I am grateful to the commentators for their thoughtful and perceptive remarks. On reading these comments and having attended a number of meetings where these issues have been discussed, I can only agree with most of the commentators that a voluntary code for making data and command syntax available does not go far enough. The way we conduct our science needs substantial restructuring. Lack of ability of the wider scientific community to scrutinize data sets and statistical analysis commands is one issue and probably a substantial one. However, lack of transparency in the process leading up to the reporting of findings and selective reporting of findings are also major issues that need addressing. Tools such as the Open Science Framework (https://osf.io/) make it possible to address many of these issues if there is the will.
However, the problem goes further: writing articles as narratives is an extremely inefficient communication process that leads to unnecessary repetition on the one hand and failure to describe key features of studies on the other. Material is repeated when, as is often the case, a single large study gives rise to multiple articles. When it comes to accurate and comprehensive reporting, even experienced researchers struggle to ensure that all the important features of a study are described adequately, despite the increasing use of guidance such as CONSORT (Consolidated Standards of Reporting Trials—see http://www.equator-network.org/). And establishing the case for a study and interpretation of findings is still a dark art that in my experience can make a huge difference to whether an article gets accepted and once it is accepted to how the findings are used.
Advances in computer science mean that we no longer have to be bound by the traditional way of communicating science: the scientific article. Already one can see moves to expand the methods of knowledge transmission in terms of shared data sets, videos, software, and so on. In terms of the formal communication of study findings, however, the journal article clearly predominates. I do not have space to go into the details here, but I think there is a much better approach if we can get there: scientists contribute to a set of formal “ontologies” (ultimately perhaps a single unified ontology). An ontology is a data structure that specifies a set of elements, their properties, and relationships between them. In science, ontologies can be created that link the following types of item: research hypotheses, formal theories, study methods, data sets, statistical command files, study findings, study conclusions, conflict of interest statements, and recommendations. There might, for example, be an ontology for clinical interventions to promote smoking cessation, or for environmental interventions to promote healthy eating, or counseling interventions to increase physical activity levels. Ontologies have already begun to be developed in public health science (e.g., [
- McKone T.E.
- Feng L.
Building a human health risk assessment ontology (RsO): a proposed framework.
Risk Anal. 2015; ([Epub ahead of print])https://doi.org/10.1111/risa.12414
2]). Such ontologies could be constructed using a standard format allowing them to be linked to create super ontologies. Under such a system, story-telling skill becomes less important and creativity can be channeled into understanding evidence, generating theories and hypotheses and finding ways of testing these. Patterns in the data can emerge that we would never be able to glean using the patchwork approach to science currently being adopted.
- McMurray J.
- Zhu L.
- McKillop I.
- Chen H.
Ontological modeling of electronic health information exchange.
J Biomed Inform. 2015; ([Epub ahead of print])https://doi.org/10.1016/j.jbi.2015.05.020
This is a grand vision, and the route from where we are to such a radically different paradigm will involve navigating uncharted and rocky waters, but it is clear that there is sufficient impetus among stakeholders to move our science in that direction—and widening agreement that the status quo is not an option [
- Glasziou P.
- Altman D.G.
- Bossuyt P.
- Boutron I.
- Clarke M.
- Julious S.
- et al.
Reducing waste from incomplete or unusable reports of biomedical research.
Lancet. 2014; 383: 267-276
- Chalmers I.
- Glasziou P.
Avoidable waste in the production and reporting of research evidence.
Lancet. 2009; 374: 86-89
- Building a human health risk assessment ontology (RsO): a proposed framework.Risk Anal. 2015; ([Epub ahead of print])https://doi.org/10.1111/risa.12414
- Ontological modeling of electronic health information exchange.J Biomed Inform. 2015; ([Epub ahead of print])https://doi.org/10.1016/j.jbi.2015.05.020
- Reducing waste from incomplete or unusable reports of biomedical research.Lancet. 2014; 383: 267-276
- Avoidable waste in the production and reporting of research evidence.Lancet. 2009; 374: 86-89
Published online: July 07, 2015
Accepted: June 23, 2015
Conflict of interest: None.
Financial disclosure: None.
© 2016 Elsevier Inc. Published by Elsevier Inc. All rights reserved.
ScienceDirectAccess this article on ScienceDirect
- Data and statistical commands should be routinely disclosed in order to promote greater transparency and accountability in clinical and behavioral researchJournal of Clinical EpidemiologyVol. 70
- PreviewThis commentary argues for clinical, public health, and health science journals routinely to invite authors to make data and statistical analysis command files underlying the findings reported in articles available as supplementary files and signal prominently articles for which this is done using a “transparency” quality marker.