Welcome to the Bayesian Statistical Science Section

The Bayesian Statistical Science Section of the ASA provides a forum for statisticians and people who have interest in the Bayesian paradigm. The broad objectives of the Section are: to encourage research on theory and methods of statistical inference and decisionmaking associated with Bayes' theorem and to encourage the application and proper use of Bayesian procedures in the behavioral, biological, managerial, engineering, environmental, legal, medical, pharmaceutical, physical, and social sciences.

Current Affairs 

2014 Mitchel Prize and Savage Awards

The Prize Committee of ISBA is pleased to announce the call for submissions for the 2014 Mitchell Prize and Savage Awards.

The winner(s) will be announced at the Joint Statistical meetings in Seattle, 2015.  The deadline for submissions is 31 May, 2014 (midnight UTC/GMT, 7pm EST, 4pm PST).

The Mitchell Prize is given in recognition of an outstanding paper that describes how a Bayesian analysis has solved an important applied problem. The prize includes a check for $1,000.00 and a plaque. Details on the Mitchell Prize, including names of past winners, eligibility details, and the on-line application procedure, can be found at http://www.bayesian.org/awards/MitchellPrize.html  .

The Savage Award, named in honor of Leonard J. "Jimmie" Savage, is bestowed each year on two outstanding doctoral dissertations in Bayesian Econometrics and Statistics, one each in “Theory & Methods” and “Applied Methodology”.  Doctoral dissertations submitted for the Savage Prize must be written in English. Up to two awards of $750.00 will be awarded.  Finalists will be notified in mid-December and invited to present their dissertation research at a special contributed session at the 2015 JSM in Seattle, with the winners announced at the meeting. For details on the Savage Award, including names of past winners, eligibility details, and the on-line application procedure, please visit http://www.bayesian.org/awards/Savage.html.

Nominations for the Mitchell and Savage Award may be made by any ISBA or SBSS member.  

For questions regarding any of the Prizes or Awards may be sent to the ISBA Prize Committee at awards@bayesian.org.

Upcoming Web-Based Lectures

Title: Applied Bayesian Hierarchical Modeling for the Social Sciences
Presenter: Jeff Gill, Washington University
Date and Time: Thursday, April 17, 12:00 p.m. - 2:00 p.m. Eastern Time
Sponsor: Section on Bayesian Statistical Science and the International Society for Bayesian Analysis

Registration Deadline: Tuesday, April 15, at 12:00 p.m. Eastern time

This webinar presents a set of multilevel models for social science research that provide a great amount of flexibility to handle data at different levels of aggregation. This applied tutorial will use Markov chain Monte Carlo tools to fit linear and nonlinear specifications with multiple levels, longitudinal features, and non-normal distributional assumptions. Content will include some theoretical discussions of modeling and estimation, but will concentrate more on as practical guidance for fitting multilevel models with JAGS software.

*** More information and registration at the webinar page.

Title: Variance Estimation in Complex Sample Surveys
Presenter: Richard Valliant, University of Maryland
Date and Time: Wednesday, April 23, 2014, 1:00 p.m. - 3:00 p.m. Eastern time
Sponsor: Survey Research Methods Section

Registration Deadline: Monday, April 21, at 12:00 p.m. Eastern time

This webinar will provide an overview of the methods for variance estimation in complex sample survey data. Two approaches: linearization and replication will be compared and contrasted. Software options will be examined for different types of estimates.

*** More information and registration at the webinar page.

Title: Using Administrative Data: Strengths and Weaknesses
Presenter: Joe Sakshaug, University of Michigan
Date and Time: Monday, May 12, 2014, 1:00 p.m. - 3:00 p.m. Eastern time
Sponsor: Survey Research Methods Section

Registration Deadline: Thursday, May 8, at 12:00 p.m. Eastern time

This webinar will provide a detailed overview of administrative data; their possible uses, strengths, and limitations. Real applications of administrative data used in a social context will be presented from projects conducted at the Institute for Employment Research in Nuremberg Germany.

Title: Statistical Aspects of Long Term Safety Cohort Studies
Presenter: Girish (Gary) Aras (Amgen, Inc.)
Date and Time: Wednesday, May 14, 2014, 12:00 p.m. - 2:00 p.m. Eastern time
Sponsor: Biopharmaceutical Section

Registration Deadline: Monday, May 12, at 12:00 p.m. Eastern time

To establish a long-term safety profile for a newly marketed drug, a long-term cohort safety study without a comparator arm is often undertaken by drug companies. Such a study routinely enrolls more subjects, and has longer study duration and less frequent scheduled visits, than a typical randomized clinical trial. In the context of such a study, we shall review basic epidemiological concepts such as prevalence and incidence associated with adverse events and discusses statistical methods employed to measure them. The study is often sized to detect adverse events that are relatively rare and/or have a long latency period, as in the case of malignancies and heart diseases. Due to the drug's mechanism of action or its membership to a drug class with a known adverse event profile, as the cumulative exposure to the drug increases longitudinally, one may anticipate an increase in incidence of certain adverse events compared to historically known background incidence. These adverse events, sometimes known as events of medical interest (EMI), are hence prospectively stated in the protocol and are followed during the study. Some of these EMI, such as infections, asthma exacerbations, and allergic rhinitis may not be rare and may exhibit seasonal variations in incidence. We shall review classification of adverse events based on prevalence that is acceptable to major regulatory agencies worldwide can be found in Report of Council for International Organization of Medical Sciences (CIOMS).

Due to the long duration and relatively lower frequency of follow up visits, dropouts are an inherent part of a long-term cohort safety study, and these factors are typically accounted for in the sample size and are modeled in the analysis. The adverse events are summarized in various ways. Estimates of cumulative incidences and annual incremental incidences based on person time of exposure/observation, cumulative event incidences, and annual incremental event rates (counting recurrent events in the numerator as opposed to counting subjects with at least one event) per person time of exposure/observation are especially popular in epidemiology literature. We shall review these measurement concepts in detail. The main limitation of these methods is that they are all based on the assumption of constant hazard over the study period. Assumption of constant hazard is equivalent to assuming that the time to event is exponentially distributed. Due to lack of better alternative, methods based on the unrealistic assumption of constant hazard rate are employed to estimate incidence rates in retrospective studies or in meta-analysis of past studies, in which dates of exposure and events are not accurately available for every subject. However, in prospective cohort studies where information at the subject level regarding dropouts and loss to follow up is readily available, more sophisticated methods such as Kaplan-Meier or life-table estimators can be employed to obtain better estimates. We shall discuss underlying assumptions and strengths and limitations of these methods as well. We may employ simulations to further characterize these methods.

The course will intermittently link the above ideas and techniques to examples from drug labels. We shall discuss how safety evidence based on such studies and are summarized or should be summarized in drug labels.

*** More information and registration at the webinar page.


From the SBSS Mixer at JSM 2013