ASA Newsroom > Joint Statistical Meetings Newsroom
JSM Late Breaking & Highlighted Sessions
Note: If you are interested in interviewing any of these presenters, please contact Jeffrey@amstat.org.
- The Role of Statisticians in Healthcare Reform
- Predictive Analytics and Data Mining: The Software-Makers
- Challenges of Data Analysis in Transportation
- The Will of the People and the Luck of the Draw: Using Statistics to Limit the Risk of Wrong Electoral Outcomes
- Statistics and the 2012 Elections
- Forecasting When You Suspect 'This Time Is Different': A Typical Day in the Life of a Macroeconomic Forecaster (Economic Outlook Speaker Luncheon)
- Supporting National Security Policy with Science, Technology, and Analysis
- AccuScore: Don't Bet on It
- Statistics and Climate
- A Study of the Quality of Family Life for Business Executives (Oral Poster Presentation)
- Climate, Health, and Vulnerability in Urban Populations
- Bright Lights, Risky Business: A Longitudinal Study of Broadway Show Success
- Growing the Business with Data-Driven Decisions
- Measuring Advertising Return on Investment
- The Perils of Polling
- Using Social Media to Engage and Motivate the Modern Statistics Student
- Adaptive Designs for Vaccine Clinical Trials
- Rising to the Challenge: Preparing Educators and Society for the Common Core State Standards
- Comparing the Suicide Rates of Veterans and Non-Veterans
- Water and Climate
- Big Data: Research and Training Challenges
The Role of Statisticians in Healthcare Reform
Speakers will present on various aspects of healthcare reform: performance measurement, cost measurement, and informing care choices for patients. Statisticians have developed population-based models that help identify and support value in health care spending by predicting total cost and its components (eg, primary care spending). It is likely that causal modeling and recent innovations in observational analysis developed by statisticians will be key in the next wave of personalizing medicine.
Speakers: Measuring Performance- Marc Elliott, RAND Corporation; Measuring Cost- Arlene S. Ash, University of Massachusetts Medical School; Developing the Evidence Base- John Adams, RAND Corporation. Thomas R. Belin, University of California at Los Angeles is the discussant. Chair: Steven Cohen, Agency for Healthcare Research and Quality
Monday, July 30, 2 pm, Hilton San Diego Bayfront, Room Sapphire MN
Predictive Analytics and Data Mining: The Software-Makers
Senior executives of major predictive analytics and data mining software firms will present their perspectives on "big data." Big data is a rapidly evolving field - here is your chance to hear from these decision-makers about their own plans, where they think predictive analytics is headed, and what is likely to surprise us. The format will be a dynamic panel-style question/answer session, moderated by industry specialist Dean Abbott (Abbot Analytics), noted for his vendor-neutral workshops on data mining.
Panelists: Daniel Fylstra, Founder and President, Frontline Systems; Ralf Klinkenberg, co-founder and Chief Business Development Officer of Rapid-I, Germany; Dan Steinberg, Founder and President of Salford Systems, San Diego; David Smith, VP, Marketing at Revolution Analytics; Thomas Hill, VP, Analytic Solutions at Statsoft, Tulsa; and Wayne Thompson, Analytics Product Manager, SAS Institute, North Carolina. Audience questions are accepted in advance (email the organizer, firstname.lastname@example.org) and will be worked into the program.
Wednesday, August 1, 2 pm, Hilton San Diego Bayfront, Sapphire D
Challenges of Data Analysis in Transportation
"When you have small data sets and simple questions, the problem is like stitching up a small cut. When you have large data sets and complicated questions, it's more like brain surgery."
As society becomes more data dependent and increasingly data rich, it is important to take a look at what data is being used for analysis and where it originates. As the amount of data increases, data management and data quality become problems in themselves. No longer can one person look over and analyze the data; when there are lots of data, many analyses can be done, and many more stories can be told. For example, in 2005 there were 43,510 highway fatalities, and that number had been stable since 1990. But then it started to drop, reaching approximately 32,788 in 2010, according to the NHTSA. This 25-percent drop is enormous and largely unexplained. A portion of it may be due to the recession (people driving less), or increased use of graduated licensure for teen drivers, or improved safety technology in automobiles and on the roads. Statistical analysis can discover how much credit is due to each of these factors.
This session addresses some of the challenges researchers face as statistics adapt to serve an increasingly data-reliant society. Four speakers from the government and private sector will discuss some of these challenges with respect to the transportation industry. Discussions will focus on new challenges in transportation statistics, including the need to fully explain the large and sudden drop in automobile fatalities in the last five years, the necessity to improve seat belt use survey data quality, multiple imputation of linked data and assessing the quality of data with methods to overcome the difficulty of working with not-so-clean data.
Panelists: David Banks, Duke University; Fan Zhang and Amrut Champaneri, US Dept. of Transportation; and Cliff Spiegelman, Texas A&M University. Organizer and Chair: Jenny R Guarino, U.S. Department of Transportation/RITA
Monday, July 30, 10:30 AM, Room 28E Convention Center
Statistics and the 2012 Elections
Participants will present papers on various aspects of the role of statistics in the upcoming elections. Topics on the agenda are:
- Decisionmaking about Redistricting - w ww.amstat.org/meetings/jsm/2012/onlineprogram/AbstractDetails.cfm?abstractid=303895
- Methodological Aspects of Pre-Election Polling - www.amstat.org/meetings/jsm/2012/onlineprogram/AbstractDetails.cfm?abstractid=303901
- Preventing Fraud versus Preventing Voting - www.amstat.org/meetings/jsm/2012/onlineprogram/AbstractDetails.cfm?abstractid=303988
- Statistical Estimation and Election Night Decisionmaking - www.amstat.org/meetings/jsm/2012/onlineprogram/AbstractDetails.cfm?abstractid=303904
Speakers: Thomas R. Belin, University of California at Los Angeles; Michael W. Traugott, University of Michigan; Arlene S. Ash, University of Massachusetts Medical School; John Lamperti, Dartmouth College; Clyde Tucker, CNN Consultant
Chair & Organizer: J. Michael Brick, Westat
Wednesday, August 1, 10:30 am to 12:20 pm, Room Sapphire EF, Hilton San Diego Bayfront
The Will of the People and the Luck of the Draw: Using Statistics to Limit the Risk of Wrong Electoral Outcomes
This session brings together senior election officials, statisticians, and election integrity advocates to discuss legislative issues, recent theoretical developments, and pilots of election auditing. One topic will be risk-limiting audits, which have a guaranteed minimum probability of correcting wrong electoral outcomes. California and Colorado have laws authorizing or requiring such audits and grants from the Election Assistance Commission to support pilots. The ASA also has endorsed Risk-Limiting Post Election Audits.
Panelists: Jennie Bretschneider, California Secretary of State, Neal Kelley, Orange County Registrar of Voters, Pamela Smith, Verified Voting Foundation, Carlsbad, CA
Monday, July 30, 10:20 am, CC, Room 31B
Forecasting When You Suspect 'This Time Is Different': A Typical Day in the Life of a Macroeconomic Forecaster (Economic Outlook Speaker Luncheon)
Real macro-economic forecasting is a personal tug-of-war between extrapolative time series forecasts offered by the left side of the brain and the intuition that "this time is different" offered by the right side. The left-dominant forecasters are the number crunchers who offer forecasts that are either simple or elaborate versions of regression toward the mean. Left-dominant forecasters may become wealthy but they do not become famous. The famous forecasters are the right-dominant macro gurus who offer stories not numbers. Their lack of numbers allows these macro gurus to celebrate their successes while not being embarrassed by their errors.
The best place to be is somewhere between the technical forecasters and the gurus, offering sophisticated access to the historical data and also wisdom regarding changes in the system that need to be accommodated to make the forecasts accurate and the stories interesting. I will discuss a number of instances in which I was confronted with a "this time is different" hypothesis. Sometimes I went "right" and sometimes "left," sometimes successfully and sometimes not.
Speaker: Edward Leamer, UCLA
Tuesday, July 31, 12:30 to 1:50 pm, Convention Center, Room 33C
Supporting National Security Policy with Science, Technology, and Analysis
Science, technology, analysis and policy --- where the rubber meets the road for national security. We will discuss the nuances of how innovations in science, technology and analysis can best support national security policy, and how national security policy needs to be modified as a result of innovations in science, technology and analysis.
Speaker: Nancy Spruill, U.S. Office of the Secretary of Defense
Wednesday, August 1, 7-8:15 am, Convention Center, Ballroom D
AccuScore: Don't Bet on It
The Wall Street Journal publishes daily statistical predictions provided to it by AccuScore.com. AccuScore is an online website that charge individuals for their predictions. AccuScore claims to simulate game outcomes 10,000 times for each contest. Sometimes final scores are published while other times probabilities are given for each team's likelihood of winning. Predictions are made for a wide variety of sports, including major league baseball, national league hockey, national league football, national league basketball, along with NCAA college football and basketball. A full season of major league baseball and NFL football was studied and the accuracy of the AccuScore predictions was examined. Linear regression and logistic regression models were fit to the data. Summary statistics were calculated. After analyzing the results, it was determined that there are simply too many variables in sports to accurately predict outcomes. The moral to the story is "save your money."
Jay Schaffer, University of Northern Colorado (Oral poster presentation)
Monday, July 30, 10:30-12:20, Convention Center, Sails Pavilion
Statistics and Climate
There is substantial evidence that the climate is changing, and that human production of greenhouse gases is a leading driver of the change. We look at different ways to model and assess statistically some of the evidence, including estimation of global temperature. In addition we present climate models, outline how to compare them to data, and discuss the uncertainty in regional predictions and how that relates to decision-making.
Speaker: Peter Guttorp, University of Washington
Wednesday, Aug. 1, 8:30 to 10:20 am, Hilton San Diego Bayfront Hotel, Sapphire MN
A Study of the Quality of Family Life for Business Executives (Oral Poster Presentation)
This survey of 2000 business executives in the United States yielded a 17.7% response rate. The questionnaire sought answers to questions about quality of life versus education, work experience, number of children, age, gender and marital status. The data produced several significant differences for the above demographics of business executives. Executives with lower education levels feel their work interferes less with family responsibility than their more educated counterparts. Older executives manage their family responsibilities better than executives younger than fifty-one. Married and widowed executives have better quality of home life than single or divorced executives. Executives that are single are the least satisfied with their careers.
Jonathan Trower, Kris Moore, and Charles Goldwaite, Baylor University
Tuesday, July 31, 2-3:50 pm, Convention Center, Sails Pavilion
Climate, Health, and Vulnerability in Urban Populations
SIMMER is a highly interdisciplinary program designed to assess current and future vulnerability to heat stress in urban environments through integration of physical and social science models. Focused on the city of Houston, the SIMMER project seeks to incorporate data from various sources including heat-related mortality and hospital admissions, a survey of Houston residents, the U.S. Census and other city and state agencies, remote sensing, high resolution climate models, and observed meteorological data. In this talk, we will discuss the SIMMER project and an initial analysis of this data, including assessments of vulnerability and a statistical model for linking mortality to vulnerability, socio-economics, weather, and climate.
Speakers: Tamara Anne Greasby, National Center for Atmospheric Research; Stephan Sain, Geophysical Statistics Project
Thursday, August 2, 10:35 am, Convention Center, Room 33A
Bright Lights, Risky Business: A Longitudinal Study of Broadway Show Success
Using a long panel of nine years data, we investigate the factors that are useful to predict the weekly attendance at Broadway shows. This is accomplished by developing a longitudinal model taking into account the capacity constraints and informative dropout in a hierarchical Bayes framework. The weekly show attendance is modeled using a re-parameterized gamma distribution with parameters representing opening week performance, timing of peak performance and decay rate. Our research indicates that opening week performance is strongly related to the show type, whether a show is a revival, and the number of star actors. Shows that received positive reviews from New York Times take longer time to reach peak performance. Weekly demand declines faster for a play and more slowly for a show that received a positive review from USA Today. The impact of Tony Nominations tends to vary substantially across shows, although generally they tend to shift demand upwards.
Speakers: Lan Nygren, Rider University; Kjell Nygren, ImpactRx; and Jeffrey Simonoff , New York University
Sunday, July 29, 4-5:50 pm, Convention Center, Room 31C
Growing the Business with Data-Driven Decisions
When faced with making decisions on a daily basis, the best decisions result not from perceptions and emotions, but from information derived from valid and objective data. Making Data-Driven decisions part of our DNA enables business growth as well as provides other strategic benefits. This roundtable will discuss Data-Driven Decisions (D3) ingredients, concepts, statistical thinking vs. statistical tools, roadmap, and case studies. We will explore several engineering, operations, and business case studies revealing scenarios where data and/or toolsets were used correctly or incorrectly with key learning points and resulting benefits, including how it presented opportunities for innovation and business growth.
Speaker: Daksha Chokshi,Pratt & Whitney Rocketdyne
Monday, July 30, 12:30 pm, Convention Center, Ballroom D
Measuring Advertising Return on Investment
As online marketers have begun to address a more diverse set of goals, traditional methods of assessing the success of online campaigns have come up short. The ultimate question for many marketers is less about how many people engaged with their marketing and more about whether their efforts persuaded people to do something outside of the online environment. While there are many tools today to assess this kind of impact, they are not scalable, and they often have issues of representativeness and privacy. Advertisers need a more general framework that allows for campaign-level measurement of the off-line impact of all of their campaigns, that assesses both in-home and out-of-home exposure, and that does so in a privacy-safe manner. In this paper the authors propose a general framework for linking campaign-based attitudinal change metrics with off-line outcomes that can be applied to virtually any advertising campaign without tracking individual-level behavior. They provide an example of how they have applied this framework to quantify how advertising for motion pictures impacts opening weekend box office.
Speakers: Sean Bruich and Fred Leach, Facebook Tuesday, July 31, 10:30 am to 12:20 pm, Room Sapphire EF, Hilton San Diego Bayfront
The Perils of Polling
Polling is an important part of the arsenal of a data scientist, and Facebook provides a huge base of users with information that allows us to run quick and targeted surveys. Unfortunately, the price one pays for this is a significant self-selection effect among respondents. We discuss how and when we can come up with good, rigorous conclusions from the numerous biases introduced by this self-selection, as well as some other issues having to do with polling in general in a low-attention-span framework as opposed to a more traditional telephone environment.
Mike Develin, Facebook
Tuesday, July 31, 10:30 am to 12:20 pm, Room Sapphire EF, Hilton San Diego Bayfront
Using Social Media to Engage and Motivate the Modern Statistics Student
In the last 10 years, several articles and case studies have been published about ways in which popular social networking tools such as Facebook, Twitter, and YouTube can be used in educational settings. Little attention has been paid, however, to specific uses of these tools within the statistics classroom. The purpose of this roundtable session is to share ideas about how social media can be incorporated into a statistics course not only to engage and motivate students, but to help students better reason through challenging course material. Discussion will center around both specific uses of social networking tools and important considerations educators should be mindful of before incorporating such tools into their own classrooms. Those interested in this topic are encouraged to bring ideas, experiences, and concerns with social media to the table so all participants can collaborate and learn from each other.
Speaker: Michelle Gaddy Everson, University of Minnesota
Wednesday, August 1, 7-8:15 am, Ballroom D, Convention Center
Adaptive Designs for Vaccine Clinical Trials
Despite the increase of our understanding of host immune responses, the sequencing of pathogen genomes, and other technological advances, important hurdles remain for developing vaccines for a variety of diseases. Developing a vaccine against diseases like the human immunodeficiency virus (HIV), tuberculosis (TB), Malaria etc poses a challenge in vaccine clinical trial studies. Adaptive clinical trial designs allow a trial to be modified in response to data acquired during the study. Such trials would rapidly screen out poor vaccine candidates, enable extended evaluation of promising candidates and provide key information on the immunological basis. This talk will provide various perspectives in utilizing adaptive trial design methods for use in vaccine clinical trials; discuss adaptive designs from regulatory prospective including the recommendations from the FDA guidance. Strength and Weakness of the approach will be discussed through examples of adaptive designs used in vaccine clinical trials.
Speaker: Ghideon Ghebregiorgis, FDA
Sunday, July 29, 2 pm, Room 28E, Convention Center
Rising to the Challenge: Preparing Educators and Society for the Common Core State Standards
The Common Core State Standards for Mathematics (CCSSM, 2010) set the stage for major changes in the mathematical education of K-12 students. The curriculum envisioned by CCSSM proposes higher expectations of students' understanding in both mathematics and statistics, leaving substantial implications for the statistical preparation of teachers. With the adoption of a common mathematics curriculum, assessing and documenting student learning over time will be useful for informing educational policy decisions motivated by the national Race to the Top initiative. However, data from student assessments are at risk of being misused, particularly when they are the sole metric for making high-stakes decisions such as evaluating a teacher's performance. Equipping K-12 teachers with the tools and knowledge needed to help students develop into statistically literate citizens requires statisticians play an active role in teachers' education. Also, statisticians have a role in educating public and political leaders of the limitations of systems evaluating teachers solely on students' test performance and developing methods to better inform the processes of teaching, learning and assessment.
Speakers: Jennifer L Green and Walter W Stroup, University of Nebraska; Gary D Kader, Appalachian State University
Thursday, August 2, 8:30 am, Room Sapphire H, Hilton San Diego Bayfront
Comparing the Suicide Rates of Veterans and Non-Veterans
We hear in the news and media that war veterans are committing suicide. This is a devastating news as soldiers are saving the lives of millions but are having a tough time trying to save their own lives. However, suicide is also common within the general population of the country. In this talk, I will discuss the suicide rates of American War Veterans, specifically if these men are at a higher risk of suicide than non-veterans. The group in focus will be men between the ages of 17 to 34. I will use data from The National Violent Death Reporting System (NVDRS). This data set includes the suicide rates in 16 states, from 2005 to 2008. Standard statistical techniques were used to analyze the data, with the aid of the statistical package SPSS 19.
Speaker: Deepon Bhaumik, Oak Park River Forest High School
Thursday, August 2, 8:50 am, Room 28D, Convention Center
Water and Climate
Organized by Peter Guttorp of the University of Washington, this session features three papers on the following topics: (1) Decline of Arctic Sea-Ice Thickness as Evidenced by Submarine Measurements, which shows that there has been a significant decline in Arctic sea-ice thickness; (2) Trends in US Precipitation Extremes, which examines and reports on some of precipitation behavior based on seasonality and records with very low precipitation; and (3) Spatio-Temporal Climate Structures That Determine Extreme Floods, which presents the synthesis of a body of recent research on the climatic context of extreme floods.
Speakers: Donald Bame Percival, University of Washington; Michael Wehner, Lawrence Berkeley National Laboratory; Upmanu Lall, Mengqian Lu, Tian Zheng, Jennifer Nakamura, Aurelien Schwarz, and Yochanan Kushnir, Columbia University; Andrew Robertson, The International Research Institute for Climate and Society; and Carlos Henrique Ribeiro Lima, Universidade de Brasilia.
Thursday, August 2, 10:30 am -12:20 pm, Room Sapphire A, Hilton San Diego Bayfront
Big Data: Research and Training Challenges
This panel explores the role of the statistical and computational sciences in visualizing and analyzing massive, complex data. In a recent article, the Economist referred to the "Data Deluge" and the effects of this influx of information on science, business, and government. Examples of big data are found in diverse fields including astronomy, finance, social networks, and medicine. With this explosion in the availability of information, faster and more efficient tools are needed to extract meaningful patterns from data. The emphases on improved tools for visualizing and analyzing data and development of efficient simulation algorithms pose a significant challenge to the Statistics community. This panel will bring together leaders from academia, industry, and government to address these challenges. The panel also will discuss workforce challenges -- education and training issues, including the development of new courses and/or curricula.
Panelists: Brian Weaver, Los Alamos National Laboratory; Nandini Kannan, The University of Texas at San Antonio; Emmanuel Yashchin , IBM; Deborah A. Nolan, University of California at Berkeley; Richard De Veaux, Williams College; James L Rosenberger, Penn State University
Tuesday, July 31, 10:30 am - 12:20 pm, Convention Center Room 31A