Context Replication plays a critical role in building cumulative scientific knowledge. However, in the context of Empirical Software Engineering (ESE), replication efforts face persistent difficulties, both technical and methodological, which hinder reproducibility and generalizability. We explored whether the problems were due to formal issues, such as under‐specification or miscommunication, or intrinsic reasons, i.e., the existing replication procedures may not meet the researchers’ needs. Objective To understand how ESE researchers conduct experiments in real‐world settings. Method We conducted an ethnographic study with an experimental software engineering group, using interviews, observations, and analysis of internal documentation. Results We have created conceptual and process models representing experimentation, replication, and synthesis in the target research group. These models align with mainstream procedures at a high level but often break down in practice during specialized tasks such as coordination or documentation. The experimental process observed in the group differs from common assumptions and textbook descriptions in terms of (1) the number and diversity of activities involved, (2) the presence of differentiated roles, (3) the granularity of conceptual elements, and (4) the varying perspectives across subareas or families of experiments. Conclusions The discrepancies between actual laboratory practices and idealized process models may hinder knowledge transfer and complicate replication. We plan to extend this research by involving additional research groups, for instance, through surveys, to examine the generalizability of our findings.

Background: Publication bias is the failure to publish the results of a study based on the direction or strength of the study findings. The existence of publication bias is firmly established in areas like medical research. Recent research suggests the existence of publication bias in Software Engineering. Aims: Finding out whether experiments published in the International Workshop on Empirical Software Engineering and Measurement (ESEM) are affected by publication bias. Method: We review experiments published in ESEM. We also survey with experimental researchers to triangulate our findings. Results: ESEM experiments do not define hypotheses and frequently perform multiple testing. One-tailed tests have a slightly higher rate of achieving statistically significant results. We could not find other practices associated with publication bias. Conclusions: Our results provide a more encouraging perspective of SE research than previous research: (1) ESEM publications do not seem to be strongly affected by biases and (2) we identify some practices that could be associated with p-hacking, but it is more likely that they are related to the conduction of exploratory research.

Background: Statistical concepts and techniques are often applied incorrectly, even in mature disciplines such as medicine or psychology. Surprisingly, there are very few works that study statistical problems in software engineering (SE). Aim: Assess the existence of statistical errors in SE experiments. Method: Compile the most common statistical errors in experimental disciplines. Survey experiments published in ICSE to assess whether errors occur in high quality SE publications. Results: The same errors as identified in others disciplines were found in ICSE experiments, where 30% of the reviewed papers included several error types such as: a) missing statistical hypotheses, b) missing sample size calculation, c) failure to assess statistical test assumptions, and d) uncorrected multiple testing. This rather large error rate is greater for research papers where experiments are confined to the validation section. The origin of the errors can be traced back to: a) researchers not having sufficient statistical training, and, b) a profusion of exploratory research. Conclusions: This paper provides preliminary evidence that SE research suffers from the same statistical problems as other experimental disciplines. However, the SE community appears to be unaware of any shortcomings in its experiments, whereas other disciplines work hard to avoid these threats. Further research is necessary to find the underlying causes and set up corrective measures, but there are some potentially effective actions and are a priori easy to implement: a) improve the statistical training of SE researchers, and b) enforce quality assessment and reporting guidelines in SE publications.

Context: Empirical Software Engineering (ESE) replication researchers need to store and manipulate experimental data for several purposes, in particular analysis and reporting. Current research needs call for sharing and preservation of experimental data as well. In a previous work, we analyzed Replication Data Management (RDM) needs. A novel concept, called Experimental Ecosystem, was proposed to solve current deficiencies in RDM approaches. The empirical ecosystem provides replication researchers with a common framework that integrates transparently local heterogeneous data sources. A typical situation where the Empirical Ecosystem is applicable, is when several members of a research group, or several research groups collaborating together, need to share and access each other experimental results. However, to be able to apply the Empirical Ecosystem concept and deliver all promised benefits, it is necessary to analyze the software architectures and tools that can properly support it.