 Proceedings
|
Invited TalksTopic 6 : Statistics education, training and the workplaceProfessionals working in many different areas are faced with the need to gather, process, analyze, interpret and/or report on data as part of their job. The use of statistical thinking, methods and tools is a necessary part of work in areas such as business, economics, government, law, politics and science, as well as for society at large. In the current workplace, more sophisticated and complex analyses are possible with a proliferation of desktop software. Workers and professionals, however, have diverse educational backgrounds with respect to these statistical tools and techniques. Many did not have the proper education and training required to perform these tasks with confidence and efficiency. The size of the statistical literature is vast, the impact of technology (communications and computing, in particular) on the methods and practices is substantial, and no single individual can aim to master all that is happening. Hence even statistically literate professionals who had their education in the past will need to revise their training and update their practice to continually learn about new ideas, techniques and approaches that appear, which may question or improve upon previous behavior and methods. These two sets of circumstances pose a substantial challenge to those who need statistical thinking, methods and tools for their professional performance. Continued statistics education and training is a worthy goal to be pursued, and a substantial part of this must occur at the workplace. The aim of this set of sessions is to bring together educators, professionals, practitioners and customers of statistical thinking, methods and tools to discuss ways in which effective statistical education and training can be developed at the workplace to help promote improved statistical practice in all avenues of professional life.Session 6A: Environmental statistics6A1: The need for teaching weighted distribution theory: illustrated with applications in environmental statisticsLyman McDonald Western EcoSystems Technology, United StatesI review weighted distribution theory and applications in observational studies where biased data arise. Recorded observations will be biased and not have the original distribution unless every observation is given an equal chance of being recorded. G. P. Patil and C. R. Rao in 1977 wrote “Although the situations that involve weighted distributions seem to occur frequently in various fields, the underlying concept of weighted distributions as a major stochastic concept does not seem to have been widely recognized.” The same quote is applicable today. Our profession is missing an opportunity to provide structure and understanding to a collection of isolated statistical methods for analysis of data from observational studies. Specifically, I present applications in study of wildlife and fish populations that should motivate undergraduate and graduate students of statistics who have an interest in environmental issues. Paper 6A2: Amarillo by morning: data visualization in geostatisticsWilliam Harper Otterbein College, United StatesIsobel Clark Alloa Business Centre, United Kingdom “Amarillo by morning, Amarillo’s where I'll be” comes from a country song released by George Strait in the 1980's. Around this time Bill moved to the Amarillo Texas area hoping to bury highlevel nuclear waste. Like a fine country song, this presentation paints a haunting visual image of the Wolfcamp aquifer underlying the waste site. If a breach occurred, how many mornings until the nuclear waste arrived at Amarillo? Using a variety of visualization tools blended with geostatistical methods, an overview of universal kriging is given at an introductory level. This project was the source of Bill's first geostatistical analysis and was sadly killed in 1988 by the US government. On the plus side, he met his lovely Texas bride Paula there. Meanwhile he and coauthor Isobel have continued their geostatistical journey teaching classes which range from 14 year-old Slovakian High School kids to PhD candidates and beyond. Paper 6A3: Statistics education in a conservation organisation — towards evidence based managementIan Westbrooke Department of Conservation, New ZealandThe Department of Conservation manages 30 percent of New Zealand, protecting biodiversity and promoting recreation. Our 1800-plus staff include several hundred science graduates. DOC's first dedicated statistician started in 2000, with improving design and analysis skills as first priority. Many staff require basic skills in data entry and exploration. Some need specialist statistical knowledge, especially modelling skills. Most data is observational in nature, with interest centring on estimation of effects (and their error). However most graduates’ training focused on experiments and hypothesis tests. We developed a course to broaden knowledge of the linear model, and extend it to glms; then added one on longitudinal data analysis using mixed models. The next priority is training in practical sampling design. For our purposes, a stronger statistical modelling approach is needed in university training, with less emphasis on hypothesis tests. Paper Session 6C: Statistics training for researchers in other disciplines6C1: Some different models for interacting with researchers and students in other disciplinesBryan Manly West Incorporated, United StatesIn this talk I discuss my experiences in terms of giving classes and workshops on statistics to researchers and students in other disciplines, working with individual researchers, writing books to introduce statistical ideas in a particular topic area to researchers and scientists, and situations where the statistician is part of a large group that is investigating some problem, and the group includes many scientists who are very knowledgeable about standard statistical methods. Paper 6C2: Statistics education at Russian agrarian universities: problems and prospectsGalina Kamyshova Saratov State Agrarian University, RussiaWe analyze Federal State Educational Standards and curricula of experts’ training for agrarian sphere of Russian economy from the point of view of statistical disciplines’ teaching. Problems are revealed and some ways of their decision which will promote improvement of quality of experts’ training are offered. Paper 6C3: Statistics for postgraduates and researchers in other disciplines: case studies and lessons learnedJohn Harraway University of Otago, New ZealandPostgraduates and researchers in many disciplines use advanced statistics procedures. Statistics backgrounds often extend to at most an introductory course on statistical methods. Effective ways of providing training in these advanced procedures must be found. Emphasizing content, prerequisites and target groups, a summary of specialized courses offered at this level over the last two years and advertised internationally is presented. Then local four day intensive workshops on advanced topics for ecologists are described. These workshops draw on research contexts familiar to participants and use appropriate software. Menu driven packages or self written programs may be used. Participants in the workshops can bring their own data or data are chosen from their discipline. The teacher is introduced to the researchers which may result in future collaboration. Student evaluations of the workshops are reported leading to recommendations for further training. Paper 6C4: Communicating the value of statistical thinking in researchGillian Lancaster Lancaster University, United KingdomIn recent years much attention has been given to statistical literacy and to stimulating interest in statistical thinking and statistical reasoning. Statistical thinking is important for understanding the world around us and is in all but recognition the cornerstone of evidence-based research. Statistical reasoning is motivated by real-world problems, which in turn promote the use and development of statistical methods of enquiry. It is well known that many researchers from other disciplines find statistics challenging, and some do not appreciate the relevance of statistical enquiry. This paper gives examples of some of the teaching strategies that have been applied within the Lancaster Postgraduate Statistics Centre when teaching students from a range of disciplines and short courses for social and health scientists. Issues concerning course structure and methods of teaching will be discussed and several experimental innovations highlighted. Paper Session 6D: Medical statistics6D1: Continuous variables: to categorise or to model?Willi Sauerbrei University of Freiburg, GermanyPatrick Royston Medical Research Council, United Kingdom Continuous variables are often encountered in life. We measure age, blood pressure and many other things. In medicine, such measurements are often used to assess risk or prognosis or to select a therapy. However, the question of how best to use information from continuous variables is relevant in many areas. To relate an outcome variable to a single continuous variable, a suitable regression model is required. A simple and popular approach is to assume a linear effect, but the linearity assumption may be violated. Alternatively, researchers typically apply cutpoints to categorize the variable, implying regression models with step functions. We illustrate problems caused by categorization and introduce fractional polynomials (FP) as a useful extension of polynomial regression. Investigating the effect of age as a prognostic factor for breast cancer, we show how conclusions depend critically on how the continuous variable is analyzed. Paper 6D2: Control in clinical trialsStephen Senn University of Glasgow, United KingdomAmongst the many types of medical scientific investigation that are possible, the randomized double-blind controlled clinical trial has a very high reputation. Without control there can be no randomization and without randomization no convincing blinding, It seems, therefore, that control is the key feature of such trials. Yet the way in which such trials are analyzed, including the way in which they are presented, shows that many trialists do not understand the value of what they have done. I illustrate the problem with various examples. One possible reason that trialists may underestimate the value of concurrent control is that they do not understand what a powerful source of bias regression to the mean constitutes. I consider how physicians can be taught to understand this difficult phenomenon. I also present a striking example of differences between populations, which can be used to teach care in using control information appropriately. Paper 6D3: The applied statistical scientist in a high-profile academic environmentGeert Molenberghs Hasselt University, BelgiumWe describe the importance of education in the high-profile, interdisciplinary academic environment facing the applied statistical scientist. Specific attention will be given to the context of biometry and medical statistics. The role can be played by international professional organizations is discussed. Paper Session 6E: Statistical applications in the workplace6E1: Training and conducting economic evaluation in public healthPhaedra Corso University of Georgia, United StatesEconomic evaluation (EE) refers to applied analytic methods used to identify, measure, value, and compare the costs and consequences of prevention and treatment strategies. Economic evaluation provides important information to assist policy makers who are faced with making funding decisions with scarce public health resources. However, the application and understanding of the statistical tools used to conduct economic evaluation has been limited in the context of evaluation training. This paper provides an introduction to three methods for conducting economic evaluation of public health programs: cost-effectiveness analysis, cost-utility analysis, and benefit-cost analysis. Each method is discussed in turn, with a special emphasis on the statistical tools used in each, and the uses and misuses of the methods in the policy arena. Paper 6E2: Issues in training physicians in the uses of statistics: what do they think they need to know?Jennifer Freeman University of Sheffield, United KingdomJim Crossley University of Sheffield, United Kingdom Fergus Hamilton University of Sheffield, United Kingdom Claire Kok Shun University of Sheffield, United Kingdom Molebedi Segwagwe University of Sheffield, United Kingdom Philip Sedgwick St George’s University of London, United Kingdom Key to any educational development work is the question of what learners need to know. In terms of teaching statistics to non-statisticians this may be examined from the perspective of both educators and students. It is easy for statistical educators to feel they know what students need to know, but it is also clear that when teaching non-statisticians, these non-specialists, who will be expected to use statistics after graduation, will have a view on what they need to know. We surveyed both current medical students and practicing doctors about what they would like to know and how it should be delivered. This informed a new curriculum covering the topics found to be most important and lead to the development of a new mode of delivery, based around the problembased learning model. The understanding produced provides valuable knowledge for both medical education and other disciplines where understanding statistics is essential. Paper 6E3: Helping make government policy analysts statistically literateSharleen Forbes Statistics New Zealand, New ZealandPaul Bucknall Statistics New Zealand, New Zealand Nathaniel Pihama Statistics New Zealand, New Zealand Evidence-based decision making increased the demand for government policy makers to have basic numeracy and statistics skills. Statistics New Zealand’s response was to create a Certificate in Official Statistics specifically for policy analysts that aims to give them the skills to critically evaluate statistical releases, research reports and published policy or media documents for their appropriateness and quality (of data, survey design, analysis and conclusions made) for some given policy question (e.g. how to reduce unemployment). Both statistical and non-statistical aspects are covered. Four of the units in the Certificate have learning done in traditional classrooms using small group workshops. Both the learning and competency based assessment are focused around real case study publications. In the final (major) unit students undertake and present an analytical report relevant to their own workplace. This paper reports on evaluations of the success of the certificate using the first three cohorts of students. Paper Session 6F: Service learning and statistics: integrating statistics education into the workplace6F1: Combining on- and off-campus service-learning in a statistics methods courseDebra Hydorn Univ of Mary Washington, United StatesMany models for implementing service-learning are possible, but the issue of meeting the needs of a community service organization within an academic time frame must be considered regardless of which model is chosen. Organizational deadlines that occur too early or too late in a course can minimize the contribution of the service project to students’ learning. Expanding the definition of ‘community’ to include the academic community allows the possibility of working with student, faculty and administrative ‘clients’ across campus whose data analysis needs may agree better with the time frame available during a single course. Service-learning in this expanded definition of community can meet the necessary criteria for academic service learning: relevant and meaningful service within the community, enhanced academic learning, and purposeful civic learning. Paper 6F2: Service-learning for statistics students in the global health arenaJulie Legler St. Olaf College, United StatesStudents work with World Health Organization (WHO) researchers contributing to their biostatistics knowledge and WHO’s mission. Two projects involving undergraduates and WHO collaborators are described briefly. A primary charge of WHO is to evaluate the global burden of disease (GBD). Students on this first project explored theoretical scenarios to learn about potential variability in GBD estimates. Students on the second project analyzed a massive dataset from the World Health Survey to describe global trends in tobacco use. Each project explored important issues that may not have been considered without the extra assistance of the students. Every other year for the past six years, this program has demonstrated that undergraduates can contribute to the analysis of global health data. Paper 6F3: STATCOM @ UHASSELT: yet another benefit for all partiesHerbert Thijs Hasselt University, BelgiumMany statistical research centers have developed a strong activity with respect to scientific collaborations with external parties. Specifically within universities besides statistical research and education the interaction with either pharmaceutical or governmental third parties has generated an enormous network with industrial partners and other academic institutions. Within Hasselt University, the Center for Statistics organizes an international Master of Statistics. Several efforts have been made to incorporate collaborating with non-statisticians into our program. This manuscript therefore focuses on (1) how a statistical training program should also contain training in consultancy skills, as well as on (2) how consultants in a university framework can play a role in training researchers. Furthermore this paper will sketch a possible optimal combination of all above mentioned aspects together with the StatCom society in order to ‘involve’ students in consultancy projects and teach them how to participate in an interdisciplinary working environment. Paper 6F4: Promoting opportunities for statistics service-learning at a large urban universityGolden Jackson Ohio State University, United StatesService-learning connects course content with service projects that address real world problems and provides a focus for student interest in learning to use and interpret statistics. Service-learning also supports leadership development, problem analysis and solving, evaluation and useful outcomes for community partners. Working with community to plan and coordinate student work requires additional resources, leading to a question whether the promise of enhanced student learning and benefit of working with community produces a sufficient return, especially in a research university, where emphasis is placed on research productivity and ability to garner external support for research. An example from a large research university reveals four elements for development and institutionalization of service-learning: (1) leadership/support from University Administration, (2) faculty development, (3) common definition and visibility, and (4) infrastructure. Suggestions for the institutionalization of service-learning in statistics will be emphasized. Paper Session 6G: Preparing for the world of work: lessons for statistics education from beyond the field6G1: Lessons from medicine for training professional statisticiansAlison Gibbs University of Toronto, CanadaTim Guimond University of Toronto, Canada In addition to knowledge of a constantly expanding collection of statistical methods, statistical consultants must have excellent communications skills and the ability to empathize with their clients’ practical problems. These characteristics that are expected of professional statisticians by their clients are also expected of physicians by their patients. In the early 1900’s, medical education in North America was restructured and standardized. All medical students followed a program of pre-clinical courses, succeeded by hands-on clinical studies. A recent trend in medical education has been to restructure education around key physician competencies. We explore a model of medical education and consider how it applies to a graduate course in statistical consulting. In particular, we discuss how we expose students to a wide variety of clients and problems, and describe the use of consulting rounds, modeled after medical rounds. Paper 6G2: Applying a model of professional learning to case studies in statistics educationAnna Reid University of Sydney, AustraliaPeter Petocz Macquarie University, Australia Lars Owe Dahlgren Linköping University, Sweden Madeleine Abrandt Dahlgren Linköping University, Sweden University students are generally aware that they are preparing for professional working life, and they use this as one focus for their learning. We have carried out interviews with over 500 students, in separate projects in Australia and Sweden, in a variety of discipline areas (including statistics). Based on these, we have developed a model of professional learning that incorporates students’ views of learning for their profession, the nature of knowledge in their discipline, and their development of professional identity. We describe this model and illustrate it by presenting case studies of two students undertaking a degree in mathematical sciences, majoring in statistics. The model gives us a framework for investigating pedagogical approaches. It allows us to examine how knowledge is presented in our courses, to analyse students’ class and assessment responses and to better support students’ professional formation. Paper 6G3: Statistical training for non-statistical staff at the Office for National StatisticsGemma Hamilton Office for National Statistics, United KingdomWithin the Office for National Statistics there is a high percentage of operational delivery staff involved in the day to day production of statistics who, although competent within their specific roles, have limited understanding of statistical concepts or the wider statistical process. The challenge facing ONS’s statistical training unit has been to develop a course that, without placing too much emphasis on technical methods, will enable staff to take a more statistical approach to their work, which should result in greater job satisfaction. A ‘statistics for non-statistical staff’ course was developed and this paper will set out the challenges faced when designing this course and will explain how these challenges were overcome. Paper 6G4: The importance of teaching statistics in a professional contextKay Lipson Swinburne University of Technology, AustraliaGlenda Francis Swinburne University of Technology, Australia This study investigates how the attitudes of marketing and psychology students differ and whether the differences are inherent from the start or develop as they progress through their degree. While statistics is equally useful for careers in marketing and psychology, third year marketing students are shown to have much less positive attitudes towards statistics than their psychology counterparts. It is suggested that this difference may be as a result of differences in the two course structures and that embedding statistics more fully into specific discipline areas improves student’ attitudes and helps to prepare them better for the workplace. Paper |