June 2009 • Issue #384 Amstat The Membership Magazine of the American Statistical AssociationNews • www.amstat.org/publications/amsn
JSM 2009: Plan Now Play Later
ALSO: ASA Endorses CNSTAT’s Principles and Practices for Federal Statistical Agency
NCSU, UM Honored for Bringing Minorities into Publications Agreement No. 41544521 Mathematics B Amstat News JUNE 2009 JUNE 2009 • Issue #384
Executive Director Ron Wasserstein: [email protected] Associate Executive Director and Director of Operations F EATURES Stephen Porzio: [email protected]
Director of Programs 3 President’s Invited Column Martha Aliaga: [email protected] 5 Board Highlights Director of Science Policy Steve Pierson: [email protected] 6 Extra! Extra!
Managing Editor 8 2009 ASA Audit Report Megan Murphy: [email protected] 13 NCSU, UM Honored for Bringing Minorities into Production Coordinators/Graphic Designers Mathematics Melissa Muko: [email protected] Lidia Vigyázó: [email protected] 14 ASA Endorses CNSTAT’s Principles and Practices for Federal Statistical Agency Publications Coordinator Val Snider: [email protected] 15 Caucus for Women in Statistics to Host Breakfast Advertising Manager During JSM Claudine Donovan: [email protected] 16 Algorithmic, Statistical Challenges in Data Analysis Contributing Staff Members Focus of MMDS 2009 Amy Farris • Rick Peterson • Eric Sampson Kathleen Wert • Elizabeth Shwaery
Amstat News welcomes news items and letters from readers on matters of interest to the association and the profession. Address correspondence Caucus for Women in to Managing Editor, Amstat News, American Statistical Association, 732 North Washington Street, Alexandria VA 22314-1943 USA, or email Statistics to Host Breakfast [email protected]. Items must be received by the first day of the preced- During JSM ing month to ensure appearance in the next issue (for example, June 1 for p. 15 the July issue). Material can be sent as a Microsoft Word document, PDF, or within an email. Articles will be edited for space. Accompanying art- work will be accepted in graphics file formats only (.jpg, etc.), minimum 300 dpi. No material in WordPerfect will be accepted. Amstat News (ISSN 0163-9617) is published monthly by the American Statistical Association, 732 North Washington Street, Alexandria VA 22314-1943 USA. Periodicals postage paid at Alexandria, Virginia, and additional mailing offices. POSTMASTER: Send address changes to Amstat News, 732 North Washington Street, Alexandria VA 22314-1943 USA. Send Canadian address changes to Station A, P.O. Box 54, Windsor 20 Feds in Space ON N9A 6J5; [email protected]. Annual subscriptions are $50 per year for nonmembers. Amstat News is the member publication of the ASA. For annual membership rates, see www.amstat.org/join or contact 23 CHANCE Highlights ASA Member Services at (888) 231-3473. 25 ASA Online Journals Provide the Tools You Need American Statistical Association 732 North Washington Street 26 JASA Highlights Alexandria, VA 22314–1943 USA (703) 684–1221 • FAX: (703) 684-2036 28 Norwood Award Proposals Due June 29
ASA GENERAL: [email protected] 29 NC State Dedicates Math, Stats Building ADDRESS CHANGES: [email protected] 30 Library Slipped? AMSTAT EDITORIAL: [email protected] ADVERTISING: [email protected] 31 staff Spotlight Web Site: www.amstat.org 33 ASA-SIAM Series Printed in USA © 2009 American Statistical Association VISION STATEMENT To be a world leader in promoting statistical practice, applications, and research; publishing statistical journals; improving statistical education; and advancing the statistics profession Feds in Space p. 20 MISSION STATEMENT Support excellence in statistical practice, research, journals, and meetings. Work for the improvement of statistical education at all levels. Promote the proper application of statistics. Anticipate and meet the needs of our members. Use our discipline to enhance human welfare. Seek opportunities to advance the statistics profession. Column Contributors Funding Opportunities Increased Funding for Science: What Does It Mean for Statistics? p. 35 This column highlights research activities that may be of interest to ASA members. These brief articles include information about new research DEPARTMENTS solicitations and the federal budget for statistics. Comments or sugges- tions for future articles may be sent to [email protected]. 42 Meetings IOLs Open Eyes to the Basics Contributing Editor Program-at-a-Glance Keith Crank has a BS in mathematics education Washington, DC, Trivia and an MS in mathematics from Michigan State University and a PhD in statistics from Purdue No Car, No Problem: Getting Around During JSM University. Prior to joining the ASA, he was a program RTI Supports Cox Race, Honors Namesake officer at the National Science Foundation, primarily in the probability program. Practical Forum Brings Together Industry, Academic Statisticians
Crank Science Policy News Thinking Outside the Urn p. 37 JSM 2009: This column is written to inform ASA members about what the ASA is doing to Plan Now promote the inclusion of statistics in policymaking and the funding of statistics research. To suggest science policy topics for the ASA to address, contact ASA Play Later Director of Science Policy Steve Pierson at [email protected]. p. 42
Contributing Editor Arlene Ash is a research professor at Boston University’s School of Medicine. A past chair of the ASA’s Subcommittee on Electoral Integrity, she has testified 52 Government News before the state legislature of Massachusetts on statis- Saying Goodbye to SOI tical issues in elections and in a court case in Florida in 2000.
57 Education CAUSE Presence Holds at JMM ‘10 Ash
ASA Board Endorses Recommendations for Contributing Editor Graduate Programs in Statistics Philip B. Stark is a professor of statistics at the University ? of California at Berkeley. He has helped develop meth- ods for risk-limiting election audits, served on a post- Washington, DC, election audit standards panel for the California secre- Trivia tary of state, and conducted four risk-limiting audits in a p. 46 California—the only such audits to date. Triv Stark Master’s Notebook A Learning Opportunity for Statisticians p. 41 MEMBER NEWS This column is written for statisticians with master’s degrees and highlights areas of employment that will benefit statisticians at the master’s level. Comments 60 People News and suggestions should be sent to Keith Crank, ASA assistant director for research and graduate education, at [email protected]. 65 Committees 67 Section News 80 Chapter News Contributing Editor 81 Calendar of Events Amarjot Kaur is director of clinical biostatistics at Merck Research Laboratories. She earned her PhD in statistics 87 Professional Opportunities from Panjab University, Chandigarh, India, and was the recipient of a national-level research fellowship.
cover design by Melissa Muko Kaur PRESIDENT’S INVITED COLUMN
In January, I shared information about three initiatives undertaken this year to address issues raised in the ASA strategic plan and promised to provide updates to members as work progressed. In this month’s column, Keith Ord, ASA treasurer and leader of the financial status initiative, summarizes his work group’s deliberations and provisional recommendations. The ASA’s spending policy and reserves are essential to the financial health of the association and, thus, its ability to carry out its mission. I thank Ord and the members of his work group for their time and attention to these important issues that will affect the association for many years to come. ~ Sally C. Morton, ASA President
Saving for a Rainy Day Keith Ord, ASA Treasurer
ecent economic events have left us only too We began our deliberations by elucidating a set well aware of the need for careful financial of operating principles for spending policy. Most of planning. Further, we recognize our fund of these statements will have a familiar ring to them. Rgood ideas usually outstrips the dollars to pay for If implemented in both letter and spirit, we believe them. As ASA President Sally Morton put it in her they will provide useful guidelines for the ASA in charge to our working group, “After personnel, the the years ahead. Each activity or project undertaken most important decisions made by an organization by the association (e.g., JASA) was first classified as are its fiscal decisions. Decisions on how to spend income-producing, self-supporting, or needing sup- reflect an organization’s values and should reflect port. Projects were then grouped into one of nine its goals. Further, the amount of funds maintained broad categories, or “buckets” (e.g., publications), as reserves determines the flexibility an organization so the overall health of major activities of the ASA will have to weather economic ups and downs could be monitored in a flexible, broad framework. and to make investments in future programs.” Spending policy is then defined at the level of indi- Accordingly, we were asked to develop (a) a spend- vidual projects, whereas reserve policy is specified at ing policy and (b) a long-term fiscal policy for the the bucket level. size of ASA reserves. We began by examining recent ASA Board deci- Spending Policy sions so we could understand the criteria currently The general principles for spending policy are applied to the assessment of new projects. We also the following: benchmarked the reserve policies of professional organizations in related fields. The outcome of this — There should be an overall rolling budget plan second exercise suggests our sister organizations are looking ahead for, say, five years that would include struggling with similar issues and do not have com- net income (or net cost) projections for each project. plete solutions either. As the financial health of the ASA is a matter of concern to us all, the purpose of — Proposals for new projects should provide both this interim report is to share the progress made to an assessment of the project’s importance to the date by the group and to invite suggestions or com- ASA and a detailed business plan. This plan should ments. These can be sent to [email protected]. include a market analysis, possibly based on a mem- In recent years, the ASA has experienced a bership survey, or at least responses from interested decline in both library and personal subscriptions sections and chapters. These new projects should be to the journals, a small but steady decline in full evaluated in the context of their cost and revenue membership, and, more recently, a decline in adver- implications, including possible cost-sharing and tising revenues (as a result of the economic down- profit-sharing with sections or chapters. turn). Historically, the ASA has not used invest- — New projects should normally be approved for a ment income to pay operating expenses. Such a step trial period of, say, three to five years. Toward the should be avoided, save in extreme circumstances. end of the trial period, the board would be pro- Our brief was not to consider new revenue streams, vided with a report on the progress of the program but we recognize action is also needed on that side and would then determine whether to support its of the balance sheet. continued operation.
JUNE 2009 Amstat News 3 — All projects should be subject to periodic review and reassessment of priorities. Projects The group was greatly assisted by the that fail to meet cost and income expectations or active participation of past and present whose relevance declines should be considered as members of the Finance Committee. candidates for termination. This should be clear- ly communicated to project sponsors at the time Members of the Working Group a project is approved by the board. The board Karen Bandeen-Roche, Finance should set up a working group to evaluate such Committee member cases and make recommendations. Christy Chuang-Stein, ASA vice president In brief, we want the ASA to be able to maintain an open door to new initiatives, but to be somewhat Keith Ord, ASA treasurer and group convener less forgiving of projects that do not live up to their Steve Porzio, ASA associate executive director potential after a reasonable trial period. Ron Wasserstein, ASA executive director Reserves Policy The reserves of the association should be sufficient to George Williams, Finance Committee enable it to operate even after sustaining major finan- member (and former ASA vice president) cial setbacks, some of which by their very nature are Other Members of the Finance Committee unforeseeable. Some activities, such as publications, can reasonably be expected to generate positive net Richard Goldstein income, as inflows and outflows tend to rise and fall Michael Kutner together. Working in this way, we were able to identify four net cost groupings: Administration, Education, Jean Opsomer Support of ASA Communities (sections and chap- ters), and Other Programs, including special projects. Frank Shen The other five buckets were regarded as reliable net income producers: Advertising, Grants, Meetings, Ji Zhang Membership, and Publications. We think reserves should be maintained to cover the four net cost group- ings in accordance with the following principles: — No reserves are required for reliably profitable — Reserves should be sufficient to enable the ASA activities, but each net cost grouping needs to be to continue to operate over an extended period of fully covered. financial hardship. — The size of the reserves should take into account Following these guidelines, we are recommend- both increasing costs over time and due allowance ing that the financial reserves should be sufficient to for increases in the cost of living. provide coverage equivalent to two years of opera- tions of the net cost groupings. The size of the gen- — The reserves need to be adjusted over time to eral reserves should be based on the average level of reflect new commitments such as hiring a perma- expenditures over the last three years, adjusted for nent new staff member. inflation using the Consumer Price Index and subject n — Insofar as is possible, year-to-year increases in to additional adjustments for any new programs. the level of reserves should be provided from investment income.
4 Amstat News JUNE 2009 Board Highlights Policy, Planning, Strategic Plans Top Board’s Agenda
olicy matters, planning, and implementation — As President-elect, Sally Morton established of strategic plans were among the important three work groups, each related to one of the planks activities of the Board of Directors during its of the ASA’s Strategic Plan. These groups reported PApril 3–4, 2009, meeting. Here are the highlights: to the board: • The 2009 Strategic Plan Work Group on — The board appointed a host of new editors to Meetings, chaired by ASA Vice President assume duties beginning in 2010: • Joe Verducci, editor of Statistical Analysis and Data Mining Board of Directors • John Gabrosek, editor of the Journal of Statistics Education Sally Morton, President • Richard Levine, editor of the Journal of Sastry Pantula, President-elect Computational and Graphical Statistics Tony Lachenbruch, Past President • Hal Stern, editor of the Journal of the American Statistical Association, Applications and Case Alicia Carriquiry, Vice President (third-year) Studies Nat Schenker, Vice President (second-year) • Jonathan Wright and Keisuke Hirano, editors of the Journal of Business & Economic Statistics Christy Chuang-Stein, Vice President (first-year) • Joe Heyse’ appointment as editor of Statistics in Biopharmaceutical Research was extended for one Geert Verbeke, International Representative year, through 2010 Karen Kafadar, Publications Representative — The board endorsed the principles and best prac- tices set forth in Principles and Practices for a Federal John Boyer, Council of Chapters Statistical Agency, 4th ed. (see Page 14) Representative (third-year) Susan Hilsenbeck, Council of Chapters — The board endorsed recommendations from a Representative (second-year) panel of faculty regarding graduate programs in sta- tistics education. (see Page 59) David Marker, Council of Chapters Representative (first-year) — Steve Pierson, ASA director of science policy, led a discussion with the board about planned activities Tom Santner, Council of Sections for the JSM congressional visits, an opportunity for Representative (third-year) members to meet with congressional staff to discuss specific issues (to be determined later this spring). David Banks, Council of Sections Details can be found at www.amstat.org/meetings/ Representative (second-year) jsm/2009/index.cfm?fuseaction=capitolhill. Jeri Mulrow, Council of Sections Representative (first-year) — The board endorsed the recommendation of the Membership Recruitment and Retention Keith Ord, Treasurer Committee to set the price for annual dues at $50 for all K–12 teachers. Ron Wasserstein, Executive Director
JUNE 2009 Amstat News 5 Nat Schenker, presented its report to the board. — Speaking of strategic planning, Sastry Pantula, The work group recommended forming a plan- ASA president-elect, led the board in a preliminary ning committee to initiate a conference on discussion of initiatives for 2010, based on the applied statistics in 2011 or 2012. A planning ASA’s Strategic Plan. Pantula indicated his intention committee is being formed, and its first meeting to focus on four of the planks of the plan during will be held at JSM. 2010: membership growth, visibility and impact in decisionmaking, public awareness, and education. • The other two strategic plan work groups, Planning for 2010 will continue through the spring Organizational Efficiency (chaired by Bob and early summer. Rodriguez) and Reserves and Spending Policy (chaired by Keith Ord), presented progress — Karen Kafadar, publications representative, led reports. Both are scheduled to make final reports a lively discussion about the future of electronic later this spring and summer. publications at the ASA. The discussion focused on a set of recommendations to the board from the Committee on Publications. The board agreed that (1) publications should move to electronic as the pri- mary medium, with paper secondary; (2) the ASA’s publications should aspire to be influential and the ASA should work to expand electronic access while Extra! Extra! maintaining the financial viability of its publishing program; and (3) the ASA should investigate the creation of a statistics portal. A work group, chaired hose words—once shout- by Committee on Publications Chair David Scott, ed by people hawking has been assembled to take further steps and bring a newspapers on street cor- proposal to the board in August. Tners everywhere—have largely gone the way of the dodo. But — Keith Ord, ASA treasurer, reviewed the ASA’s you can receive the present-day financial summary for 2008. The board also received equivalent in your email for every the 2008 audited financial report. journal the ASA publishes. Simply go to http://pubs. — The board reviewed 10 proposals for funding amstat.org, click on the journal from the Member Initiatives Program and fund- you’re interested in, and then ed three of them: Census@School in the USA, click on “Sign up for TOC Statisticians Influence Climate Change Deliberations Alerts.” With that, the table of on Capitol Hill: A Case Study, and Increasing the contents of every issue will land Value of the First Course in Statistics. in your in-box with links to the issue and each article. Even if — The Council of Sections Governing Board and you don’t subscribe to a particu- the Council of Chapters Governing Board each lar journal, go ahead and get the reported on their activities since December 2008. TOC Alert so you’ll know what’s Both councils met in Alexandria at the ASA office going on in the most important in February. statistical journals in the world. Get ahead of the curve. Get — The board approved launching an annual fund TOC Alerts! n drive, scheduled to begin in June of 2009. n
6 Amstat News JUNE 2009 JUNE 2009 Amstat News 7 2009 ASA Audit Report
8 Amstat News JUNE 2009 2009 ASA Audit Report
JUNE 2009 Amstat News 9 2009 ASA Audit Report
10 Amstat News JUNE 2009 2009 ASA Audit Report
JUNE 2009 Amstat News 11 2009 ASA Audit Report
12 Amstat News JUNE 2009 NCSU, UM Treasures from Honored for Bringing ASA Minorities into Archives Mathematics
he Department of Statistics at North Carolina State University and Department of Mathematics at the University of TMississippi were honored recently as Mathematics Programs That Make a Difference by the American Mathematical Society. “These two departments have outstanding records in recruiting members of under-represented groups and mentoring them to successfully complete their graduate degrees,” said Alejandro Adem of the University of British Columbia, chair of the selection committee for the award. “The AMS Committee on the Profession was extremely impressed with their accomplishments, which can serve as a model for other departments in the United States.” The Department of Statistics at NCSU has made diversity of students and faculty a top priority. Out of 40 faculty members in the department, 11 are female, three are African-American, and two are Hispanic. In the past 10 years, 15 minority students have earned master’s degrees and two have earned PhDs. The department has about 160 graduate stu- dents, including nine African-Americans and four Hispanics; more than 50% are female. The southern United States is home to many of the nation’s African-Americans, and yet the universities in that part of the country have not historically been large producers of African- American PhDs in mathematics. But the University of Mississippi has started to reverse this trend, becoming August 1972 letter to ASA President-elect Jerome a national leader in nurturing and mentoring Cornfield reporting on the implementation of the African-Americans in doctoral study. Over the past study of future goals of the ASA. American Statistical decade, 11 African-Americans have earned math- Association Records, MS 349, Special Collections ematics doctorates from Ole Miss. That the univer- Department, Iowa State University Library. sity was once a symbol of educational segregation in the United States gives this success story The URL for the archive is www.lib.iastate.edu/spcl/ special importance. manuscripts/MS349.html. To read about these programs and the award, visit www.ams.org/prizes/make-a-diff-award.html. n If you have questions, email [email protected] or call (515) 294-6672.
JUNE 2009 Amstat News 13 ASA Endorses CNSTAT’s Principles and Practices for Federal Statistical Agency
The Principles and Practices is an invaluable stan- “dard of guidance for leadership and staff of every About the National Academies’ federal statistical agency. It helps sort out the Committee on National Statistics evolving challenges to accomplishing their mis- The Committee on National Statistics (CNSTAT) sions in difficult times. Perhaps most important, in was established in 1972 at the recommendation my experience, the book is a place to repair when of the President’s Commission on Federal Statistics to improve the statistical methods and one is pressed to some improper expediency. It is information on which public policy decisions also actually appreciated by higher-level political are based. The committee serves as an appointees who want to ‘do the right thing’ in an integrative force for the nation’s decentralized federal statistical system through its wide- awkward situation, but have not been prepared by ranging studies on statistical applications their experience to see what that is. The endorse- in public policy and its ongoing review of ment by the American Statistical Association adds statistical policy activities of the executive branch and Congress. the force of the size and wisdom of that leading national organization of the profession. The committee convenes expert panels to conduct studies on the data and methodology ~ C. Louis Kincannon, U.S. Census” Bureau, needed to improve our understanding of Director 2002–2008 and Deputy Director 1982–1992 the U.S. population, economy, environment, public health, crime, education, immigration, poverty, welfare, terrorism, and other public policy topics. The committee also furthers the application of statistics to better implement and evaluate federal programs and works to t a recent meeting, the ASA Board of improve statistical methods for application to Directors voted to endorse Principles and public affairs; private sector decisionmaking; and Practices for a Federal Statistical Agency social, economic, and other scientific research. A(P&P), 4th Ed. The publication is a project of the Committee on National Statistics (CNSTAT), a For more information about the committee, standing unit of the National Academy of Sciences, its reports, and its projects, visit www7. which issued the first edition in 1992. nationalacademies.org/cnstat. “Principles and Practices is the authoritative guide for federal statistical agencies,” said Sally Morton, ASA president and CNSTAT member. “It builds on years of experience observing, advising, P&P was developed in response to requests that and interacting with the federal statistical agencies CNSTAT provide advice on what constitutes an and represents a great wealth of accumulated wis- effective federal statistical agency. Second and third dom. The endorsement of the ASA Board signals editions of the publication were issued in 2001 and our strong support of the federal statistical agencies, 2005, respectively, updating the document to pro- and we will use P&P to provide guidance for the vide current editions to newly appointed cabinet ASA’s science policy initiatives.” secretaries at the beginning of each presidential
14 Amstat News JUNE 2009 administration. The fourth edition was published — They must achieve and maintain a strong posi- early this year. P&P has been widely cited and used tion of independence from the appearance and by the Office of Management and Budget, the U.S. reality of political control General Accountability Office, and many federal agencies. Principle 4 was added for the fourth edition to In the new edition, CNSTAT presents and com- underscore the need for statistical agencies to con- ments on four basic principles that statistical agencies trol their own programs so they can be impartial must employ to successfully carry out their mission: and trusted by all. The book also discusses 11 important practices that — They must produce objective data that are rel- enable statistical agencies to live up to the four princi- evant to policy issues ples. These practices include a commitment to qual- ity and professional practice and an active program — They must achieve and maintain credibility of methodological and substantive research. among data users Principles and Practices for a Federal Statistical Agency, 4th Ed., is available from The National — They must achieve and maintain trust among Academies Press at www.nap.edu. n data providers
Caucus for Women in Statistics to Host Breakfast During JSM
he Caucus for Women in Statistics will host a breakfast on Monday, August 3, from 6:30 a.m. to 8:00 a.m. at the Embassy Suites, 900 10th St. NW (located between I and K Streets) in TWashington, DC. The breakfast is open to both members and non- members of the caucus and those attending or not attending JSM. There will be the traditional roundtables with specific topics for discussion, as well as tables with no specific topics of discussion. The meal will be the full breakfast buffet offered by the Embassy Suites. Planned topics for discussion include work/life balance, negotiating salary and other job benefits, comparison of statisticians’ jobs across set- tings, and legal rights and wrongs in the workplace. Additional round- tables may be added at a later date, and topics are subject to change. The cost of the breakfast is $13 for members, $15 for nonmembers, and $8 for full-time students. To reserve a place at the breakfast, email Anna Nevius at [email protected] by July 22, 2009. For more information about the Caucus for Women in Statistics, go to http://caucusforwomeninstatistics.com. n
JUNE 2009 Amstat News 15 Algorithmic, Statistical Challenges in Data Analysis Focus of MMDS 2008 Michael W. Mahoney, Lek-Heng Lim, and Gunnar E. Carlsson
was originally motivated by the to tabulate and process the data at complementary perspectives hand to find interesting patterns, brought by the numerical linear rules, and associations. A differ- algebra and theoretical computer ent view of the data, more com- science communities to matrix mon among statisticians, is one of algorithms in modern informatics a particular random instantiation applications. of an underlying process describ- ing unobserved patterns in the Diverse Approaches to world. In this case, the goal is Modern Data Problems to extract information about the Graph and matrix problems were world from the noisy or uncertain common topics for discussion, data observed. largely because they arise natu- The two perspectives need not rally in data mining, machine be incompatible: Statistical and learning, and pattern recognition. probabilistic ideas are central to For example, a common way to much of the recent work on devel- model a large social or informa- oping improved approximation tion network is with an interac- algorithms for matrix problems. tion graph model, G = (V,E), in Much recent work in machine which nodes in the vertex set V learning draws on ideas from represent “entities” and the edges both areas, and, in boosting the in the edge set E represent “inter- regularization parameter (i.e., the actions” between pairs of entities. number of iterations), also serves Alternatively, these and other data as the computational parameter. sets can be modeled as matrices, Given the diversity of possible he 2008 Workshop on since an m x n real-valued matrix perspectives, MMDS 2008 was Algorithms for Modern A provides a natural structure for organized loosely around six Massive Data Sets encoding information about m hour-long tutorials that intro- T(MMDS 2008) was held at objects, each of which is described duced participants to the major Stanford University last June. The by n features. themes of the workshop. goals of MMDS 2008 were (1) to It is worth emphasizing the explore novel techniques for very different perspectives that Large-Scale Informatics: modeling and analyzing massive, have historically been brought Problems, Methods, high-dimensional, and nonlin- to such problems. A common and Models early structured scientific and view of the data, in particu- On the first day, participants Internet data sets and (2) to bring lar among computer scientists heard tutorials by Christos together computer scientists, interested in data mining and Faloutsos of Carnegie Mellon statisticians, mathematicians, and knowledge discovery, has been University and Edward Chang of data analysis practitioners to pro- that the data are an accounting, Google Research, in which they mote cross-fertilization of ideas. or record, of everything that presented an overview of tools MMDS 2008 occurred in the happened in a particular setting. and applications in modern large- wake of MMDS 2006, which From this perspective, the goal is scale data analysis.
16 Amstat News JUNE 2009 Faloutsos began his tuto- or Hadoop that supports data- large-scale social and information rial, “Graph Mining: Laws, intensive distributed computa- networks available, however, Generators, and Tools,” by tions running on large clusters generative models that are struc- motivating the problem of data of hundreds, thousands, or even turally or syntactically more flex- analysis on graphs. He described hundreds of thousands of com- ible are increasingly necessary. a wide range of applications in modity computers. By introducing a small exten- which graphs arise naturally, and sion in the parameters of a gen- he reminded the audience that Algorithmic Approaches erative model, of course, one can large graphs that arise in modern to Networked Data observe a large increase in the informatics applications have Milena Mihail of the Georgia observed properties of generated structural properties that are very Institute of Technology described graphs. This observation raises different from traditional Erdös- algorithmic perspectives on devel- interesting statistical questions Rényi random graphs. Although oping better models for data in her about model overfitting, and these structural properties have tutorial, “Models and Algorithms it argues for more refined and been studied extensively in for Complex Networks.” She systematic methods of model recent years and used to devel- noted that a rich theory of power parameterization. This observa- op numerous well-publicized law random graphs has been tion also leads to new algorith- models, Faloutsos also described developed in recent years. With mic questions, which were the empirically observed properties the increasingly wide range of topic of Mihail’s talk. that are not well reproduced by existing models. Building on this, Faloutsos spent much of his talk describing several graph- mining applications of recent and ongoing interest. Edward Chang described other developments in web- scale data analysis in his tuto- rial, “Mining Large-Scale Social Networks: Challenges and Scalable Solutions.” After review- ing emerging applications—such as social network analysis and per- sonalized information retrieval— Chang covered several other appli- cations in detail. In all these cases, he emphasized that the main per- formance requirements were “scal- ability, scalability, scalability.” Modern informatics applica- tions such as web search afford easy parallelization (e.g., the overall index can be partitioned such that even a single query can use multiple processors). Moreover, the peak performance of a machine is less important than the price-performance ratio. In this environment, scalability up to petabyte-sized data often means working in a software framework such as MapReduce
JUNE 2009 Amstat News 17 problem. A common choice for the penalty is the ℓp-norm of the coefficient vector a = (a1, a2, … , an). This interpolates between the subset selection problem (g = 0) and ridge regression (g = 2) and includes the well-studied lasso (g = 1). For g # 1, sparse solu- tions are obtained, and for g $ 1, the penalty is convex. Although one could choose an optimal (l, g) by cross vali- dation, this can be prohibitively expensive. In this case, so-called path-seeking methods that can be used to generate the full path of optimal solutions {â(l) : 0 # l # } in time that is not much The Geometric their application to natural image more than that needed to fit a Perspective: Qualitative statistics and data visualization. single model have been studied. Analysis of Data Statistical and Machine Friedman described a generalized path-seeking algorithm, which A very different perspective was Learning Perspectives provided by Gunnar Carlsson of efficiently solves this problem for Stanford University, who gave an Statistical and machine learning a much wider range of loss and overview of geometric and topo- perspectives on MMDS were penalty functions. logical approaches to data analy- the subject of a pair of tutorials Jordan, in his tutorial, “Kernel- sis in his tutorial, “Topology and by Jerome Friedman of Stanford Based Contrast Functions for Data.” The motivation underly- University and Michael Jordan Sufficient Dimension Reduction,” ing these approaches is to provide of the University of California at considered the dimensionality insight into the data by impos- Berkeley. Given a set of measured reduction problem in a super- ing a geometry on it. Part of the values of attributes of an object, vised learning setting. Methods problem is thus to define useful x = (x1, x2, … , xn), the basic such as principal components metrics—especially as applica- predictive or machine learning analysis, Johnson-Lindenstrauss tions such as clustering, clas- problem is to predict or estimate techniques, and Laplacian-based sification, and regression often the unknown value of another nonlinear methods are often depend sensitively on the choice attribute y. used, but their applicability is of metric—and two design goals In his tutorial, “Fast Sparse limited since the axes of maxi- have recently emerged. First, don’t Regression and Classification,” mal discrimination between two trust large distances. As distances Friedman began with the com- classes may not align well with are often constructed from a sim- mon assumption of a linear model the axes of maximum variance. ilarity measure, small distances in which the prediction One might hope there exists a reliably represent similarity, but . Unless the low-dimensional subspace of large distances make little sense. number of observations is much the input space X, which can be Second, trust small distances larger than n, however, empiri- found efficiently and retains the only a bit. After all, similarity cal estimates of the loss function statistical relationship between X measurements are still very noisy. exhibit high variance. To make and the response space Y. These ideas suggest the design the estimates more regular, one Jordan showed that this prob- of analysis tools that are robust typically considers a constrained lem of sufficient dimensionality to stretching and shrinking of or penalized optimization prob- reduction (SDR) could be for- the underlying metric. Much of lem. The choice of an appropriate mulated in terms of conditional Carlsson’s tutorial was occupied by value for the regularization param- independence and evaluated in describing these analysis tools and eter l is a classic model selection terms of operators on reproducing
18 Amstat News JUNE 2009 kernel Hilbert spaces (RKHSs). methods, dimensionality reduc- ple, nearly every statistician com- Interestingly, the use of RKHS tion and graph partitioning mented on the desire for more ideas to solve this SDR problem methods, and co-clustering and statisticians at the next MMDS; cannot be viewed as a kerneliza- other matrix factorization meth- nearly every scientific computing tion of an underlying linear algo- ods, participants heard about a researcher told us they wanted rithm, as is typically the case when variety of data applications. more data-intensive scientific such ideas are used (e.g., with In all cases, scalability was a computation at the next MMDS; SVMs) to provide basis expan- central issue, motivating discus- nearly every practitioner from an sions for regression and classifica- sion of external memory algo- application domain wanted more tion. Instead, this is an example rithms, novel computational applications at the next MMDS; of how RKHS ideas provide algo- paradigms such as MapReduce, and nearly every theoretical com- rithmically efficient machinery to and communication-efficient lin- puter scientist said they wanted optimize a much wider range of ear algebra algorithms. Interested more of the same. There is a lot of statistical functionals of interest. readers are invited to visit http:// interest in MMDS as a develop- mmds.stanford.edu, where pre- ing interdisciplinary research area Conclusions and sentations from all speakers can at the interface between com- Future Directions be found. puter science, statistics, applied In addition to other talks on The feedback received made it mathematics, and scientific and the theory of data algorithms, clear that MMDS struck a strong Internet data applications. Keep machine learning and kernel interdisciplinary chord. For exam- an eye out for future MMDSs. n
LATEX
LATEX LATEX
MacKichSOFTWARE,an INC.
JUNE 2009 Amstat News 19 Feds in Space JSM session highlights government survey research in cyberspace and outer space Janice Lent, Energy Information Administration
or statisticians seeking a career path, salary is and testing not the main issue. Had money been our top of the system goal (had we been … well … ‘pigs’), we would present huge chal- Fhave sought careers in commerce, law, or invest- lenges. But O’Neill ment banking. Let’s face it: We just don’t like the knows her work will aid daily grind of repetitive, routine work. We want the nation’s understanding of challenging abstract problems, opportunities to its energy situation for decades to come. stretch our minds, and a chance to leave our mark Darcy Anne Miller of the National Agricultural on the world. Statistics Service (NASS) will discuss the “Census by Hot Research Projects: A Taste of the Future Satellite” methods NASS uses to collect data about for Future Government Statisticians, a topic- American farms. NASS employs satellite imagery to contributed session at JSM 2009, will feature four create a cropland data layer for use in geographi- young statisticians who discovered the research cal information systems. Miller is also involved in projects they could tackle in federal statistical agen- exploring new uses of data-mining techniques to cies offered great outlets for their creative ener- refine remotely collected data and improve data gies. Organized by members of the Innovation quality. The innovations she will discuss aid in data Workgroup, which functions under the auspices of dissemination, allowing complex agricultural sta- the Interagency Council for Statistical Policy, the tistics to be displayed in easily understood formats. session will focus on hot research projects under way All in all, the new methods NASS is developing are in the federal statistical system. quickly putting the old, busy-looking data tables in These talks will show that data collection—and the “horse-and-plow” category. even analysis—is not just for humans anymore. Yes, Have you ever wondered what the IRS does with the U.S. Census Bureau has had to hire squadrons all that data you have to fill in on your tax forms? of enumerators to canvass the tough spots on the Believe it or not, you’re not wasting your time. decennial map, but many federal surveys can now Though your individual information is kept con- be administered online by computer—or even from fidential, some of it is aggregated with other data space by satellite. and used for statistical purposes. Ronald Walsh of Energy data is a case and point. As America’s the IRS Statistics of Income (SOI) Division will talk energy picture evolves, the Energy Information about how statisticians at the IRS are putting that Administration (EIA) is tooling up to collect more vast reservoir of tax data to good use for researchers data on the new energy sources and alternative fuels and the American people. Walsh will discuss several slowly gaining market share. To support the effort, innovative projects that the SOI Statistical Support Grace O’Neill, who recently joined EIA, is involved Section is currently spearheading. in developing a standardized Internet-based data Of course, the U.S. Census Bureau is the agen- collection system for the agency. cy that originally put government statistics “on Working with interdisciplinary teams—includ- the map,” so to speak. You may be interested to ing energy technology experts, IT specialists, and know that young statisticians there are still at economists—O’Neill finds herself consistently it—developing high-tech tools for integrating learning something new. The final system will sup- and visualizing geographically detailed data. port all phases of data collection, from questionnaire Matthew Graham will discuss and demonstrate design to data quality checks. Given the diversity OnTheMap, a dynamic new online mapping and data of users and types of data collected, development reporting software.
20 Amstat News JUNE 2009 OnTheMap takes inspiration from C. J. new technologies to gain fresh insight into America’s Minard’s impressive diagram of Napoleon’s march challenges from the data we collect. on Moscow. With interactive data manipulation Nancy Gordon, the associate director for stra- and graphics tools, however, OnTheMap reaches tegic planning and innovation at the U.S. Census beyond the 19th-century masterpiece, integrating Bureau, will serve as discussant. She will round out administrative records, decennial census data, and the session with an overview of the exciting research sample survey data to generate user-requested dis- opportunities available for statisticians in the federal plays while preserving respondents’ confidential- statistical system. Gordon will be happy to discuss ity. Graham will show us how the U.S. Census questions and issues raised from the floor. So, what Bureau is marrying creative minds with powerful is it like to be a government statistician today? Come and see—you may be surprised! n
Are You Seeking a Statistician?
Reserve an Executive Suite at the 2009 Joint Statistical Meetings and let us help you! Each Executive Suite employer receives: A PRIVATE, FURNISHED BOOTH for your exclusive use throughout the meeting. Use it as an office, interview booth, or meeting space. It will give you the freedom to schedule interviews with potential employees whenever you like, without the need to reserve space on an hourly basis.
REGISTRATION FOR THE JSM PLACEMENT SERVICE, which includes up to 10 Contact Amy Farris at position listings. Also, up to five representatives and three guests from your organiza- [email protected] for tion will have access for conducting onsite interviews. more information. ONLINE ACCESS TO APPLICANT DATA AND RÉSUMÉS IN ADVANCE. Beginning June 1 and continuing through September 30, search our online service Registration will for qualified candidates based on specific criteria. Print résumés and pursue contacts remain open before JSM starts, and continue to access this resource after the meeting for those through July 16. last-minute openings.
AN ONSITE COMPUTER AND PRINTER IN YOUR SUITE. Eliminate waiting in line at the message center. Research and communicate with new applicants as they register.
SPECIAL RECOGNITION ONLINE AND ONSITE. Executive Suites are designated in the online listing of employers and noticed by candidates. Leave your prospective employees with a lasting impression of your company’s commitment to hiring talented statisticians. You also will receive JSM Partners Program credit for your purchase! , PERSONALIZED SERVICE from JSM Career Placement staff members, who will answer your questions and help with your interview traffic.
To reserve your Executive Suite, register online as soon as possible. Executive Suites are limited and have sold out in each of the last four years! www.amstat.org/meetings/jsm/2009/placement
JUNE 2009 Amstat News 21 22 Amstat News JUNE 2009 CHANCE Highlights Baseball, Graphics Play Major Role in Newest Issue Mike Larson, CHANCE Editor
he two themes of CHANCE 22(2) are base- role can statisticians/biostatisti- ball and graphics. Mark Glickman’s Here’s cians play in design, modeling, to Your Health column, Jonathan Berkowitz’s and validation? TGoodness of Wit Test puzzle column, and a couple In his Goodness of Wit Test of letters to the editor complete the issue. column, Berkowitz gives us a For the baseball lover, there are three articles. For bar-type cryptic puzzle with one statisticians in general, issues of defining measures of additional solving requirement. performance, selecting a comparison group, adjust- Last, but not least, two read- ing for confounders, and modeling also appear in ers submitted letters. Susan these articles. Jim Albert compares the career trajec- Aref comments on Stephanie tories of pitchers. The rate at which pitchers allow Land’s article (21[4]), “‘We All hits and walks, adjusted for the pool of pitches and Survived’ and Other Failings of number of batters, is modeled using piecewise qua- Risk Perception,” and confirms dratic functions. Flexible, multilevel modeling is that it is, indeed, safer to be born used to create individual trajectories that adjust for now than in the past. Ray Stefani important factors and allow nuanced comparisons discusses efforts to reproduce across pitchers. results from tables discussed in Did drugs play a role in Roger Clemens’ later articles, also in 21(4), by Brian Clauser and Stephen career performance? Brian Schmotzer, Pat Kilgo, Stigler. The details presented here could be of use and Jeff Switchenko estimate the effect of perfor- to instructors and others working to understand and mance-enhancing drugs on offensive production in explain the tables. baseball. The outcome measure is runs created per In other news, CHANCE, working through the 27 outs. Information about drug use comes from ASA, conducted a survey of lapsed subscribers. the “Mitchell Report.” To address sensitivity to Approval for the survey was obtained from the ASA’s the statistical model, several models are estimated Survey Review Committee. An email message with a and contrasted. link to a web survey was sent to individuals who had Finally, Don Chance takes a look at Joe let their subscription to CHANCE expire in the last DiMaggio’s 1941 56-game hitting streak. Adjusting few years. The purpose of the survey was to learn for hit opportunities and making other assumptions, something about why people stopped subscribing does DiMaggio have the highest probability of ever and what they liked or disliked about CHANCE. having such a streak? Forty-four former subscribers responded. The three winners of the Will Burtin graphics According to them, the technical level is about right contest (announced in 21(4) are Mark Nicolich, and the amount and quality of graphics is accept- Dibyojyoti Haldar, and Brian Schmotzer. Their able, although there appears to be some room for graphs and biographies begin on Page 43. Howard improvement. Topics with the high response were Wainer’s Visual Revelations column presents sev- current events, education, environment, and health/ eral additional graphs and examines the lessons medical. Other topics of strong interest were clinical learned through the contest. We wish to thank the trials, economics, government, graphics, legal issues, 64 entrants to the contest for their efforts. Based and surveys. One respondent added the following on the positive response to the contest, we plan to comment: “I want to see interesting applications. conduct a similar one in the near future. The area doesn’t matter.” In Glickman’s Here’s to Your Health column, In summary, we interpret the results of the sur- Xiaofei Wang, Herbert Pang, and Todd Schwartz vey to support the mission of CHANCE: provide study cancer biomarkers: What are they? How are accessible and interesting articles on diverse topics in they used for prediction and classification? What which probability and statistics play critical roles. n
JUNE 2009 Amstat News 23 24 Amstat News JUNE 2009 ASA Online Journals Provide the Tools You Need
ou know the ASA’s journals represent the best PDF Plus is better than a web page in one respect. in statistical science, covering a broad range You can download the file and take it with you of application areas while advancing the lat- anywhere. All the links work. And you get links to Yest in theory. What you may not know is that the external web pages and internal links to figures, ASA’s online presence is now more robust, more tables, equations, and references. flexible, and more powerful than ever. The ASA now gives you the tools you need to access the lit- Short on Time? erature you want and need. And you can put this Download the abstracts you want and read access to use right away. them later. ASA members are already able to access the Journal of the American Statistical Association, the Journal Found an article you need to cite? Download the of Economic & Business Statistics, The American BiBTeX entry and put it in your BiBTeX database. Statistician, and Statistics in Biopharmaceutical It beats retyping every time. Research for no extra cost. Simply log in to the Members Only area of www.amstat.org to give these Got Supplements? journals a try. Here’s what you can expect: R code, data sets, extended appendices … What do you need? The ASA now posts article supplements Search Anyone? right next to the article in the table of contents. If Simple and advanced full-text searches of the ASA’s you are a subscriber, you can access all the good journals. Search across all journals, within a given stuff and put it to work for you with a simple click. journal, or even within a single issue. No more being rerouted to other web sites!
Want to keep on top of a research topic? Sign up Really, there isn’t room to list all the great features for an email alert when a paper matching your the ASA online journals provide. You need to see search term is posted. them for yourself by logging in to Members Only at www.amstat.org/membersonly. n PDF Plus When is a PDF like a web page? When it is a PDF Plus. Take a look at the references in an ASA journal PDF Plus article. If there is a blue arrow Don’t forget to let us know how we’re doing. Drop a line to next to the citation, you’ll find a number of ways [email protected] if you have a comment about the ASA’s to access that article, or at least its information. online journals. Better yet, if it is an article that appeared in an ASA journal since 2001, the link will take you right to the abstract.
JUNE 2009 Amstat News 25 JASA Highlights June Issue Filled with Diversity of Papers Len Stefanski, David Banks, and Dalene Stangl
he Journal of the American Statistical Association editors JASA Table of Contents Tare delighted to introduce Applications and Case Studies their new colleague, Hal Joint Models for the Association of Longitudinal Binary and Stern, from the Department Continuous Processes with Application to a Smoking Cessation Trial of Statistics at the Xuefeng Liu, Michael J. Daniels, and Bess Marcus University of California at Irvine. Stern is the incom- Bayesian Analysis of Cancer Rates from SEER Program Using Parametric and Semiparametric Joinpoint Regression Models ing coordinating editor Pulak Ghosh, Sanjib Basu, and Ram C. Tiwari and editor of the Applications and Case Nonparametric Priors for Ordinal Bayesian Social Science Models: Studies section. He has Specification and Estimation Jeff Gill and George Casella already begun to handle the new A&CS sub- missions. As is customary, David Banks will con- A Statistical Framework to Infer Functional Gene Relationships tinue to handle papers that are already in process from Biologically Interrelated Microarray Experiments through the end of 2009. Siew Leng Teng and Haiyan Huang The editors want to particularly thank Mark Assessing Sexual Attitudes and Behaviors of Young Women: Kaiser for chairing the search committee that select- A Joint Model with Nonlinear Time Effects, Time Varying ed Stern. There were many excellent candidates. The Covariates, and Dropouts Pulak Ghosh and Wanzhu Tu profession is fortunate to have such depth of strength among its members, and even more fortunate that so Inferring Optimal Peer Assignment from Experimental Data many are willing to volunteer for this job. Debopam Bhattacharya Looking to the future, the ASA is undertaking a Sensitivity Analysis for Equivalence and Difference in an review of its approach to electronic publication. Observational Study of Neonatal Intensive Care Units David Scott is heading up a committee charged Paul R. Rosenbaum and Jeffrey H. Silber with reporting to the ASA Board in June and Random Effects Models in a Meta-Analysis of the Accuracy of Two August on the financial and technical feasibility of Diagnostic Tests Without a Gold Standard significantly expanding the scope of journal ser- Haitao Chu, Sining Chen, and Thomas A. Louis vices members receive. Repeated Measurements on Distinct Scales with Censoring— A Bayesian Approach Applied to Microarray Analysis of Maize Applications and Case Studies Tanzy Love and Alicia Carriquiry The section contains its usual diversity of papers. Nonparametric Signal Extraction and Measurement Error in the There are two by Pulak Ghosh, one on Bayesian Analysis of Electroencephalographic Activity During Sleep analysis of cancer data and one on the assessment Ciprian M. Crainiceanu, Brian S. Caffo, Chong-Zhi Di, and of sexual attitudes and behaviors. Genetic analysis Naresh M. Punjabi is represented in three papers—by Val Johnson, by Nonparametric Residue Analysis of Dynamic PET Data with Tanzy Love and Alicia Carriquiry, and by Gilles Application to Cerebral FDG Studies in Normals Durrieu and Laurent Briollais. Finbarr O’Sullivan, Mark Muzi, Alexander M. Spence, David M. Chris Paciorek and Jason McLachlan use Mankoff, Janet N. O’Sullivan, Niall Fitzgerald, George C. Newman, data on pollen sediments to map ancient for- and Kenneth A. Krohn ests. Hongtu Zhu and a research team of eight Modeling Hazard Rates as Functional Data for the Analysis of coauthors develop a new noise model for magnetic Cohort Lifetables and Mortality Forecasting resonance imaging. Kristin Lennox, David Dahl, Jeng-Min Chiou and Hans-Georg Müller Marina Vannucci, and Jerry Tsai apply spheri- Density Estimation for Protein Conformation Angles Using a cal data methods to analyze protein conformation Bivariate von Mises Distribution and Bayesian Nonparametrics angles. Greg Ridgeway and John MacDonald study Kristin P. Lennox, David B. Dahl, Marina Vannucci, and Jerry W. Tsai racial bias in traffic stops. Debopam Bhattacharya
26 Amstat News JUNE 2009 develops a procedure for optimally assigning college Log-Linear Models for Gene Association roommates, exploiting estimated interaction effects Jianhua Hu, Adarsh Joshi, and Valen E. Johnson to boost academic performance. And Catriona Mapping Ancient Forests: Bayesian Inference for Spatio-Temporal Queen and Casper Albers study intervention and Trends in Forest Composition Using the Fossil Pollen Proxy Record causality in traffic flows. Christopher J. Paciorek and Jason S. McLachlan This does not begin to exhaust the range of appli- Regression Models for Identifying Noise Sources in Magnetic cations in this issue, nor does this short catalog do Resonance Images justice to the authors. It is an exciting collection of Hongtu Zhu, Yimei Li, Joseph G. Ibrahim, Xiaoyan Shi, papers, and showcases the broad impact of our field. Hongyu An, Yashen Chen, Wei Gao, Weili Lin, Daniel B. Rowe, and Bradley S. Peterson Theory and Methods Intrinsically Autoregressive Spatio-Temporal Models with The Theory and Methods Section leads with the Application to Aggregated Birth Outcomes discussion paper “On Consistency and Sparsity Jonathan D. Norton and Xu-Feng Niu for Principal Components Analysis in High Sequential Design for Microarray Experiments Dimensions,” by Iain M. Johnstone and Arthur Gilles Durrieu and Laurent Briollais Yu Lu. The authors argue that for principal compo- Doubly Robust Internal Benchmarking and False Discovery Rates nents analysis with large-p-small-n data “some ini- for Detecting Racial Bias in Police Stops tial reduction in dimensionality is desirable before Greg Ridgeway and John M. MacDonald applying any PCA-type search for principal modes, Intervention and Causality: Forecasting Traffic Flows Using a [and that] the initial reduction in dimensionality is Dynamic Bayesian Network best achieved by working in a basis in which the sig- Catriona M. Queen and Casper J. Albers nals have a sparse representation.” Discussion papers are by B. Nadler; D. Whitten, T. Hastie, and R. Theory and Methods Tibshirani; and J. Ramsay. On Consistency and Sparsity for Principal Components Analysis in Functional data are addressed in “Estimating High Dimensions Derivatives for Samples of Sparsely Observed Iain M. Johnstone and Arthur Yu Lu Functions, with Application to Online Auction Estimating Derivatives for Samples of Sparsely Observed Functions, Dynamics” by Bitao Liu and Hans-Georg Müller with Application to Online Auction Dynamics and “On the Concept of Depth for Functional Bitao Liu and Hans-Georg Müller Data” by Sara López-Pintado and Juan Romo, whereas shrinkage in high-dimensional problems On the Concept of Depth for Functional Data Sara López-Pintado and Juan Romo is the topic of “Partial Correlation Estimation by Joint Sparse Regression Models” by Jie Partial Correlation Estimation by Joint Sparse Regression Models Peng, Pei Wang, Nengfeng Zhou, and Ji Zhu Jie Peng, Pei Wang, Nengfeng Zhou, and Ji Zhu and “Shrinkage Estimation of the Varying Shrinkage Estimation of the Varying Coefficient Model Coefficient Model” by Hansheng Wang and Hansheng Wang and Yingcun Xia Yingcun Xia. Bayesian Mixture Labeling by Highest Posterior Density Bayesian contributions include “Bayesian Weixin Yao and Bruce G. Lindsay Mixture Labeling by Highest Posterior Density” Prior Distributions from Pseudo-Likelihoods in the Presence of by Weixin Yao and Bruce G. Lindsay and “Prior Nuisance Parameters Distributions from Pseudo-Likelihoods in the Laura Ventura, Stefano Cabras, and Walter Racugno Presence of Nuisance Parameters” by Laura Ventura, Confidence Intervals for Population Ranks in the Presence of Ties Stefano Cabras, and Walter Racugno. and Near Ties Theory and Methods is rounded out with inter- Minge Xie, Kesar Singh, and Cun-Hui Zhang esting contributions on confidence intervals in “Confidence Intervals for Population Ranks in the Computationally Efficient Nonparametric Importance Sampling Jan C. Neddermeyer Presence of Ties and Near Ties” by Minge Xie, Kesar Singh, and Cun-Hui Zhang; importance sampling A Class of Transformed Mean Residual Life Models with Censored in “Computationally Efficient Nonparametric Survival Data Jan C. Neddermeyer Liuquan Sun and Zhigang Zhang Importance Sampling” by ; survival analysis in “A Class of Transformed Mean A Multivariate Extension of the Dynamic Logit Model for Residual Life Models with Censored Survival Data” Longitudinal Data Based on a Latent Markov Heterogeneity by Liuquan Sun and Zhigang Zhang; binary lon- Structure Francesco Bartolucci and Alessio Farcomeni gitudinal data in “A Multivariate Extension of the Dynamic Logit Model for Longitudinal Data Based Hunting for Significance with the False Discovery on a Latent Markov Heterogeneity Structure” by Martin Posch, Sonja Zehetmayer, and Peter Bauer Francesco Bartolucci and Alessio Farcomeni; and
JUNE 2009 Amstat News 27 multiple testing in “Hunting for Significance with are intelligent opponents and uncertain outcomes. the False Discovery Rate” by Martin Posch, Sonja The paper describes several formulations of adver- Zehetmayer, and Peter Bauer. sarial risk problems and provides a framework that extends traditional risk analysis tools. Research Reviews challenges are outlined, examples from corporate In the Reviews section, David Rios Insua, Jesus competition illustrate ideas, and the topic’s rel- Rios, and David Banks present a paper titled evance to counterterrorism and national defense “Adversarial Risk Analysis.” The authors review meth- is highlighted. n ods for the analysis of decisionmaking when there
Norwood Award Proposals Due June 29
he University of Alabama their terminal degree, have made tical science. Self-nominations at Birmingham’s Section outstanding contributions to the are acceptable. on Statistical Genetics statistical sciences, and, if select- Send nominations to David B. Tand Department of Biostatistics ed, are willing to deliver a lec- Allison, professor and head of the in the School of Public Health ture at the award ceremony. For Section on Statistical Genetics, request nominations for details, visit www.soph.uab.edu/ssg/ Department of Biostatistics, the eighth annual Janet L. norwoodaward/aboutaward. RPHB 327, University of Norwood Award for Outstanding To nominate a candidate, send Alabama at Birmingham, 1665 Achievement by a Woman in the a full curriculum vitae accom- University Blvd., Birmingham, Statistical Sciences. The award panied by a letter of not more AL 35294-0022. To send elec- will be conferred on Wednesday, than two pages describing the tronic submissions, email Allison September 16, 2009. The recipi- nature of the candidate’s con- at [email protected]. ent will be invited to deliver a lec- tributions. Contributions may The deadline for receipt ture at the UAB award ceremony, be in the development and of nominations is Monday, all expenses paid. Additionally, evaluation of statistical meth- June 29, 2009. The winner will the award winner will receive a ods, teaching of statistics, be announced by Friday, July plaque and $5,000 prize. application of statistics, or any 4, 2009. n Those who are eligible are other activity that can arguably women who have completed be said to have advanced statis-
28 Amstat News JUNE 2009 NC State Dedicates Math, Stats Building $32M structure named SAS Hall after software company
orth Carolina State University formally Statistics of Department NCSU the of courtesy Hall, Marc By dedicated its new mathematics and statis- tics building on Friday, May 1. At a dedi- Ncation celebration attended by more than 200 fac- ulty, students, friends, and alumni, NC State Chancellor James L. Oblinger announced for the first time that the building will be named SAS Hall, in honor of the founders of the software company based in Cary, North Carolina. The 119,000-square-foot building will house state-of-the-art classrooms, computer labs, tutorial centers, and meeting and study space for students and faculty from NC State’s mathematics and sta- tistics departments. A partnership with Cisco will improve digital communications for students by providing access to live and on-demand video con- tent from anywhere on campus. From left: D. McQueen Campbell, chair of the board of trustees; Dan Construction of the $32 million building was Solomon, dean of the College of Physical and Mathematical Sciences; Loek made possible by the Higher Education Bond Helminck, head of the Department of Mathematics; John Sall, cofounder Referendum passed by North Carolinians in 2000, of SAS; Jim Goodnight, cofounder of SAS; Sastry Pantula, head of the as well as by gifts from private donors, including a Department of Statistics; and Jim Oblinger, chancellor substantial contribution from SAS. SAS was born out of a research project that began in the NC State Department of Statistics in the early Statistics of Department NCSU the of courtesy Hall, Marc By 1970s. Since then, the company has grown into one of the largest software providers in the world. Two of the company’s founders, CEO Jim Goodnight and Executive Vice President John Sall, as well as their spouses, remain close partners and staunch supporters of the department and university. “At SAS, we believe it is vital for students in the mathematical and statistical sciences to learn in an environment that provides state-of-the-art facili- ties and instructional technologies,” Sall said. “It’s also critical that they participate in the kind of col- laborative initiatives they’ll experience in the work place. That type of environment produces the type of employee and person we want at SAS, and it’s the type we want to produce at NC State. That’s why we decided to make a significant contribution SAS Hall, named in honor of the founders of the Cary, North Carolina, soft- toward ensuring that this building would become ware company a reality.”
JUNE 2009 Amstat News 29 “The technology resources in this building will indeed have a very positive impact on teach- ing, learning, consulting, and [doing] research in our departments,” said Sastry Pantula, head of the Department of Statistics at NC State. “I must say, so will the physical layout of the spaces within this beautiful building. A lot of thought has gone into the overall sense of community and collabo- ration we wanted these spaces to convey.” Pantula continued, “We are very grateful to the donors for their support and for their vision, to the taxpayers of North Carolina, to Dean Solomon, and to every single person involved with this masterpiece, from the conception to completion.” NC State boasts a long-standing tradition of By Marc Hall, courtesy of the NCSU Department of Statistics excellence in teaching and research in statistics. Its Department of Statistics is among the nation’s old- est and most prestigious, having been founded by renowned statistician Gertrude Cox in 1941. The university currently ranks fifth nationally in total R&D expenditures and in competitive federal R&D expenditures in the mathematical and statistical sci- ences. It also received the Departmental Teaching and Learning Excellence Award at NC State a few years ago. “We now have a state-of-the-art facility that is worthy of the stature of our students and Jackie Hughes-Oliver, codirector of graduate programs, faculty,” said Solomon. n and Anthony Franklin, a PhD student
Library Slipped? t is no secret that libraries are under tremendous budget pressure right now. Who isn’t? But when the library where you work lets the subscription lapse on your favorite ASA journal, well that’s a Istep too far. And if your library doesn’t have a subscription, let them know you would like to have those vital journals available as soon as possible. You can now recommend ASA titles right from http://pubs.amstat.org. Click on any title and you’ll find a link to rec- ommend the title to your librarian at the top. Your ASA journals are a vital link for you. But they are also valuable resources to students and practitioners of many disciplines. Has your librarian heard from you? Maybe it’s time to let them know about sta- tistics, the ASA, and the journals you love. n
30 Amstat News JUNE 2009 Freeware for Statistical Computing Staff Spotlight Elizabeth Shwaery
FPL Statistics Unit Java and HTML forms/Perl information for statisticians www1.fpl.fs.fed.us/poster897.html
TeX Users Group www.tug.org
The R Project for Statistical Computing www.r-project.org
FreeMacWare Scientific software applications for Mac computer users While planning an event in Sydney, Australia, Elizabeth meets a koala bear. www.freemacware.com/category/ scientific-mac-freeware
s the newest member of I planned there took me all RDV the American Statistical over the world, allowing me to Free data viewer software Association team, I would spend several weeks in Australia, http://code.google.com/p/rdv/?gclid=CPb_ Alike to take this opportunity to Europe, and Asia. I look forward jLH06JgCFQFvGgod8xFm1g introduce myself. My name is to bringing my event planning Elizabeth Shwaery, and I joined skills to the ASA and, specifically, MD Anderson Cancer Center the ASA as a meetings planner to the Joint Statistical Meetings Department of Biostatistics & Applied about a month ago. this August. Mathematics Software Download Site In 2005, I graduated from On a personal note, I was http://biostatistics.mdanderson.org/ James Madison University with born and raised in Springfield, SoftwareDownload a degree in public relations and Virginia. I am recently engaged a minor in political communi- and planning my wedding in The web sites listed here are provided as a courtesy. The ASA does cation. After college, I moved to Washington, DC, in October. not control, monitor, or guarantee the information contained in any other entity web sites and shall not be liable for any damages of any Orlando, Florida, where I worked Being a professional event plan- kind. If you have a web site statisticians would find useful, and you for a retail property management ner is a definite asset when plan- would like to see it listed, please send it to [email protected]. company planning store grand ning a wedding! openings and shopping events. I am looking forward to meet- I decided to move back to the ing many of you this summer at Washington, DC, area after about JSM. Should you have any JSM- two years, and, in 2007, I joined related questions, please feel free the Corporate Executive Board as to contact me. I can be reached at a meetings manager. The events [email protected]. n
JUNE 2009 Amstat News 31 32 Amstat News JUNE 2009 ASA-SIAM Series Publishing with the Series Has Benefits
s our authors know, pub- while a proofreader also checks lishing with the ASA- the proofs. SIAM Series on Statistics Aand Applied Probability has many Worldwide marketing benefits. Highlighted below are The ASA and SIAM have a com- a few. bined membership of approxi- mately 30,000 individuals and a Personal service marketing database of more than When you publish with the 100,000 to whom series books ASA-SIAM series, you’re not just are marketed. Also, through a another author. You’re an ASA- partnership between SIAM and SIAM author. That means you Cambridge University Press will receive individual attention (CUP), our books are exhibited from our staff and not get lost in and distributed outside North the shuffle. Acquisitions Editor America by CUP staff. Sara Murphy will be your main contact, and other in-house staff Three-tiered pricing will work through Murphy or All books in the series have three directly with you to meet or exceed prices: a list price, a price for ASA your expectations. We are happy and SIAM members (30% dis- to accommodate authors’ special count), and a price for conference requests whenever possible. attendees who are not members of either society (20% discount). Competitive royalties Shipping fees are waived for onsite and complimentary or orders at conferences. discounted copies The series offers royalties that are Books that never go out competitive with commercial pub- of print lishers and escalate after a predeter- One of the key features of the mined number of copies have sold. book series is that the books do Upon publication, authors receive not go out of print unless they are a number of complimentary copies replaced by new editions. The time and a 40% discount on additional you invest in writing for the ASA- copies intended for personal use. SIAM series is rewarded with an indefinite shelf life for your book. Meticulous copy editing While some publishers print their Service to the profession books from camera-ready copy, Publishing with the ASA-SIAM each of our books undergoes a series not only gets your book thorough copy edit by an in-house into the hands of readers, but or United States–based freelance also serves the profession by per- editor to correct grammatical, petuating the work done by these stylistic, and punctuation errors. partner societies. The copy editor also addresses If you are interested in consid- the book’s layout to ensure an ering the ASA-SIAM series as the easy-to-read finished product. outlet for your next book, contact The copy editor’s work is then Sara J. Murphy at murphy@siam. quality checked by a production org. We look forward to working editor before proofs are created. with you. n Authors receive proofs to check
JUNE 2009 Amstat News 33 34 Amstat News JUNE 2009 Funding Opportunities Increased Funding for Science: What Does It Mean for Statistics? Keith Crank, ASA Assistant Director for Research and Graduate Education
n an address to the National Academy of Sciences, President Barack Obama reiterated his plans to double the budgets of the National IScience Foundation (NSF), the Department of Energy’s (DOE) Office of Science, and the National Institute of Standards and Technology (NIST). Originally, this doubling was to have occurred over a 10-year period, but with flat funding in fiscal years 2007 and 2008, it will need to double in eight years to meet that goal. (This may not happen. It is not clear whether the current goal is to double these budgets by 2016 or 2018.) How will this affect statistics? As academic fund- ing of statisticians by NIST and DOE’s Office of Science is small or nonexistent, we will focus on NSF. We start with what might be considered a typ- ical grant. Consider an award to a principal investi- gator (PI) with a graduate student (GS). The award period, but if this is a person near the center of the might look something like the following: award distribution now, he or she should be mov- ing into the upper tail of the distribution over the next 8–10 years. Instead, we will consider someone Salary ($90,000 for $20,000 at this stage of his or her career 8–10 years from now academic year) (i.e., someone who is probably currently in graduate Fringe Benefits (30%) $ 6,000 school). What are some reasonable assumptions for the changes in the award? GS Stipend $20,000 For a new associate professor (0–1 years in rank), GS Tuition $15,000 the salary at the 75th percentile increased by an aver- Travel $ 2,500 age of about 4% per year between 2003 and 2007. For the 90th percentile, the increase was about 5% Supplies $ 500 per year, and for the 95th percentile, the increase Indirect Costs (50%) $32,000 was about 6% per year. (As we are considering a Total One-Year Award $96,000 typical NSF awardee, the salary is likely to be in the upper tail of the distribution.) For my calculations, I used a 5% increase in salary per year. For the other items, it is more difficult to find data. I assumed the If the Statistics Program currently has $12 mil- fringe benefit rate and indirect cost rate would stay lion to spend, it could make 125 grants like this one. the same from year to year. I put in a 4% increase in (I don’t know the current budget for the Statistics GS stipend, an 8% increase in tuition, a 3% increase Program. The figure used here is for illustration.) in travel, and a 2% increase in supplies. Now, what happens over an eight- or 10-year This results in an award size of $146,000 after period? We could follow this person and see what eight years and $164,500 after 10 years. If the pro- his or her grant might look like at the end of the gram budget doubles to $24 million, this would
JUNE 2009 Amstat News 35 Program’s budget. This may not be the case, espe- cially if there is no increase in demand (in the form of more proposals to the program). So, while I have tried to caution against exuberance, I do not want to discourage statisticians from applying for funding. It is also appropriate to compare the results of doubling the budget with other alternatives. If the budget were to increase by 3% per year (i.e., keep- ing up with inflation), the number of awards would decrease by about one-sixth after eight years and by increase the number of awards to 164 over eight about one-fifth after 10 years. Being even more pes- years (31.5% for the period, 4–5 awards per year) simistic, if the budget were flat over that time peri- or 148 over 10 years (18.2% for the period, 2–3 od, the number of awards would decrease by about awards per year). 34% (eight years) and 40% (10 years). So, while a doubling of the NSF budget over an The bottom line: We should be optimistic, but eight- to 10-year period sounds really great, the real not euphoric. effect is much more modest. And if the number of To contact me, send an email to keith@ statistics faculty increases, the increasing number of amstat.org. Questions or comments about this article, NSF awards may not be able to keep pace. as well as suggestions for future articles, are I have assumed a doubling of the NSF budget always welcome. n would also result in a doubling of the Statistics
36 Amstat News JUNE 2009 Science Policy News This month’s guest columnists, Arlene Ash and Philip B. Stark, have been very active in election integrity work. I’m grateful for their instrumental role in shaping and guiding the ASA science policy work in this area, particularly when it comes to interacting with election officials. I especially appreciate their willingness to write this month’s column, which provides an update on this rapidly developing area and suggests a road map for further activity. ~Steve Pierson, ASA Director of Science Policy, [email protected] Thinking Outside the Urn Statisticians make their marks on US ballots Arlene Ash, Boston University School of Medicine, and Philip B. Stark, University of California at Berkeley
ree, fair, and accurate elec- — Designing, printing, and mail- tions are the cornerstone of ing election information, such as democracy, yet troubling sample ballots Ffailures of equipment, software, and procedures continue in the — Managing early, absentee, and United States. For example, Election Day voting we have all heard that some — Educating poll workers and ballots “went missing” in the public Minnesota’s 2008 famously close Senate race recount. — Allocating staff and equip- Elsewhere in 2008: Florida ment for polling places lost more than 3,400 ballots — Preparing and delivering poll initially counted on Election books and other voting materials Day; a bug in commercial elec- and equipment to election sites tion software dropped 197 bal- routine quality-control tests, lots from the totals in Humboldt — Maintaining a secure chain and—ultimately—assessing the County, California; the same of custody of all relevant mate- uncertainty in election outcomes. voting machines used in 34 rials while developing prelimi- The Help America Vote Act states lost votes in Ohio; several nary, first-reported, and certified of 2002, responding to problems states reported vote-flipping on vote counts in the 2000 presidential elec- electronic voting machines; and tion, encouraged jurisdictions to thousands of phantom votes were — Reporting Election Day results retire punchcard voting. Many switched to electronic voting. reported in Washington, DC, — Performing a full canvass and Unfortunately, most of the new inflating the apparent number of reporting final election results votes to 4,759 from a group of “direct-recording electronic” 326 actual votes. — Conducting routine audits, (DRE) voting machines pro- U.S. elections are complicated, and sometimes full recounts of duced no paper trail, so the only involving at least the following: election tallies possible checks on machine totals rely on the accuracy of the same — Quality checking and public — Registering voters and main- machines being tested. reporting taining up-to-date registries Independent verification is the keystone of good electoral audit- — Certifying (the usually many) Statisticians have much ing. Hand counting paper records candidates and measures to be on to offer election administration, is a good check on machine tal- each ballot including specifying data col- lies, as the two methods tend to lection and reporting require- err for different reasons and in — Designing, testing, and print- ments, monitoring the integ- ing (or programming) ballots, different ways. Jurisdictions that rity of data and data processing, scan voter-marked paper bal- with up to hundreds of variants designing and computing with by language and jurisdiction lots or use DREs that produce a large databases, conducting voter-verifiable paper audit trail
JUNE 2009 Amstat News 37 — To verify that vote-tallying inference from stratified samples, machines are functioning properly sampling with probability pro- portional to error bounds, mul- —To provide for continuous tiple testing, sequential testing, improvement in the conduct and combinatorially complex of elections games. Audit protocols must also meet constraints of cost, time, Statisticians can help. For exam- complexity, and transparency ple, they can do the following: to diverse audiences—the pub- lic, election integrity advocates, — Characterize the distribu- legislators, and election officials. tion and nature of vote-counting Because audits must work within errors for different voting tech- the practical limitations of elec- nologies and different methods tions and the legislative process, of hand counting statisticians should be prepared — Develop and improve the to collaborate across many disci- efficiency of practical methods to plines: political science, computer control the risk that an incorrect science, the social and behavioral outcome will go undiscovered sciences (usability studies and survey research), forensics, and — Estimate the cost and logis- “usable security”—security that tical requirements of various is adequate to the task without approaches to auditing and the impairing usability. Statisticians PhD student Luke Miratrix and Philip Stark (right) audit trade-offs among rigor, risk, time, must “think outside the urn.” the votes for Measure W in Yolo County, California, in and cost Methodological research in November of 2008. Miratrix works with Stark developing — Inform policy when there are election auditing is relevant to risk-limiting methods. competing risks and benefits—for financial auditing, experimental example, between the risk of dis- design and analysis, and many other fields in which one seeks (VVPAT) can independently enfranchisement and that of vote fraud, between cost and waiting to draw inferences about a finite check the accuracy of machine a priori time at the polls, between cost and population that satisfies subtotals, typically subtotals by bounds, but for which para- precinct. Vote-tabulation audits accuracy of vote-count technology, between security and ease of use metric approximations could be (also called post-election audits) inaccurate. Preliminary results that compare hand and machine — Identify processes and technol- suggest that techniques recently counts of a random sample of ogies that can improve accuracy or developed for election auditing subtotals can help ensure incor- enable less painful compromises are more efficient than those cur- rect outcomes are caught and rently used in financial auditing. corrected. Indeed, post-election — Develop a knowledge base audits uncovered many of the Risk-Limiting, Vote- problems cited above. The role of statisticians was Tabulation Audits discussed by David Marker, John Statisticians helped draft According to Principles and Best Gardenier, and Arlene Ash in the Principles and Best Practices for Practices for Post-Election Audits, June 2007 issue of Amstat News. Post-Election Audits (see http:// a risk-limiting audit has a pre- Here, we review recent progress, electionaudits.org/principles), specified, minimum chance of a propose a road map for further and the American Statistical full manual count whenever the progress, and suggest ways for Association Board of Directors outcome of the election is wrong. statisticians to contribute. endorsed its statistical content. By definition, an outcome is The document states the following “wrong” if it disagrees with the goals of vote-tabulation audits: A Challenging Statistical Problem outcome a full hand count would show. A risk-limiting audit is — To deter fraud At first glance, election audit- only as good as its audit trail. ing appears to be a straightfor- No trail, no audit. Only with — To promote public confidence ward sampling problem. In fact, in elections a complete and accurate audit audits raise questions about non- trail—ensured through a secure — To find error, whether acci- parametric tests for the mean of chain of custody—can any audit dental or intentional nonstandard distributions, exact provide real assurance.
38 Amstat News JUNE 2009 A risk-limiting audit can be four confirmed the provisional couched as a hypothesis test. outcomes without requiring a full The “risk” is the chance that the hand count. audit stops before a full hand Risk-limiting audits are typi- count when the outcome is cally performed in stages. Each wrong. To control that risk sta- stage involves drawing a probabil- tistically, it makes sense to take ity sample of batches, comparing the null hypothesis to be that a the preliminary results with hand full hand count would contradict counts of the audit trail for those the apparent outcome. A Type batches, and calculating a p-value I error occurs if the audit stops for the hypothesis that the out- short of a full hand count when come of the election is incor- the outcome is wrong. The goal rect. The calculation involves the is to efficiently control the Type reported votes by batch for the I error rate—that is, to count as entire race, the observed discrep- few ballots as possible while limit- ancies between the preliminary ing the risk of stopping too early counts and the hand counts, the when a full count would change sampling design (including sam- the outcome. ple sizes, stratification, and the Early statistical work on vote- protocol for advancing from one tabulation audits includes SAFE stage to the next), and the desired (statistically accurate, fair, and limit on the risk. If the discrepan- efficient), which gave the num- cies in a large enough sample are ber of audit batches to select by sufficiently small, there is strong simple random sampling as a evidence that the outcome is cor- function of the margin of victory, rect, so the audit can stop; other- precinct sizes, and a parameter wise, the audit progresses to the specifying the maximum plausi- next stage—to collect more evi- ble level of error in any precinct. dence, perhaps eventually requir- SAFE ensures that if the outcome ing a full hand count. is wrong, there is a large chance This sample-and-test strategy the audit will uncover at least requires a transparent, public, one precinct with an error, even trustworthy method of gener- if the error is concealed in as few ating random samples, public precincts as possible. SAFE is not counting, published procedures, a fully fleshed-out, risk-limiting and calculations and theorems protocol; it provides no guidance that few elections officials or for what to do when the audit others have the background to finds discrepancies, which are understand. Perhaps the biggest virtually inevitable even when the open problems for methodolo- outcome is right. gists in post-election audits now Philip B. Stark of the University concern logistical barriers to of California at Berkeley and his adoption: developing simpler, colleagues have developed several more efficient, and more usable methods for risk-limiting audits. methods; developing efficient In collaboration with county elec- ways to audit dozens of contests tion officials, they conducted four simultaneously; and developing risk-limiting audits in California data ‘plumbing,’ software tools in 2008: Marin County (a small and turn-key solutions that make Tally sheets for auditing Measure W, Yolo County, measure requiring a supermajor- the methods accessible to elec- California, November 2008 ity and a county-wide measure), tions officials and transparent to Santa Cruz County (County the public. Another important Supervisor, District 1), and Yolo task is to estimate the costs and County (a bond measure). The specific resources needed to sup- audits limited the risk—the Type port credible audits. I error rate—to at most 25%. All
JUNE 2009 Amstat News 39 Political Climate for outcome. Although no current procedures available. Legislators Post-Election Audits or pending legislation actually might be unwilling to revisit Some states have enacted vote requires risk-limiting audits, we election audits repeatedly, so tabulation audit laws that require commend states for moving jurisdictions that legislate meth- comparing machine tallies with a toward statistically motivated and ods can be stuck with poor meth- hand count of a sample of ballot justified audits. A common flaw ods indefinitely. batches. Other states are consider- of many bills is that they attempt to legislate details of sampling Collect election and audit data ing such laws. Federal legislation and make it accessible. A central requiring audits has been intro- and statistical calculations, rather than enunciating principles (e.g., repository for data—including duced several times. The laws and election results at the precinct bills vary in their goals and effect “the procedure shall have at least a 90% chance of requiring a full level or below, voting system fail- on election integrity. Some have ures, errors uncovered by audits, no enunciated goal, but require hand count whenever that count would show a different outcome”) etc.—would be extremely valu- that the audit trail receive some able. Metadata should include scrutiny. Some seek ‘quality con- and leaving implementation details to regulation. The push details about the voting technol- trol,’ in a loose sense. Some seek ogy, the vote-counting technol- to ensure that electoral outcomes for overly detailed prescriptions seems to arise from mutual sus- ogy, the audit, the hand-counting are correct. States draw their ini- procedures, and so on. tial audit samples differently; picion among the stakeholders, few require additional sampling including local, county, and state Help the public and elections if the initial audit finds errors. election officials, legislators, and officials understand basic sta- Some use a fixed number of election integrity advocates. tistical principles that apply to batches per county across all con- audits. For instance, it is a com- Vote-Tabulation Audit mon misconception that correct- tests, or a number that increases Road Map with population. Some require ing the error found in a random auditing a fixed percentage of From our experience in talking sample of batches corrects the error batches or votes. Some have with election officials, auditing in the entire contest—the idea tiered designs, whereby one of elections, and working with elec- that each item in a random sample three percentages of ballots are tion integrity advocates, we see stands for a larger number of items audited, depending on the appar- the following priorities: in the population is not universally ent margin of victory. appreciated. Similarly, jurisdictions A few bills now being con- Build the information infra- have passed laws requiring that the sidered seek to limit the risk of structure for good audits (i.e., audit sample be selected before the certifying an incorrect electoral data plumbing). Audits can only preliminary results are published. be done efficiently if data are Statisticians can help explain why promptly and publicly available that undermines the security the in machine- and human-readable audit would otherwise give. formats at the level of auditable ASA Science Policy Actions batches of ballots. Surprisingly, Statisticians possess the exper- this is rare. Commercial elec- tise to discover and characterize • The ASA sends a letter regarding tion management software seems the nature, frequency, and sourc- effects of IT consolidation on the IRS to be the bottleneck. Adopting es of problems in our elections Statistics of Income (SOI) Division and a standard terminology and data and reduce the risk that electoral letter regarding the next SOI director formats would help states ask outcomes do not truly reflect the • The 50 slots for the ASA congressional for useful data and make it easier will of the voters. The ASA as an visits filled in just two weeks, with for voting machine and software organization and several of its representatives being 25% students vendors to supply it. National members are working to develop and coming from 25 states or international standards should and disseminate model language, be considered. techniques, and best practices for • The ASA signs a letter to congressional our elections. Election officials Legislate auditing principles, leaders thanking them for their efforts have begun to come to us for not implementation details. to pass the Edward M. Kennedy guidance. We welcome your help Methods for risk-limiting audits Serve America Act (H.R. 1388) to meet the important challenge are evolving quickly. Overly pre- of measuring, improving, and Further information can be found at scriptive laws prevent jurisdic- ensuring the accuracy and integ- tions from using the most effi- www.amstat.org/outreach/scipolicyletters.cfm. rity of U.S. elections. n cient, reliable, and transparent
40 Amstat News JUNE 2009 Master’s Notebook A Learning Opportunity for Statisticians JSM Continuing Education courses provide spectrum of secondary learning Amarjot Kaur, ASA Committee on Applied Statisticians
eonardo da Vinci once said “Learning never In addition to attending conferences, there are exhausts the mind.” It is from that perspec- other venues where statisticians can reap the ben- tive that initiatives in continuing education efits of continuing education. Many ASA chapters Lseem to be so relevant to the statistical world. are active in sponsoring full-day symposia or short Continuing education represents a spectrum of sec- courses on current topics. Also, several ASA sec- ondary learning. It is an important facility that tions sponsor webinars delivered by subject-matter helps professionals keep up with the dynamic world, experts. These activities usually involve a fee, but the where change never stops and learning never ends. fees are nominal. Furthermore, many software ven- The Joint Statistical Meetings offers a variety dors offer informative webinars for free. of Continuing Education courses and Computer The form and objective of learning may vary, Technology Workshops. These courses and work- but there are many excellent avenues through which shops benefit statisticians by updating their knowl- statisticians can enrich their knowledge. Try some edge of existing techniques or helping them learn out for yourself. Visit the JSM online program at an entirely new technique. Speaking from experi- www.amstat.org/meetings/jsm/2009/onlineprogram to ence, I have learned a great deal from the variety view the scheduled Continuing Education courses of courses I have attended during JSM over the and Computer Technology Workshops. n years. For example, my first exposure to data-min- ing techniques, such as decision trees, was during a Continuing Education course several years ago. What I learned was something useful that I could contribute to my workplace. Glancing through the list of JSM offerings from the last five years, there seems to have been plenty from which to choose. At each of the recent meet- ings, there were approximately 30 Continuing Education courses and about 10 Computer Technology Workshops with applications in areas such as survival analysis, multivariate analysis, mod- eling, longitudinal analysis, Bayesian methodologies, data-mining techniques, spatial statistics, graphical displays, genetic sampling, adaptive designs, time series, survey sampling, and specific software orien- tation. The courses are offered in two-day, one-day, and half-day formats during the first four days of JSM, and the computer workshops are offered in two-hour intervals on the fifth day. There are other conferences that provide short courses and workshops, as well. These are great opportunities for master’s-level statisticians. Attending a course while at a conference or attend- ing a conference while at a course is like “killing two birds with one stone,” as it is more economical than attending the two events separately.
JUNE 2009 Amstat News 41 JSM 2009: Plan Now Play Later
With all there is to do in Washington, DC, it may be hard for ASA members to decide which sessions and activities to attend during JSM 2009. In preparation for this year’s meeting, the program chair and other ASA volunteers share their suggestions with Amstat News readers. Also, make sure to visit the ASA sections department in this issue to view the sponsored sessions, roundtables, and Continuing Education courses being offered.
4242 AmstAmstaatt NewsNews JUNEJUNE 20092009 MEETINGS
IOLs Open Eyes to the Basics Wendy Martinez, JSM 2009 Program Chair
ave you ever wanted to Spatial Data Analysis know the basics of a August 3, 8:30 a.m. new area of statistics? Noel Cressie of The Ohio State HHave you ever been baffled by University will present a lecture some of the talks at JSM and to get you thinking in a new wanted to have some background statistical way—spatially. He before you attended the sessions? will provide useful information Well, we have some opportunities about models for expressing spa- for you! tial dependence and emphasize Consider attending the hierarchical or multilevel statis- , Introductory Overview Lectures tical models. This IOL will help (IOLs) at the Joint Statistical attendees better understand and Meetings. These special invited ses- appreciate other spatial statistics will enable researchers to handle sions are open to all JSM attendees sessions at JSM. problems in many of the sciences. for no extra charge, and you do not have to sign up ahead of time to Designing Longitudinal Statistical Learning and attend them. There are five IOLs— Studies Data Mining one for each day of the meet- August 4, 8:30 a.m. August 6, 8:30 a.m. ing—so take advantage of these Donna Spiegelman of the Harvard David Hand of Imperial College, lectures and expand your School of Public Health will talk London, will provide an interest- statistical horizons. about the challenges involved ing lecture on statistical learning Largely About Largeness: in designing longitudinal stud- and data mining and examine the ies, along with various methods past, present, and future of these Models and Views for for addressing these problems. two domains. He will describe the High-Dimensional Data Software is available at www.hsph. nature of the current tools, as well August 2, 4:00 p.m. harvard.edu/faculty/spiegelman/ as the drivers for future develop- First, Antony Unwin of the optitxs.html to implement the ments. This IOL on the last day University of Augsburg will talk techniques to be described. of JSM will help attendees focus about ways to explore high- Spiegelman encourages attendees on the exciting future of statistics dimensional data through visu- to download it and bring their in these areas. n alization. His talk will include laptops to the lecture. well-known graphical tools adapted for this purpose and Causal Inference in new ideas to spark your interest. Statistics After you learn how to visually August 5, 2:00 p.m. Poster Presentations discover the structure in data, Judea Pearl of UCLA will intro- David Banks of Duke University duce causal inference in statistics. There are more than 500 poster will discuss the “curse of dimen- He will introduce basic principles presentations lined up for JSM this year. sionality” and how it arises and simple mathematical tools Every poster will be available for viewing whenever one does analysis in that are useful when solving prob- for four hours. Presenters are scheduled high dimensions. In particular, lems involving causal inference. to be there with their posters during this problem becomes worse The framework Pearl will describe one of the regular session times. Poster when the number of dimensions includes principles based on non- presentations encourage up-close exceeds the number of observa- parametric structural equation interaction with presenters, so make tions. Banks will describe some models and ideas from logic and sure to stop by, admire your colleagues’ theories and methods to help graph theory. The resulting calcu- work, and ask lots of questions. with these challenges. lus of causes and counterfactuals
JUNE 2009 Amstat News 43 44 Amstat News JUNE 2009 JSMProgram-at-a-Glance2009 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 am am am am am pm pm pm pm pm pm pm pm pm pm pm pm am
FRIDAY Committee and Business Meetings and Other Activities 7:30am–9pm July 31, 2009 Registration 7:30am–7:30pm Cyber Center 7:30am–6pm SATURDAY Speaker Management Room 10am–6pm August 1, 2009 Career Placement Service Registration ONLY 9am–5pm ASA Marketplace 12pm–5:30pm Committee and Business Meetings and Other Activities 7:30am–6pm Continuing Education Courses 8:30am–5pm Registration 7:30am–8:30pm SUNDAY Cyber Center 7:30am–10:30pm Speaker Management Room 9am–7pm August 2, 2009 Expo 2009 1pm–6pm Committee and Business Meetings and Other Activities 7:30am–8:30pm Continuing Education Courses 8:30am–5pm ASA Marketplace 9am–5:30pm First-Time Career Placement Service 1pm–6pm Attendee Orientation & Reception Opening Mixer 6pm– Technical Sessions 2pm–5:50pm 7:30pm 8pm–10:30pm Registration 7:30am–6pm MONDAY Cyber Center 7am–10pm Speaker Management Room 7am–6pm August 3, 2009 Expo 2009 9am–6pm Committee and Business Meetings and Other Activities 7am–10pm Continuing Education Courses 8am–5pm Career Placement Service 8am–6pm Round- ASA Marketplace 9am–5:30pm tables Speaker & Technical ASA IMS with Technical Sessions Round- Sessions President’s Student Mixer Presidential Coffee 8:30am–12:20pm tables 2pm–3:50pm Invited Address with Lunch 6pm–8pm 7am– Poster Sessions 12:30pm– Poster Sessions Address 8pm– 8:15am 8:30am–10:20am 1:50pm 2pm–3:50pm 4pm–5:50pm 9:30pm 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 am am am am am pm pm pm pm pm pm pm pm pm pm pm pm am Registration 7:30am–4:30pm Cyber Center 7am–10pm TUESDAY Speaker Management Room 7am–6pm Expo 2009 9am–6pm August 4, 2009 Committee and Business Meetings and Other Activities 7am–9pm Continuing Education Courses 8am–5pm ASA Career Placement Service 8am–6pm President's Round- ASA Marketplace 9am–5:30pm Informal Dance Address & tables Speaker & Technical Party Technical Sessions Awards with Round- Sessions Deming 9:30pm–12am 8:30am–12:20pm tables 2pm–3:50pm 8pm– Coffee with Lunch Lecture 7am– Poster Sessions 12:30pm– Poster Sessions 4pm–5:50pm 9:30pm 8:15am 8:30am–10:20am 1:50pm 2pm–3:50pm Registration 7:30am–4:30pm WEDNESDAY Cyber Center 7am–8pm August 5, 2009 Speaker Management Room 7am–6pm Expo 2009 9am–2:30pm Committee and Business Meetings and Other Activities 7am–7:30pm Computer Technology Workshops 8am–4:45pm Career Placement Service 8am–2:30pm ASA Marketplace 9am–5:30pm Round- tables Speaker & Technical Technical Sessions Round- COPSS with Sessions 8:30am–12:20pm tables 2pm–3:50pm Awards and Coffee with Lunch Fisher Lecture 7am– Poster Sessions 12:30pm– Poster Sessions 4pm–5:50pm 8:15am 8:30am–12:20am 1:50pm 2pm–3:50pm Registration 7:30am–10:30am THURSDAY Cyber Center August 6, 2009 7am–10:30am Speaker Management Room 7am–12pm Committee and Business Meetings and Other Activities 7am–11:30am ASA Marketplace 7:30am–10:00am Technical Sessions 8:30am–12:20pm
JUNE 2009 Amstat News 45 Washington, DC Triv a?
1. The Lincoln Memorial has 36 columns. 6. The National Archives houses the famous docu- What do the 36 columns represent? ments dedicated to American freedom and democ- A) Number of 10,000 of soldiers who died in the Civil War racy: Declaration of Independence, Bill of Rights, and Constitution. Who were the authors of these B) The number of president Lincoln was documents? (An individual can be chosen more C) The number of states Lincoln held together before his than once.) death on April 15, 1865 Declaration of Independence George Washington D) Number of 100,000 of slaves Lincoln freed when he Bill of Rights Benjamin Franklin signed the Emancipation Proclamation on January 1, 1863 Constitution John Adams 2. How tall is the Washington Monument? James Monroe A) 444 feet James Madison B) 555 feet Thomas Jefferson C) 666 feet 7. What was Abraham Lincoln watching in Ford’s D) 777 feet Theater on the fateful night of April 14, 1865? 3. Who was the first president to serve a full term in A) “Our American Cousin” the White House? B) “How to Lose a Guy in 10 Days” A) George Washington C) “Ion” B) John Adams D) “King Lear” C) Thomas Jefferson E) “Macbeth” D) James Madison 8. If you were to run all the hallways of the 4. What famous politician sold his eclectic Pentagon, how far would you have run? book collection to the U.S. government, which A) 2.5 miles started the Library of Congress? B) 6.8 miles A) Thomas Jefferson C) 11.9 miles B) Benjamin Franklin D) 17.5 miles C) George Washington D) John Smith 9. Before Virginia took back the land they gave to make the District of Columbia, how many square 5. This individual was credited for helping to push miles was our nation’s capital? Congress to move the Supreme Court from the A) 10 miles2 basement of the Capitol to the building we see B) 100 miles2 today. Need more help? This individual has the dis- 2 tinction of being the only head of two branches of C) 1,000 miles the U.S. government. Still more? He has the distinc- D) 10,000 miles2 tion of being both the heaviest U.S. president and chief justice to date. 10. Only two people have the power to arrest the president. Who are they? A) William H. Taft A) The vice president and the secretary of state B) Grover Cleveland B) The secretary of war and his/her deputy C) James A. Garfield C) The Senate sergeant-at-arms and his/her deputy D) Rutherford B. Hayes D) The mayor of Washington, DC, and his/her lieutenant
mayor
10) C 10) B 9) D 8) A 7)
6)i. Thomas Jefferson ii. James Madison iii. James Madison Madison James iii. Madison James ii. Jefferson Thomas 6)i.
A 5) A 4) C 3) B 2) C 1) Answers:
46 Amstat News JUNE 2009 No Car, No Problem: Getting Around During JSM
The DC subway system, known as the Metro by locals The Circulator is a bus system that operates several routes in and near downtown Washington, DC.
he Joint Statistical Meetings will take place passes through the National Mall and ends at the in downtown Washington, DC, this year, Southwest Waterfront. Other Circulator routes go to which makes public transportation a good Georgetown and Adams Morgan, which are popular Ttravel option. The DC subway system, known as travel destinations in DC that do not have Metro the Metro by locals, has a station adjacent to the stations in the immediate area. The Circulator will Washington Convention Center called the Mount accept SmarTrip for payment of fare. Similar to Vernon Square Station. Metrobus, SmarTrip users automatically receive a Metro fare can be paid using either paper fare transfer when paying Circulator fare that allows free cards or a plastic SmarTrip Card; both can be pur- reboarding of any Circulator within a three-hour chased at vending machines at Metro stations. Paper period. For more information about Circulator, visit fare cards can be used for Metro travel only. The www.dccirculator.com. plastic SmarTrip Card has a one-time $5 charge in addition to the amount added to it to cover rides. Theodore Roosevelt Island and the It can be used for both Metro and Metrobus travel Mount Vernon Trail and is required to obtain transfer discounts between Just across the Potomac River from downtown DC Metro and Metrobus, or vice versa. Metrobus fare is the Mount Vernon Trail, a popular trail for walk- paid with SmarTrip is slightly less than cash fare. ers, runners, and bicyclists. The Mount Vernon Trail Metrobuses run from both BWI and Dulles extends 15 miles south to Mount Vernon. A map airports to Metro stations, offering an economical of the trail is available at www.nps.gov/archive/gwmp/ travel option. Reagan National Airport has an adja- mvtmap.html. Pedestrians, runners, and bicyclists cent Metro station. Schedules for the BWI Airport can access the trail from downtown DC using paths Metrobus (B30) and the Dulles Airport Metrobus on several bridges that cross the Potomac River. (5A) are available at www.wmata.com. Both routes The bridge that carries I-395 traffic from DC to terminate at a Metro station with Green Line ser- Virginia has a footpath that starts near the Jefferson vice, which is one of the two Metro lines that service Memorial. The Theodore Roosevelt Bridge near the the Mount Vernon Square Metro station adjacent to Lincoln Memorial also has a footpath. the Walter E. Washington Convention Center. The northern end of the Mount Vernon Trail Paper copies of Metrobus schedules and Metro is at Theodore Roosevelt Island, which is accessi- and Metrobus maps are available in the lobby of ble only by footbridge. The island has a system of the Washington Metropolitan Area Transit trails and an outdoor memorial to Roosevelt. More Authority (WMATA) headquarters building at 600 information about the island is available at www. 5th Street NW. nps.gov/archive/gwmp/tri.htm. The island is within walking distance of the Rosslyn Metro station, The Circulator via a path and bridge that link the northern end The Circulator is a bus system that operates several of the Mount Vernon Trail to a system of hiker- routes in and near downtown Washington, DC. biker trails in Virginia (the Custis-I-66 Trail and the One of the Circulator lines is a north/south route, W. & O.D. Trail). n with the northern end at the convention center. It
JUNE 2009 Amstat News 47 RTI Supports Cox Race, Honors Namesake Roy W. Whitmore, RTI International
This year, as RTI celebrates its own “50th anniversary, the institute proudly honors Cox and her contributions to the field of applied statistics. ”
Getrude Cox
elected into the International Statistical Institute in 1949, serv- ing as president of the American Statistical Association in 1956 (a position currently held by Sally Morton, who directs statistical and epidemiological research at RTI), and being elected to the National Academy of Sciences in 1975. RTI has recognized Cox by naming a building on its main campus for her and, begin- ning in 2003, joining with the Washington Statistical Society From left: William G. Cochran, Gertrude M. Cox, George W. Snedecor, and C. P. to sponsor the Gertrude M. Cox Winsor in 1940 Award for contributions in sur- vey methodology, experimental design, biostatistics, or statistical n honor of the 20th anniver- early planning for a new research computing. sary of the Gertrude M. Cox institute in North Carolina. That This year, as RTI celebrates its Scholarship, sponsored by the institute became RTI. When the own 50th anniversary, the institute IASA Committee on Women in institute was fully formed, Cox proudly honors Cox and her con- Statistics and the Caucus for became director of the statistics tributions to the field of applied Women in Statistics, RTI research division, a position she statistics. We congratulate the International is lending support held until her retirement in 1964. organizers of the Cox Scholarship to this year’s Cox Scholarship Cox’s team was highly sought and the race on two decades of Race. An anticipated event dur- after from the beginning. All but supporting the future achieve- ing the annual Joint Statistical two of RTI’s first 20 research con- ments of women in statistics. Meetings, the race will take place tracts were in statistics, including For more information about on Tuesday, August 4. the institute’s first education proj- the scholarship or to register for RTI proudly traces its early ect, its first private-sector contract, the race in advance, visit http:// history to the “first lady of sta- and its first international project. caucusforwomeninstatistics.com tistical science,” who led the Cox’s notable achievements or www.amstat.org/committees/ committee that, in 1958, did the include becoming the first woman cowis. n
48 Amstat News JUNE 2009 RTI Supports Cox Race, The 20th Annual Honors Namesake GERTRUDE COX SCHOLARSHIP RACE Roy W. Whitmore, RTI International 5k Race and 2.5k Fun Run/Walk at JSM 2009, Washington, DC
The Caucus for Women in Statistics, in conjunction with the ASA, presents the 20th annual Gertrude Cox Scholarship Race at the Joint Statistical Meetings 2009 in Washington, DC. All proceeds will benefit the Gertrude M. Cox Scholarship in Statistics. This year, the race will be sponsored by RTI International.
The Race: Competitive 5k race and a 2.5k fun run/walk, running concurrently When: Tuesday, August 4, time to be announced (to start and finish before 8:30 a.m.) Where: Location and logistical information will be posted at e Caucus for Women in Statistics web site as it becomes available (www.caucusforwomeninstatistics.com) and at e Caucus for Women in Statistics hospitality table at the convention center during JSM 2009 How Much: e entry fee is $30.00 Registration ose interested in participating are encouraged to register early by mail, but Procedure: may also register during JSM at the hospitality table for the Caucus for Women in Statistics at the Convention Center, near the registration area. All participants must sign a registration form and waiver. T-shirts for all pre-registered runners will be distributed at the hospitality table.
REGISTRATION FORM (each participant must complete and sign)
______Name
______Address City
______State/Province ZIP/Postal Code Phone
Required for placement and timing in the 5k race: SEX: ❏ M ❏ F Age ______EVENT: ❏ 5 k Race ❏ 2.5 k Fun Run/Walk T- SHIRT SIZE: ❏ S ❏ M ❏ L ❏ XL
THE FINE PRINT: I understand that running a road race is a potentially hazardous activity. I will not enter and run unless I am medically able and properly trained. I agree to abide by any decision of a race official relative to my ability to complete the run safely. I assume all risks associated with running in this event, including, but not limited to, falls; contact with other participants; and effects of weather, traffic, and course conditions. All such risks are known and appreciated by me. Having read this waiver, knowing these facts, and in consideration of your accepting my entry, I, for myself and anyone entitled to act on my behalf, waive and release the race directors, the race committee, and all sponsors from all claims of liabilities of any kind arising out of my participation in this event, even though such liability may arise as a result of negligence or carelessness on the part of the persons named in this waiver.
______Signature Date
______Parent or guardian (if under 18)
Make check payable to The Gertrude Cox Scholarship Fund and mail with registration to: We regret to inform that we will not be able to accept registrations online this year. For questions, contact: Anna Nevius, Treasurer Marcia A. Ciol ([email protected]) or 7732 Rydal Terrace Anna Nevius ([email protected]) Rockville, MD 20855-2057
JUNE 2009 Amstat News 49 Practical Forum Brings Together Shop Industry, Academic Statisticians ASA Tammy Massie, FDA, and Carmen Mak, Schering-Plough he 13th annual FDA/ Bootstrap Methods and Industry Statistics work- Permutation Tests by shop is scheduled for Tim Hesterberg TSeptember 23–25 at the Capital Hilton in Downtown Washington, Three plenary sessions will DC. It is sponsored by the focus on regulatory and statisti- Biopharmaceutical Section of the cal issues in global harmonization American Statistical Association, and future directions in safety with cooperation from the Food analysis. Concurrent sessions will and Drug Administration cover timely topics in adaptive Statistical Association. More than designs, noninferiority trials, bio- 600 statisticians from academia, markers, quality of clinical pro- industry, and the FDA will gather grams, designs of integrated safety for lively discussions on topics of and efficacy analysis plans, CMC interest to all. issues, food safety, devices trials, The workshop will feature a vaccine trials, and veterinary trials. day of short courses followed by This year, the luncheon dis- two days of plenary and concur- cussions have been expanded to rent presentations, as well as lun- two days and follow a new for- cheon discussions on a variety of mat. Boxed lunches will be served; topics. Back by popular demand, workshop participants will be able the following eight short courses to pre-select lunch at the time of will offer practical solutions to conference registration. This will real-life problems: allow participation at the round- tables to be available on a first- come, first-served basis. In addi- Adaptive Design by Martin tion, roundtables will be smaller Posch and Sue-Jane Wang to foster better discussion. Purchase T-shirts, Data Monitoring Committee by The Capital Hilton is a pre- Tom Fleming and Janet Wittes mier meeting site, located in the books, JSM Proceedings, heart of downtown Washington, Analysis of Longitudinal Data DC. A block of sleeping rooms and gift items! with Missing Data by has been arranged for workshop Diane Fairclough attendees. To receive the spe- cial rates, attendees must book Visit the ASA’s online Cox Regression in Practice by their reservations by August 26. Brenda Gillespie marketplace at Visit www.amstat.org/meetings/ www.amstat.org/asastore Special Topics in Survival fdaworkshop for additional reg- Analysis by Lee-Jen Wei istration and hotel information. Conference organizers and others Bayesian Adaptive Methods by interested in up-to-date informa- SAVE 10 % Peter F. Thall tion, including the preliminary on your first purchase by Benefits: Risk Assessment by workshop programs and luncheon discussion topics, can visit http:// entering ASASTORE at checkout. Scott R. Evans fdaindustry09.blogspot.com. n
50 Amstat News JUNE 2009 REGISTRATIONFORM 2009 FDA/Industry Statistics Workshop Sponsored by the ASA Biopharmaceutical Section with cooperation from the FDA Statistical Association 3EPTEMBER n s #APITAL (ILTON7ASHINGTON $# INSTRUCTIONS 1. Print or type all information and retain a copy for your records. 2. Use a separate form for each registrant. 3. Mail form with payment to FDA/Industry Statistics Workshop Registration, P.O. Box 221, Annapolis Junction, MD 20701. Fax form (credit card only) to (301) 206-9789. www.amstat.org/ 4. Registration form must be received by August 26, 2009, to be processed at the reduced rate. meetings/fdaworkshop Forms Received Without Payment Will Not Be Processed.
ATTENDEE INFORMATION MEAL PREFERENCE Lunch on Thursday, September 24, and Friday, September 25, is included with your workshop registration. To assist in planning, please select the area of discussion that most interests you. See Page 2 of this form for the list of topics. Name 3ELECT ONE OF THE FOLLOWING MENU OPTIONS FOR EACH DAY Thursday, September 24 Friday, September 25 Preferred Name for Badge Topic:______Topic:______❑ Regular ❑ Regular ❑ Vegetarian ❑ Vegetarian ASA ID # (if known) ❑ Not attending lunch ❑ Not attending lunch (Regular=turkey sandwich; Vegetarian=veggie wrap) All boxed lunches include Organization chips, a piece of fruit, and a cookie.
REGISTRATION FEES Workshop Fee (required) Address By August 27– August 26 September 15 ❑ Registrant $270 $295 $______❑ Academic (nonstudent) $230 $255 $______❑ Biopharm Section Member $230 $255 $______❑ City State/Province ZIP/Postal Code Government $130 $155 $______❑ Student $130 $155 $______Short Courses—7EDNESDAY, September 23 Country (non-U.S.) Add-ons to Workshop Fee: $100 each before Aug. 27; $105 each Aug. 27–Sept. 15 8:00 a.m.–12:00 p.m. Phone ❑ SC1: Adaptive Designed Clinical Trials—Martin Posch & Sue-Jane Wang $______❑ SC2: Benefit: Risk Assessment—Scott R. Evans $______Email ❑ SC3: Analysis of Longitudinal Studies with Missing Data— Diane Fairclough $______In case of emergency, list the name and phone number of the person ❑ SC4: Cox Regression in Practice—Brenda Gillespie $______we should contact (remains confidential). 1:00 p.m.–5:00 p.m. Emergency Contact’s Name ______❑ SC5: Special Topics in Survival Analysis—Lee-Jen Wei $______Relationship to Registrant ______❑ SC6: Bootstrap Methods and Permutation Tests— Tim Hesterberg $______Telephone Number ______❑ SC7: Recent Advances in Bayesian Adaptive Methods for Clinical Tests—Peter F. Thall $______❑ Check here if you would like your ASA customer contact ❑ information updated with your meeting contact information. SC8: Data Monitoring Committees—Tom Fleming & Janet Wittes $______This meeting is ADA accessible. TOTAL FEES: $______❑ Please check here if you need special services due to a disability and attach a statement regarding your needs. PAYMENT ❑ Check/money order payable to the American Statistical Association (in U.S. dollars on U.S. bank) CANCELLATION POLICY #ANCELLATIONS RECEIVED BY !UGUST WILL BE REFUNDED LESS A PROCESS- Credit Card ❑ VISA ❑ MasterCard ❑ American Express ING FEE AND LESS A PROCESSING FEE FOR EACH SHORT COURSE #ANCELLATIONS RECEIVED BY 3EPTEMBER WILL BE REFUNDED LESS A PROCESSING FEE AND LESS A PROCESSING FEE FOR EACH SHORT COURSE 2EQUESTS FOR REFUNDS RECEIVED AFTER Card Number 3EPTEMBER WILL NOT BE HONORED !LL CANCELLATIONS MUST BE MADE IN WRITING TO [email protected] VIA FAX TO OR MAILED TO &$!)NDUSTRY Exp. Date CVS# (3- digit number on back of card) 3TATISTICS 7ORKSHOP 2EGISTRATION 0/ "OX !NNAPOLIS *UNCTION -$ *0URCHASE ORDERS WILL NOT BE ACCEPTED .O EXCEPTIONS !3! &EDERAL