Bayesian Data Analysis, U. Wisconsin, Madison 2012

## Doing Bayesian Data Analysis University of Wisconsin, Madison Three full days: July 11 - 13, 2012

 Figure 1. Why you should attend the workshop. (Notice that the Bayesian analysis reveals many credible regression lines, for which the slopes and intercepts trade off, instead of just one "best" line.)

The workshop shows you how to do Bayesian data analysis, hands on, with free software called R and JAGS. The intended audience is grad students, faculty, and other researchers from across the University of Wisconsin at Madison area, who want a ground-floor introduction to Bayesian data analysis. No mathematical expertise is presumed. Bayesian analysis provides rich information about the relative credibilities of all candidate parameter values. Bayesian analysis applies seamlessly to small samples, large samples, unbalanced designs, missing data, outliers, etc. Bayesian analysis software is flexible and can be used for a wide variety of data-analytic models. More info about why to go Bayesian is linked below.

Agenda:

• Day 1 (Wed. July 11): Bayesian reasoning generally. Bayes' rule, grid approximation, and R. Example: Estimating the bias of a coin. Markov Chain Monte Carlo and JAGS. Example: Estimating parameters of a normal distribution. HDI, ROPE, decision rules, and null values. Robust Bayesian estimation of differences between two groups.
• Day 2 (Thu. July 12): The t test and null hypothesis significance testing (NHST). The perfidy of p values and the con game of confidence intervals. Sequential testing in NHST and Bayes. The generalized linear model. Simple linear regression. Exponential regression. Multiple linear regression. Logistic regression. Ordinal regression.
• Day 3 (Fri. July 13): Hierarchical models: Estimates of means at individual and group levels. Shrinkage. Examples with beta distributions: therapeutic touch, meta-analysis of ESP. Bayesian hierarchical one way ANOVA. Multiple comparisons and shrinkage. Example with unequal variances. How to report a Bayesian analysis.
• As time permits: Model comparison as hierarchical modeling. The Bayes factor. Doing it in JAGS. Two Bayesian ways to assess null values; estimation is better than model comparison.

Each day's schedule will include the following sessions: 9:00-10:20 presentation, 10:20-10:40 break, 10:40-12:00 presentation, 12:00-1:30 lunch on your own, 1:30-2:50 presentation, 2:50-3:10 break, 3:10-4:30 presentation. Exact time for each topic may flex in response to audience interest and interactive computer discussion.

 Planning to attend? Please register in advance: Register by contacting Kevin Belt, kbelt@psych.wisc.edu. Registration information includes detailed location information.

 ! Install Software Before Arriving: You are invited to bring your personal notebook computer so you can try running the data analyses as they are presented. Before arriving at the workshop, please install the software, all of which is free. For complete installation instructions, please refer to this blog entry. Also be sure to get the additional programs from this site.
 Why go Bayesian? See Figure 1, above. But beyond that, sciences from astronomy to zoology are changing from 20th-century null-hypothesis significance testing to Bayesian data analysis, because Bayesian analysis provides rich information with flexible application to numerous models. Read more: An article that shows the rich information provided by Bayesian estimation in the context of analyzing data from two groups. Kruschke, J. K. (in press). Bayesian estimation supersedes the t  test. Journal of Experimental Psychology: General*. An article that explains a critical flaw of p-values in null hypothesis significance testing, and two different Bayesian approaches to assessing null values. Kruschke, J. K. (2010). Bayesian data analysis. Wiley Interdisciplinary Reviews: Cognitive Science*, 1(5), 658-676. An article that emphasizes advantages of Bayesian data analysis and the fact that Bayesian data analysis is appropriate regardless of the status of Bayesian models of cognition. Kruschke, J. K. (2010). What to believe: Bayesian methods for data analysis. Trends in Cognitive Sciences*, 14(7), 293-300. An article that explains two Bayesian methods to assess null values, and which one is typically more informative. Kruschke, J. K. (2011). Bayesian assessment of null values via parameter estimation and model comparison. Perspectives on Psychological Science*, 6(3), 299-312. *Your click on this link constitutes your request to the author for a personal copy of the article exclusively for individual research.
Who is the instructor? John Kruschke is seven-time winner of Teaching Excellence Recognition Awards from Indiana University, where he is Professor of Psychological and Brain Sciences, and Adjunct Professor of Statistics. He has written an introductory textbook on Bayesian data analysis; see also the articles linked above. His research interests include the science of moral judgment, applications of Bayesian methods to teaching and learning, and models of attention in learning, which he has developed in both connectionist and Bayesian formalisms. He received the Troland Research Award from the National Academy of Sciences. He is action editor for the Journal of Mathematical Psychology, and is on the editorial boards of Psychological Review, the Journal of Experimental Psychology: General, among others.

Recommended textbook: Doing Bayesian Data Analysis: A Tutorial with R and BUGS. The book is a genuinely accessible, tutorial introduction to doing Bayesian data analysis. See some published reviews of the book here. Other endorsements and information about the book can be read here.

 Bayesian data analysis is not Bayesian modeling of cognition. Data analysis involves "generic" descriptive models (such as linear regression) without any necessary interpretation as cognitive computation. The rational way to estimate parameters in descriptive models is Bayesian, regardless of whether or not Bayesian models of mind are viable. The concepts and methods of Bayesian data analysis transfer to other Bayesian models, including Bayesian models of cognition. Read more at this blog entry.