However, software engineering lacks a systematic approach for synthesis: Today, most syntheses in software engineering use narrative, informal summaries. These narrative reviews suffer from a number of weaknesses; in particular, they are subjective and thus often incorrect.