Replication studies considered harmful

TL;DR

This paper argues that many replication studies in software engineering are uninformative due to wide prediction intervals, and advocates for meta-analysis instead of replication to better estimate effects.

Contribution

It demonstrates through simulation that replication often confirms results without adding meaningful knowledge, and promotes meta-analysis as a more effective approach.

Findings

Most replications are confirmatory due to wide prediction intervals.

Replication efforts with under-powered studies are often scientifically wasteful.

Meta-analysis provides a better estimate of the true effect size.

Abstract

CONTEXT: There is growing interest in establishing software engineering as an evidence-based discipline. To that end, replication is often used to gain confidence in empirical findings, as opposed to reproduction where the goal is showing the correctness, or validity of the published results. OBJECTIVE: To consider what is required for a replication study to confirm the original experiment and apply this understanding in software engineering. METHOD: Simulation is used to demonstrate why the prediction interval for confirmation can be surprisingly wide. This analysis is applied to three recent replications. RESULTS: It is shown that because the prediction intervals are wide, almost all replications are confirmatory, so in that sense there is no 'replication crisis', however, the contributions to knowledge are negligible. CONCLUSIONS: Replicating empirical software engineering experiments, particularly if they are under-powered or under-reported, is a waste of scientific resources. By contrast, meta-analysis is strongly advocated so that all relevant experiments are combined to estimate the population effect.