The BaSiS Group
Address correspondence to:
Constantine Gatsonis, gatsonis@stat.brown.edu , tel: 401 863 9183, or
Steve Goodman: sgoodman@jhmi.edu, tel: 410-955-4596.
·
Enormous methodologic and computational advances in the last two decades have
brought Bayesian statistics into the
mainstream of scientific study design and analysis.
·
Whereas considerable experience and consensus
exists in the conduct and reporting of frequentist analyses, the corresponding consensus regarding
Bayesian analyses is only now emerging among statisticians, and is effectively
unknown outside of statistics.
·
While a variety of major scientific journals have published
articles explaining and encouraging the use of Bayesian methods, they are
rarely utilized, particularly for the purpose of statistical inference.
·
As with many new quantitative methodologies,
many journal editors are uncertain how Bayesian analyses should be critiqued or
presented to allow for proper assessment by readers of both the results and the
methodology.
·
Most authors are wary of submitting research
using Bayesian methods, fearing that these are non-standard and unwelcome in
scientific journals.
·
The purpose of the BaSiS initiative is to
develop guidelines and an accompanying
explanatory document on standards for reporting Bayesian analyses, for the use
of authors, reviewers, and journal editors.
·
The initiative is modeled after recent
similar initiatives in the medical
literature (notably CONSORT).
However, the BaSiS guidelines do not
cover the entire structure of a
scientific paper, as is the case in the omnibus guidelines that have been
proposed. Instead, they are focused on those aspects of scientific reporting, which are salient for a
Bayesian analysis.
· The process of developing the BaSiS documents included… [brief description of process]
· The BaSiS group began with a literature search for published guidelines on reporting scientific results and in particular, Bayesian methods [currently underway ] and solicited the opinion of a panel of expert methodologists in order to formulate the first draft of the guidelines and accompanying statement.
· The drafts were discussed in a special session at the 6th Workshop on Case Studies in Bayesian Statistics (Carnegie Mellon University, Sept. 2001).
· Comments on the revised drafts were solicited from a broad spectrum of stakeholders, including statistical methodologists, journal editors and researchers in medicine, biology, the behavioral and the social sciences.
· Final guidelines were developed and were made accessible to interested parties via the web.
·
It is expected that the guidelines will be
incorporated into the Instructions for Authors section of journals that adopt
them, with reference to the explanatory document, which will be published in a
high profile medical or methodologic journal.
·
Short description of the findings from the
search (including BAYESWATCH, CONSORT, QUORUM, and MOOSE).
·
Description of the guidelines, with explanation
of rationale for each one.
· Need for standards in reporting Bayesian methods.
o Will encourage both consistency and proper conduct of Bayesian methods. This will become increasingly important as user-friendly software (e.g. WinBUGS) and increasing acceptability of Bayesian methods result in more analyses conducted by inexperienced analysts.
o Will help increase the use of Bayesian methods by eliminating the perception that they are non-standard or unwelcome in scientific journals.
o Will provide guidance to editors , reviewers and readers for both the presentation, interpretation and critique of Bayesian analyses.
o Consistency of format will help readers better understand such analyses.
· Comparison with standards for frequentist methods and specific scientific/analytic methodologies (e.g. RCTs, meta-analyses, cost-effectiveness analyses).
· Intended use of the guidelines.
· Plans for monitoring the use of the guidelines, and for possible updates in the future.
2. Bailar J. , Mosteller F. Guidelines for Statistical Reporting for Medical Journals: Amplifications and Explanations. Ann Int Med. 1988; 108: 266-73.
3. Bruns DE, Huth EJ, Magid E, Young DS. Toward a checklist for reporting of studies of diagnostic accuracy of medical tests. Clin Chem. 2000; 46(7):893-5.
5. Hughes, M. Reporting Bayesian Analyses of Clinical Trials. Stat Med . 1993; 12:1651-1664.
6. Information for Authors: Annals of Internal Medicine, (paragraph on presentation of Bayesian results) WWW.annals.org/shared/manu_format.html
7. Lang TA, Secic M. How To Report Statistics in Medicine. Philadelphia: American Coll of Physicians; 1997.
8. Lehmann HP, Goodman SN. Bayesian communication: a clinically significant paradigm for electronic publication. J Am Med Inform Assoc. 2000 7(3):254-66.
9. Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D, Stroup DF. Improving the quality of reports of meta-analyses of randomised controlled trials: the QUOROM statement. Quality of Reporting of Meta-analyses. Lancet. 1999;354:1896-900.
10. Moher D, Schulz KF, Altman DG, Lepage L. The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomised trials. Lancet. 2001 Apr 14;357:1191-4.
11. Siegel JE, Weinstein MC, Russell LB, Gold MR. Recommendations for reporting cost-effectiveness analyses. Panel on Cost-Effectiveness in Health and Medicine. JAMA. 1996;276:1339-41.
12. Spiegelhalter DJ, Myles JP, Jones DR, Abrams KR. Bayesian methods in health technology assessment: a review. Health Technol Assess. 2000;4:1-130.
13. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, Moher D, Becker BJ, Sipe TA, Thacker SB. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. A proposal for reporting. JAMA. 2000;283:2008-12.
14. Uniform requirements for manuscripts submitted to biomedical journals. International Committee of Medical Journal Editors. JAMA 1997; 277:927-34.
BaSiS guidelines
List of members of the BASIS group