Experimental Evaluation of Software and Systems in Computer Science

Critiques of the quantity and quality of empirical evaluations in software engineering have existed for quite some time. However such critiques are typically not empirically evaluated. This paper fills this gap by empirically analyzing papers published by ICSE, the prime research conference on Software Engineering. We present quantitative and qualitative results of a quasi-random experiment of empirical evaluations over the lifetime of the conference. Our quantitative results show the quantity of empirical evaluation has increased over 29 ICSE proceedings but we still have room to improve the soundness of empirical evaluations in ICSE proceedings. Our qualitative results point to specific areas of improvement in empirical evaluations.