Theme Issue: Bayesian Methods In Medical Research
The Journal of Medical Internet Research (JMIR) is inviting submissions for an e-collection of the journal that will be dedicated to the use of Bayesian Methods in Medical research.
The misuse of P values continues to be an issue in medical and social science. The hunt for true and false has led to misinterpreting non-statistical significance as absence of effect and using large sample sizes to identify effect sizes of no real-world significance. The focus on dichotomization of evidence has led to fear of type I errors, which have in turn led to questionable practices of power calculations, fear of multiplicity, and concerns of P value spending. Change is however coming, albeit slow, and it is going to require a re-education and possibly much debate before a new consensus is found on how to deal with evidence from scientific studies.
One approach that has been proposed is to use Bayesian methods to analyze data. Evidence would then be considered a continuous measure, and would not be dichotomized into significant or non-significant. Researchers would be responsible for assessing the evidence collected using a holistic view, taking into consideration study design, potential biases, costs involved in rolling out the intervention, risks to patients, results from previous studies, etc.
JMIR has recently published a tutorial that compares the traditional null hypothesis testing approach with a Bayesian approach (A gentle introduction to the comparison between null hypothesis testing and Bayesian analysis: reanalysis of two randomized controlled trials), and is now seeking papers in which authors use Bayesian methods. JMIR is also offering the possibility for authors to upload their data to JMIR Data if they would like support with the analysis (please see details below).
SUBMISSION OF PAPERS
Submitted papers should include use of Bayesian methods, including, but not limited to: analyses of outcomes of randomized trials, trial designs (eg. Bayesian group sequential designs), analyses of observational studies, and use of Bayesian methods for explorative analyses. JMIR guidelines for manuscripts apply.
Manuscripts should be sent through the online system at http://www.jmir.org/author.
In submission step 1, authors must choose the section “E-collection: Bayesian Methods in Medical Research” (Guest editor: Marcus Bendtsen) (see also How do I submit to a theme issue?).
All submitted manuscripts will undergo a full peer review process consistent with usual rigorous editorial criteria for JMIR. Accepted papers will be published in JMIR, or may be transferred to JMIR Public Health and Surveillance, JMIR Medical Informatics, JMIR Mental Health, JMIR Human Factors, JMIR Res Protoc or another JMIR sister journal, according to focus and impact of the paper. All papers will appear together in an e-collection guest edited by the academics listed below. Papers rejected for the theme issue may still be considered for regular issues.
For this special issue, Article Processing Fees are discounted by 20% (saves up to $500).
Authors who are interested to contribute to the special issue, but do not have the resources to conduct the Bayesian analysis on their own, can use JMIR Data to access support for their analysis. The editors will conduct the statistical analysis for a selection of projects, and then allow the original authors to interpret and discuss the findings in a submission to the special issue. Read more about JMIR Data here: https://data.jmir.org/announcement/view/138 and submit your dataset here: http://data.jmir.org/author. Contact the guest editor below once you have submitted your dataset.
Marcus Bendtsen, Linköping University, firstname.lastname@example.org