On Implicit Assumptions Underlying Software Engineering Research

Abstract

Background: Software engineering research articles should make precise claims regarding their contribution, so that practitioners can decide when they might be interested and researchers can better recognize (1) whether the given research is valid, (2) which published works to use as stepping stones for their own research (and which not), and (3) where additional research is required. In particular, articles should spell out what assumptions were made at each research step. Question: Can we identify recurring patterns of assumptions that are not spelled out? Method: This is a position paper. It formulates impressions, but does not present concrete evidence. Results: Assumptions that are wrong or assumptions that are risky and not explicit threaten the integrity of the scientific record. There are several recurring types of such assumptions. The frequency of these problems is currently unknown. Conclusion: The software engineering research community should become more conscious and more explicit with respect to the assumptions that underlie individual research works.