We have been examining how scientific literature has the potential to stimulate, but also to obstruct, the advancement of knowledge.
Focussing on clinical research, this work contributes to the ongoing debate on research quality and transparency, providing evidence to foster the production of solid scientific knowledge.
Our focus so far has been on the selective reporting of research findings in clinical trials, defined as the publication of only part of the findings originally recorded during a research study, based on the results.
Selective reporting can lead to concerns ranging from publishing flawed scientific knowledge, to skewing medical evidence, to wasting time and resources invested in the conduct of research.
In our first study, we investigated selective reporting and the contextual factors associated with it. Using risk of bias ratings assessed based on expert judgment and presented in systematic reviews of clinical literature, we explored whether selective reporting is associated with the source of institutional support and the type of innovation evaluated.
We found that the odds of selective reporting are higher for industry-funded studies than for publicly-funded studies; however, this effect is restricted to studies where at least one author is industry-affiliated. The results also suggest that selective reporting is more likely in projects exploring radical innovation, compared to those investigating incremental innovation.
The paper has been published in the journal Research Policy
In a second study with Prof Paola Criscuolo (Imperial College) and Prof Ammon Salter (University of Bath) we asked the question of whether systematic reviews could be used to signal errors to the scientific community.
Considering the case of systematic reviews in health care, we explored whether risk of bias ratings communicated within these reviews may help shift scientists’ attention towards published research that is at a low risk of bias due to selective reporting. Using a matched-sample control group we found that, after potential bias is signalled in systematic reviews, publications at high risk of bias attract less attention – as indicated by fewer follow-on citations – when compared to a control group of low risk of bias publications.
We extended our analysis by considering those cases where risk of bias is unclear, and by examining how different features of the rating system may affect the magnitude of the main effect. The findings provide evidence about whether systematic reviews can play a role in signalling biases in the scientific literature, over and above their established role of synthesising prior research.
This paper appears in the journal Research Policy
A working paper with Dr Jan-Michael Ross (Imperial College) explores the role of rivalry in influencing the selective reporting of key research outcomes in such publications.
We define “rivalry” as a subset of competition that captures the subjective relationships between two competitors. Focusing on the context of comparative trials of pharmaceutical drugs, we propose that the chances of selectively reporting drugs’ side effects in scientific articles are higher when the drugs investigated in the trial are rivals than when the investigated drugs are non-rival opponents.
Using data on head-to-head clinical trials for antidepressants and expert-driven assessments of selective reporting, we find that, controlling for structural competition, product-level rivalry arising from relative attributes and prior interactions has distinct effects on selective reporting.
Selective reporting is less likely when the information being released concerns products that share similar attributes, but more likely when there is a history of competition between the products.
An early version of this research appears in the Academy of Management Proceedings