Faculty - Graduate Certificate in Applied Statistics
James L. Rosenberger, PhD
Program Chair
Dr. Rosenberger has interests in statistical applications and the design of experiments for disciplines including agriculture, ecology, genomics, medicine, and transportation. Complex designs are frequently encountered in situations where physical restrictions, ethical limitations, or fiscal constraints prevent application of straightforward comparative experiment protocols. Nonstandard research designs often require nonstandard statistical analyses to reflect the degree of uncertainty and validity of an experiment.
Another area of interest has been statistical computing methodology and algorithms. He developed the algorithms for balanced, orthogonal analysis of variance, and more general linear model routines (now in the Minitab statistical package) that provide estimates and hypothesis tests for unbalanced data often encountered in observational studies.
Dr. Rosenberger served as department head from 1991 to 2006. Previously he served as the founding director of the Statistical Consulting Center. In 2012, Dr. Rosenberger was elected to the position of vice president of the American Statistical Association (ASA), and formerly was chair, program chair, and newsletter editor of the Statistical Computing Section of the ASA. During a leave from 1998 through 2000 he served as the statistics program director at the National Science Foundation. He was formerly an editor of the journal Statistical Science.
- Steven F. Arnold
- Steve Bai
- Indrani Basak
- Prasanta Basak
- Srabashi Basu
- Mosuk Chow
- Eugene J. Lengerich
- Megan Romer
- Scott Roths
- Laura J. Simon
- Aleksandra B. Slavković
- Andrew J. Wiesner
- Derek S. Young
Steven F. Arnold, PhD
Dr.Steven Arnold studies the effect of reducing by invariance before applying likelihood procedures. Examples indicate that in many situations the likelihood procedures are improved, and in nearly all situations they are no worse for this reduction. Certain types of invariance reductions may lead to generalization of the idea of ancillary statistics. Unfortunately, however, there are situations in which reducing by invariance may adversely affect likelihood procedures. Dr. Arnold hopes to find out when the invariance reduction can be taken safely and when it may lead to trouble.
His main research interest is statistical inference for models involving patterned covariance matrices. Although some attention is paid to methods for testing for the adequacy of these models, the primary emphasis is on finding procedures for drawing inference for the mean vector, when we assume that the covariance matrix has the assumed structure. Recently he has been studying antedependence models.
A second research interest is in improved estimators of the distribution function of samples of independently identically distributed random variables. One goal of this research is to find improved methods of doing bootstrapping, a procedure in which repeated samples are taken from the estimated distribution function and used to draw conclusions about the true distribution function, using very few assumptions.
Steve Bai, PhD
Dr. Steve Bai has been a mathematical reviewer for the U.S. Food and Drug Administration (FDA) since 2005 after earning a doctorate in statistics from Penn State. Throughout his career in Center for Drug Evaluation and Research (CDER) and the FDA, he has reviewed many regulatory submissions seeking indications in cardio/renal, neurology, and psychiatry. Based on his review work, his areas of research interest include adaptive design, non-inferiority trials, meta-analysis, multiplicity adjustment, and multi-regional trials.
Indrani Basak, PhD
Dr. Indrani Basak’s main research interests are in analytic hierarchy process (AHP), robust statistical methods, and censored data problems. AHP is a decision-making tool for problems with multiple criteria and multiple alternatives, especially useful in the business field. Dr. Basak studied methods of assigning statistical values to alternative priorities. AHP is related to paired comparison methods of considering alternatives two at a time.
Dr. Basak has also developed new applications within the established field of robust statistical analysis. These are ways of determining if the statistical methods being used are still valid when the assumptions they are based on are relaxed. Some of her interests in those analysis concern survival analysis and reliability of systems which are useful in the medical and engineering sciences.
Dr. Basak’s recent area of work is in statistical inferential and prediction methods in the setup of progressive censoring, a special censoring method which is relatively new. Censoring occurs when exact survival times are known only for a portion of the individuals or items under study. The complete survival times may not have been observed by the experimenter either intentionally (to reduce the cost and time of the experiments) or unintentionally (subjects may have withdrawn from the study or could not be tracked down).
Prasanta Basak, PhD
Dr. Basak’s research interests have been in the field of finite mixture models, record statistics and paired comparison methods. In finite mixture models, the distribution of a process is assumed to be a weighted combination of two or more distributions. The main challenges in this field are to estimate the parameters of the distributions and the number of component distributions. For record statistics, which are closely related to order statistics, we can make conclusions based on a set of records. In pairwise comparison experiments, we compare two items or treatments based on certain criteria. Recently, Dr. Basak’s research interest has extended to prediction problems and progressive censoring methods. In particular, he is interested in problems such as predicting a future record value given past record values, prediction of failure times in progressively censored samples and parameter estimation for three-parameter lognormal distribution under progressive censoring.
Srabashi Basu, PhD
Dr. Srabashi Basu has been pursuing research in applied and theoretical statistics for about two decades. She has worked with epidemiologists and medical researchers both in the United States and in India. She has provided service to a major airline and had spearheaded the development of a forecasting suite for passenger arrival and resource allocation. In her current position she is working for one of the largest contract research organizations (CRO) in the drug development domain. Dr. Basu is also very interested in the upcoming field of analytics and undertakes various consulting problems in different domains, such as media, marketing, risk analytics, and revenue management.
Dr. Basu has original research publications in a number of refereed journals in statistics, biostatistics, and medicine. She has several book chapters to her credit and is an international contributor to a business statistics text book.
Dr. Basu also likes to teach and was with the Indian Statistical Institute for nearly ten years. She has been teaching regularly for Penn State’s World Campus since the spring of 2009. In India she works as a corporate trainer for various players in analytics domain and for Indian Institutes of Management.
Mosuk Chow, PhD
Dr. Chow's areas of research interest include biostatistics, statistical decision theory, Bayesian inference, and sampling methods. An important question in statistical decision theory is to characterize the set of all optimal procedures. An admissible procedure is optimal in the weak sense that it cannot be outperformed by another procedure completely in all circumstances. It is thus desirable to find necessary conditions for admissible procedures. Her work in decision theory involves finding such necessary conditions, investigating the admissibility properties of various estimators for problems arising from biology, genetics, and fishery.
Since for most cases a necessary condition for admissibility is that the procedure corresponds to a generalized Bayes rule, Dr. Chow's research also covers Bayesian inference. With recent advances in Bayesian computation methods, she has used Markov chain Monte Carlo methods in some of her work. Currently she is interested in Bayesian inference for aggregated distributions under various sampling schemes and Bayesian approach to problems related to biostatistics.
Eugene J. Lengerich, VMD, MS
Dr. Lengerich is professor in the Department of Public Health Sciences at Penn State’s College of Medicine, and director of the Community Sciences and Health Outcomes (CSHO) Core at the Penn State Hershey Cancer Institute. He brings 24 years of experience in epidemiology with a focus on participatory research, public health informatics and disease surveillance, and health disparities. His research has been funded by the National Cancer Institute, Centers for Disease Control and Prevention, Health Resources and Services Administration, Lance Armstrong Foundation, American Cancer Society, Pittsburgh Affiliate of Susan G. Komen for the Cure®, and the Pennsylvania Department of Health. He has led the analysis of registry data to quantify geospatial patterns in disease incidence and morbidity, developed a model digital cancer atlas for Pennsylvania, and used these analyses to design intervention strategies. This research has also led to the development of visualization software and methods.
Prior to joining the faculty at Penn State, he was the state chronic disease epidemiologist for North Carolina, during which time he directed the Behavioral Risk Factor Surveillance System (BRFSS) and evaluated the quality of the North Carolina Central Cancer Registry. Since 2004, he has been the principal investigator of the Northern Appalachia Cancer Network, which was recognized nationally in 2009 by the Association of Public and Land-grant Universities for engaged research.
Dr. Lengerich directs an online graduate program in public health preparedness, with an emphasis on disease and syndromic surveillance. Dr. Lengerich received epidemiologic training via the Epidemic Intelligence Service program and the Preventive Medicine Residency at the Centers for Disease Control and Prevention in Atlanta, Georgia. Dr. Lengerich enjoys teaching epidemiology and mentoring pre- and post-doctorate scholars.
Megan Romer, PhD
Dr. Megan Romer’s main interest is in teaching statistics. She has been teaching online since 2009 when she earned her doctorate from Penn State's Department of Statistics. She enjoys discussing statistical concepts and problems with students. Before returning to school to finish her doctorate, Dr. Romer worked in clinical trials as a senior research support associate. Dr. Romer’s primary area of research is in incomplete data.
Scott Roths, PhD
Dr. Scott Roths earned his doctorate in August, 2011 from Penn State and has been teaching as a faculty lecturer since the fall of 2011. He also has considerable teaching experience at the undergraduate level as well.
Dr. Roths is currently investigating ways to control the false discovery rate (FDR) and false discovery proportion (FDP) in multiple testing situations. The FDR and FDP are often the preferred error quantities to control because they do not become prohibitively restrictive, unlike the traditional family-wise error rate, when the number of tests is very large. Some notable examples of large-scale testing are genetic profile analysis and astronomical source detection. Among the most popular methods to control the FDR are the stepwise procedures, where test statistics or p-values are ordered by significance and compared to pre-specified cutoffs. These procedures are easy to use, but they focus only on the binary indicators for whether the test statistics exceed their cutoffs or not. He is investigating the effect of this discretization on the power and variance of these procedures and methods that do not use such indicators.
Laura J. Simon, PhD
Dr. Simon is primarily interested in teaching statistical concepts to undergraduate math majors and to nonstatisticians in the biostatistical and health sciences fields. As her courses illustrate, she is a proponent of active, hands-on, and web-based statistical education. Dr. Simon was nominated for the 2004 Eberly College of Science's C. I. Noll Award for Teaching Excellence. She was also nominated by two students and subsequently named to Who's Who Among America's Teachers in 2004 and 2005.
Dr. Simon co-authored nine units of the web-based introductory statistics text, Visualizing Statistics (Cybergnostics, Inc.). She is a reviewer of statistical education resources submitted to CAUSEweb.org and merlot.org. Other professional interests include biostatistical consulting, repeated measures modeling, clinical trials development, research data management, and web-page programming.
Before moving to the University Park campus in 1996, Dr. Simon worked at the Center for Biostatistics and Epidemiology at the Penn State College of Medicine at the Milton S. Hershey Medical Center. While there, she consulted with more than 40 medical researchers and provided overall leadership to the data management unit on numerous projects, including the National Institutes of Health-funded National Interstitial Cystitis Database Study.
Prior to working at Penn State, Dr. Simon was a statistician at Auke Bay Research Laboratories in Auke Bay, Alaska, and a visiting summer statistician at Rohm and Haas Research Laboratories in Spring House, Pennsylvania.
Aleksandra B. Slavković, PhD
Dr. Slavković's past and current research interests include usability evaluation methods, human performance in virtual environments, statistical data mining, application of statistics to social sciences, algebraic statistics, and statistical approaches to confidentiality and data disclosure. Her Ph.D. dissertation work focuses on statistical methodologies for disclosure limitation and data confidentiality, and presents new theoretical links between disclosure limitation, statistical theory, and computational algebraic geometry. It is a unique and interesting integration of diverse results from conditional specification of joint distribution, graphical models, disclosure limitation, and algebraic statistics.
Dr. Slavković served as a consultant to the National Academy of Sciences/National Research Council Committee to Review the Scientific Evidence on the Polygraph in 2001 and part of 2002. In 2003 she received an honorable mention for the best student paper from the Committee on Statisticians in Defense and National Security of the American Statistical Association.
Andrew J. Wiesner, PhD
Dr. Andrew Wiesner's primary research interests involve educational statistics. He has presented several faculty workshops on interpreting item statistics to improve exams and the fundamentals of test item writing. Currently his research entails how frequent testing of students can improve their overall performance and increase learning. This includes the use of several web tools that are available to deliver online assessments.
Derek S. Young, PhD
Dr. Derek Young’s primary research interests are mixture models and tolerance intervals. Mixture models are statistical models that represent the presence of subpopulations in an overall population, but without a mechanism that identifies to which subpopulation each observation belongs. There are many interesting computational and theoretical questions around these models and Dr. Young’s research focuses on mixture structures within regression models. Statistical tolerance intervals are the lesser-known cousin to confidence and prediction intervals. They are statistical bounds for a specified proportion of the sampled population at a given confidence level. Tolerance intervals are often used in engineering applications as well as in research in the pharmaceutical industry.
Dr. Young is also interested in statistical computing and algorithms. Computational aspects of Dr. Young’s research in mixture models and tolerance intervals include the use of expectation-maximization (EM) algorithms and Monte Carlo methods. Most of his work is available in the two R packages that he developed and currently maintains: mixtools and tolerance.
After receiving his doctorate from Penn State in 2007, Dr. Young worked at the Bettis Atomic Power Laboratory in Pittsburgh, Pennsylvania as a senior statistician. His work there concerned developing statistical models and statistical process control procedures as they pertained to the Naval Nuclear Propulsion Program. He was also responsible for internal statistical training of the laboratory engineers and scientists. Since 2011, Dr. Young has been a Research Mathematical Statistician with the U.S. Census Bureau in Washington, DC. His work and research have been focused on improving embedded experimental designs in surveys as well as the application of tolerance intervals for complex surveys. Since 2008, Dr. Young has also concurrently been a lecturer of statistics at Penn State.