Online Program

Return to main conference page
Program-at-a-Glance
Keynote Address | Concurrent Sessions | Poster Sessions
Short Courses (full day) | Short Courses (half day) | Tutorials | Practical Computing Demonstrations | Closing General Session with Refreshments

Last Name:

Abstract Keyword:

Title:

     

Viewing Tutorials onlyView Full Program
     
Saturday, February 17
T1 Engage the Room: Mastering Your Personal Presentation Style
Sat, Feb 17, 2:00 PM - 4:00 PM
Salon A
Instructor(s): Duncan Burl Gilles, Art of Problem Solving

Download Handouts
As confident as we may be in the quality of our work, presentation can make or break the impact it has. Engaging the room and communicating clearly can make the difference between an unimpressed, bored audience and a thrilled audience eager to learn more. This course will focus on presentation techniques that help you communicate your ideas effectively and in an engaging manner. You’ll be trained on ways to draw your audience into your talk, engage them in active listening and thinking, and use your voice and the space of the room to command attention and convey your message. These are skills applicable in many areas – whether presenting your work to clients, teaching in the classroom, one-on-one interviews or discussions, and even CSP talks! After the talk, participants will have the chance to send a short video of a talk to the presenter for review and feedback.

Outline & Objectives

The primary objective of this course is to help participants understand some of the qualities of captivating speakers, and to know how they can develop these qualities in themselves. There are many ways to engage an audience, so we’ll also discuss how to utilize your personal qualities and strength to engage a room. Participants will also receive some direct feedback, either by volunteering to give a short talk in front of the group or sending a video to the presenter afterwards.

This course is applicable to anyone who wishes to improve their presentation and public speaking skills, both in front of groups and one-on-one. Specifically, the course will be discussing:
1) Knowing your strengths
2) Knowing your audience
3) Engage your audience
4) Taking advantage of space
5) Using presentation aids wisely
6) (Time Permitting) Presentations from the group

About the Instructor

Duncan Gilles (MS) has been a teacher, faculty manager and teacher trainer for over a decade. For 5 years he managed and trained Kaplan Test Prep’s SAT, GRE, GMAT, LSAT and MCAT faculty in the New England Region and currently manages the teacher pool at the Art of Problem Solving – an online math school. He has experience training and giving feedback to presenters in multiple environments – physical classes, online video-based classes, online text-based classes and one-on-one meetings. He’s been recognized as an Elite Teacher for Kaplan, and has trained or provided feedback for over 200 presenters/teachers in the course of his work.

Relevance to Conference Goals

While we hope that our work will stand on its own, how we present our work when engaging with clients and customers can have a big effect on how well it is received. This course will help participants develop their communication skills, giving them immediately actionable tips on how to better engage with their audience. In addition, participants will have a chance to get personalized feedback on their own work, either by giving a short talk at the conference itself, or sending in a video to the presenter afterwards. In addition to affecting their personal presentations, this course will enable participants to be better representatives of their organizations.

Software Packages

None.

 
T2 Applying Propensity Score Methods to Observational Studies Using R and SAS
Sat, Feb 17, 2:00 PM - 4:00 PM
Eugene
Instructor(s): Wei Pan, Duke University

Download Handouts
Observational studies are common in applied settings but pose threats to the validity of causal inference due to selection bias in the data. Propensity score methods have been increasingly used as a means of reducing selection bias to enhance the causal claims. A training course on the application of propensity score methods to observational studies using commonly used statistical software would be beneficial for applied statisticians and researchers to improve the quality of their observational studies. With this objective, the proposed course will introduce basic concepts and practical issues of propensity score methods, including matching, stratification, and weighting; the instructors will facilitate hands-on activities of applying propensity score methods to observational studies with real-world examples using R and SAS. No prior knowledge of propensity score methods or computer programming is required. Participants are encouraged to bring their own laptop computers for hands-on activities.

Outline & Objectives

Outline:

(1) Mini-lecture on:
• Basic concepts of propensity score methods.
• Various strategies of outcome analysis after matching, stratification, and weighting.
• Issues and developments in propensity score methods.

(2) Demonstration and hands-on activities:
• Demonstrating propensity score matching, stratification, and weighting using R and SAS with real-world data.
• Evaluating covariate balance for propensity score matching.
• Estimating treatment effects after matching, stratification, and weighting.
• Interpreting the results from statistical software for propensity score methods.

Objectives:

(a) Understand the basic concepts and practical issues of propensity score methods.
(b) Discuss why, when, and how to apply propensity score methods in applied settings.
(c) Understand the limitations of propensity score methods.
(d) Learn how to use R and SAS for propensity score methods on observational data.
(e) Interpret the results of propensity score methods.

About the Instructor

Dr. Wei Pan is Associate Professor and Director of the Research Design and Statistics Core at Duke University School of Nursing. Propensity score methods are one of his major research interests. He has published and presented numerous articles on propensity score methods in the past 10 years. Dr. Haiyan Bai is Associate Professor of Quantitative Research Methodology at the University of Central Florida. Her research areas include propensity score methods, resampling methods, research design, and measurement and evaluation. She has published many journal articles on propensity score methods in the past 10 years. Both Drs. Pan and Bai have provided more than 10 professional workshops and training courses on propensity score methods at national conventions, such as annual meetings of the American Statistical Association, the American Public Health Association, the American Evaluation Association, and the American Educational Research Association. They also recently published a book entitled, "Propensity Score Analysis: Fundamentals and Developments."

Relevance to Conference Goals

In this course, the participants will be able to understand why, when, and how to apply propensity score methods to observational studies in applied settings and implement propensity score methods in R and SAS. Through step-by-step hands-on activities on real-world examples or their own research data, participants will be able to produce actual analysis results with graphical and statistical presentations and learn how to interpret them. This course will indeed benefit the participants applying propensity score methods using statistical software as a best practice in statistical analysis, design, and consulting. This course will also increase participants’ overall analytical capacities to improve the quality of observational studies in applied settings. This course is appropriate for applied statisticians, researchers, and scientists. It provides opportunities for them to enhance their career development.

Software Packages

R and SAS (or SAS University Edition which along with R is free to public).

 
T3 A Workshop on Validation of Discrete Response Statistical Models
Sat, Feb 17, 2:00 PM - 4:00 PM
Portland
Instructor(s): Raul Eduardo Avelar Moran, Texas A&M Transportation Institute
Count models are widely used to analyze discrete data in various fields. When the intent of the analysis is prediction, model validation is an important step before the model can be offered with confidence to final users. This tutorial will discuss when and why to validate, and will demonstrate model validation techniques specific to discrete response models, such as Poisson and Negative Binomial Generalized Linear Regression Models.

Outline & Objectives

The objectives of the tutorial are:
1. Participants will acquire working knowledge of validation of discrete response models for various applications.
2. Participants will learn to apply three different validation techniques appropriate for discrete response models.
The tutorial will provide general background and motivation for model validation. A brief description of three validation techniques will be provided. The techniques will be demonstrated using a sample data set. Finally, the tutorial will offer a window for discussion and questions.

About the Instructor

Dr. Avelar is an Associate Researcher Engineer at the Texas A&M Transportation Institute (TTI). His areas of expertise include: transportation safety, roadway and pedestrian operations, data management and processing, and statistical modeling.
Since initially joining TTI in 2012 as a Post-Doctoral Research Scientist, Dr. Avelar has been involved in more than 40 transportation-related research projects for state and federal sponsors in the United States. His role in the vast majority of these projects has been as a transportation statistician, with main responsibilities for data management and statistical analyses.
Dr Avelar is the recipient of the 2016 Patricia F. Waller Paper Award from the Transportation Research Board (TRB) for best paper in the area of safety and system users; the 2016 Outstanding Paper Award from the TRB Committee on Safety Data, Analysis and Evaluation ANB20; the 2015 D. Grant Mickle Award from TRB for best paper in operations and maintenance; and the 2014 Outstanding Paper Award, from the TRB Committee on Pedestrians ANF10.

Relevance to Conference Goals

Data Science and big data rely on discrete response models to develop predictions. Although there is generally good understanding of why and how to use discrete response predictive models, validation of these models is often underrated or skipped altogether. This tutorial will make the case for why we validate and will provide tools to perform validation for discrete response models.

Software Packages

R, R Studio.

 
T4 Tools for Connecting R, SAs, and Stata to Word: A Practical Approach to Reproducibility
Sat, Feb 17, 2:00 PM - 4:00 PM
Salon D
Instructor(s): Abigail S. Baldridge, Northwestern University; Leah J. Welty, Northwestern University

Download Handouts
Reproducibility, wherein data analysis and documentation is sufficient so that results can be recomputed or verified, is an increasingly important component of statistical practice. “Weaving” tools such as R Markdown facilitate reproducibility by combining narrative text and analysis code in one plain-text document, but are of limited use when manuscripts or reports must be generated in MS Word (e.g. due to journal requirements or client preference). This course will: (1) summarize how weaving tools create Word documents, and the ensuing limitations; and (2) introduce an alternate approach using recently released StatTag software. StatTag is a free, open-source program that embeds results (values, tables, figures, or verbatim output) from R, SAS, or Stata directly in Word such that they can be automatically updated if code or data changes. This course is intended for a broad audience; prerequisites are experience preparing documents in Word and conducting analysis in any one of R, SAS, or Stata. The workshop will provide practical, hands-on examples drawn from R, SAS, and Stata, and will include an overview of weaving approaches as well as an introduction to StatTag.

Outline & Objectives

Outline:
1. Introduction to reproducibility
2. Overview of “weaving” tools, worked examples
3. Practical limitations of “weaving” tools with Word
4. An alternate approach using StatTag
5. StatTag instructions and features
6. Hands-on exercises using StatTag to connect Word with R, SAS, and Stata
Objectives:
Provide an overview of reproducibility, focusing on the practical challenges of preparing manuscripts or reports in Word.
Demonstrate how “weaving” tools may be used to generate a Word document. Illustrate the limitations of this approach when documents are subsequently edited in Word.
Present an alternate approach using StatTag software. Provide participants the knowledge and skills to use StatTag to embed values, tables, figures, or raw output from R, SAS, or Stata in Word documents so that: (1) statistical results may be automatically updated if data or models change; (2) the Word document may be edited and formatted as usual without losing the connection to the statistical code.
Provide sample StatTag files for R, SAS, and Stata.
Support a more robust research process by eliminating the need for statistical output to be copied and pasted in to Word documents or reports.

About the Instructor

Leah J. Welty, PhD, Associate Professor in the Department of Preventive Medicine-Biostatistics at Northwestern University, directs the Biostatistics Core Resources within the Northwestern University Clinical and Translational Sciences Institute. She is also the president of the Association of Clinical and Translational Statisticians. She has led the development of StatTag (www.stattag.org), and in addition has delivered 8 invited talks and published one manuscript on reproducible research.

Abigail S. Baldridge, MS, Biostatistician in the Department of Preventive Medicine at Northwestern University, is a statistician and project manager for several studies in cardiovascular epidemiology in addition to helping develop and test StatTag. She teaches an MPH course entitled Programming for Statistical Analysis.

Both instructors have extensive, first-hand experience trying to conduct reproducible research while preparing manuscripts in Word: not only do all of their collaborators prefer Word’s familiarity and editing features, but many of the medical journals they publish in require or strongly recommend that manuscripts be submitted in Word.

Relevance to Conference Goals

This workshop was designed to address an important challenge for applied statisticians: how to ensure reproducibility when circumstances require using Word to prepare and edit reports or manuscripts.
Participants will enhance their professional skills by learning how to integrate document preparation in Word with statistical analysis: no longer will a correction to a dataset or change in model parameters entail re-copying results in to a Word document. The workshop will present two approaches – “weaving” and StatTag – with particular focus on the latter, which was recently released and works with R, SAS, or Stata.
The tools presented encourage collaboration and open communication in addition to reproducibility. For example, StatTag provides a link between the statistical results presented in a Word document and the statistical code and data that generated them: double-clicking on any “tagged” result in the Word document pulls up a dialog box displaying the statistical code that created the result. Statisticians and their collaborators may work separately on statistical code and Word documents, but use StatTag to maintain connections between the two.

Software Packages

Participants will need a computer running Windows or macOS with Microsoft Word (2010 or higher for Windows, 2016 for macOS) and one of R v3.2 or higher, Stata 14 or higher or SAS 9.4 or higher. They will need to be able to download and install software (for Windows, this requires “Administrative” rights).
The workshop will use StatTag software, which is free, open source, and publically available at http://stattag.org