TOPIC 3: Statistics education at the post-secondary level
The subject statistics is a multidimensional concept, and involves statistical science, theory, methodology as well as practical statistical skills applied to different disciplines. The topic “teaching statistics at the post secondary level” will have different meanings dependent on the discipline, the aim of the education, the statistical method, the theoretical and applied level of statistics. Hence, this topic covers a wide variety of students, statistical contents and theoretical levels.
Inter-disciplinary cooperation between researchers, teachers and users will often be a prerequisite of a successful education. The aim of this topic at ICOTS-7 is to provide multi-disciplinary and multi-conceptual examples of teaching statistics at the post secondary level. Sessions for different applications and levels, for future statisticians and for non-statistical students are planned. Educational techniques such as case studies, projects, problem-based learning and learning by computers will be presented. The teaching of non-parametric methods, multivariate methods, and models will also be on the schedule. The goal is to engage statistics educators - those affiliated in statistics departments as well as in other departments - in a discussion concerning the statistics curriculum and the teaching techniques.
SESSION 3A: Statistics as a service subject in courses
3A1: LETTING STUDENTS UNDERSTAND WHY STATISTICS IS WORTH STUDYING
Eric R. Sowey
The University of NSW, Australia
Whether they are studying statistics as a disciplinary major or through service courses, students will be more motivated towards what they are learning, and will retain a richer recall of it, if they feel they are doing something worthwhile. I have previously argued that three elements in teaching are salient in giving a sense of worthwhileness: showing that statistics is interesting, useful, and substantial. The first two of these elements are already well discussed. But letting statistics be seen as a substantial discipline, in the sense of being resilient to challenging questioning prompted by students’ own curiosity, has not been previously addressed in the statistics education literature. Here I show the kinds of challenging questions which serve this goal. The answers given need not be overly-detailed: what matters is that they satisfy students’ curiosity. In this way they strengthen the students’ sense that statistics is worth the effort of study.
3A2: FROM STATISTICIAN TO OPHTHALMOLOGIST VIA NURSING, OCCUPATIONAL THERAPY AND PHYSIOTHERAPY
Angus K. McFadyen
Glasgow Caledonian University, Scotland
Bridging the gap between the statistical specialist and students from clinical areas must be, at least initially, the primary responsibility of the statistician. If the relevance of statistics is to be appreciated by undergraduates from other academic areas then the statistics must be taught in a relevant context using knowledge gained by the statistician from collaborative research. This paper outlines the experiences of one academic statistician who has taken a long and winding journey through various clinical areas. The journey has been enjoyable, creating close links with academic colleagues and seeing the appreciation shown by students who realise that the dreaded “statistician” has a working knowledge of their area of practice which improves credibility and enhances rapport. Ways of encouraging collaborative work with academics from other areas will be discussed as will the nerve-wracking but rewarding experiences of presenting at clinical conferences.
3A3: TEACHING STATISTICAL CONCEPTS, FUNDAMENTALS AND MODELLING
Timothy E. O’Brien
Loyola University Chicago, United States
Over the past several years, statistical educators have been involved in teaching “service courses”, where our audiences often come from the social sciences and humanities. As such, it is unlikely that these students will ever become “producers” of statistics – rather, they will be “consumers” or “users” of statistics. Thus, the courses we teach these students should reflect this fact – instead of focusing on calculations and derivations, our courses are becoming much more conceptual. This paper highlights some of the author's experiences with the transition from “number crunching” to “conceptual” basic statistics courses, focusing on in-class activities, student projects, writing assignments, and using computer packages.
3A4: STATISTICS EDUCATION IN THE NETHERLANDS AND FLANDERS: AN OUTLINE OF INTRODUCTORY COURSES AT UNIVERSITIES AND COLLEGES
Pieternel Verhoeven
Roosevelt Academy, The Netherlands
This paper summarizes the results of a study into ways in which introductory Statistics courses are taught at universities and honors colleges in The Netherlands. In interviews, course coordinators described the teaching methods used, the student population, they assessed the developments in the statistics curriculum and considered future developments, focusing on Statistics for social science majors in their first year. These interviews form a part of a major study into educational and student determinants of course outcomes with respect to statistics courses in The Netherlands and Flanders. Regarding the first part of this study i.e. educational determinants, the results show a diversity of teaching methods, group sizes (from 12 students to more than 500) and assessment tools. Statistical course content is, in general, equal across social science departments.
SESSION 3B: Teaching robust methods
3B1: THE HISTORICAL DEVELOPMENT OF ROBUST STATISTICS
Elvezio Ronchetti
University of Geneva, Switzerland
We focus on the historical development of robust statistics by highlighting its contributions to the general development of statistical theory and applications. The basic robustness concepts and tools should be included in a natural way both in undergraduate and graduate statistics and econometrics curricula. We argue that this is more effective than treating robust statistics as a special (advanced) topic and we illustrate this point by means of an example drawn from economics and finance.
3B2: INTERACTIVE LEARNING TOOLS
David S. Zamar
Simon Fraser University, Canada
Ruben H. Zamar
University of British Columbia, Canada
Although the intuitive idea of robustness is simple and appealing some key robustness concepts and measures are mathematically involved and hence difficult to grasp by students. We show how interactive graphic tools can be used to motivate and demonstrate the use of robust methods. We also show how interactive graphical tools can also be used to help understand and “visualize” main robustness concepts such as influence function and breakdown point. We developed interactive tools for simple linear regression and multivariate location and covariance matrix. Due to space limitations we focus on regression and only briefly comment on the multivariate location and covariance matrix.
3B3: THE TEACHING OF ROBUST STATISTICS FOR REGRESSION
Víctor J. Yohai
Universidad de Buenos Aires and CONICET, Argentina
We present some ideas on how to teach Robust Regression. We motivate the importance of robust estimation using a real data set and briefly discuss why diagnostic procedures based on the least squares estimates do not guarantee the detection of outlier observations. We also introduced regression M-estimates and discuss why they require an estimate of the error scale. Finally we introduce the class of S-estimates to obtain a robust estimate of the error scale.
SESSION 3C: Mentoring of graduate students
3C1: MENTORING STATISTICIANS IN FORMAL DEGREE PROGRAMS: THE MASTERS IN BIOSTATISTICS AT THE UNIVERSIDAD DE CHILE (1983-2005): ENTERING THE 21ST CENTURY
Claudio Silva, Irene Schiattino, Francisco Cumsille, Gabriel Cavada, Rosa Montaño
Universidad de Chile, Chile
This paper describes the process of renewal of the Masters in Biostatistics program according to the technologic, occupational and academic new realities present at the beginning of this century in a developing country. In 1983 the University of Chile created an academic program of Masters in Biostatistics oriented toward preparing highly qualified personnel in Biostatistics to perform university teaching functions, as well as research consulting in the biological and medical sciences. In 2002, the Masters in Biostatistics program underwent a re-engineering process that is reflected in changes in the lecture series, including revision of course content, requiring students to work real-life problems from the beginning of their degree program, introduction of computer simulation as a teaching and research tool, and discussion of both parametric and non-parametric inference procedures.
3C2: AS GOOD AS IT GETS: CHALLENGES IN TEACHING APPLIED STATISTICS
Julio M. Singer
Universidade de São Paulo, Brazil
The role of mentoring (i.e., direct supervision) in bridging the gap between statistical methodology taught in the academic environment and its actual application to practical problems is considered. The particularities of the Brazilian case are highlighted and the thirty-year experience acquired by the Statistical Laboratory at the University of São Paulo is described. Topics ranging from interaction with clients, problem specification, statistical modelling, report writing to multimedia presentation of results are covered. The main difficulties and positive aspects of the program are identified. Guidelines to improve it are also discussed.
3C3: MENTORING GRADUATE STUDENTS IN APPRENTICESHIP POSITIONS AS RESEARCH ASSISTANTS: THE EXPERIENCE AT THE UNIVERSITY OF NORTH CAROLINA AT CHAPEL HILL
Shrikant I. Bangdiwala
The University of North Carolina at Chapel Hill, United States
Graduate training programs in applied statistics are the formal method to provide future professionals with the necessary theoretical and methodological tools for a successful career as a statistician. Mentoring is considered as another form of education at the post secondary level, and should complement the formal coursework with individualized hands-on experiences with real world problems in addition to providing students with career counseling. This manuscript describes the mentoring process at the Department of Biostatistics of the School of Public Health of the University of North Carolina at Chapel Hill.
3C4: MENTORING IN THE FINAL PROJECT OF A BACHELOR IN STATISTICS
María Teresa Blaconá
Universidad Nacional de Rosario, Argentina
In many degrees it is requested that students should do a final project under the guidance of a professor at the final stage of the bachelor’s program. This is the case of the bachelor program in statistics (Licenciatura en Estadística) of the School of Statistics of the National University of Rosario, Argentina. In fact, the lecturer in charge of guiding the students is not only the advisor but also the mentor, because in many cases the project consists of solving not only an academic problem but also a real-life problem with up-to-date methodologies. Therefore, the professor should help the student deal with situations that would arise in his or her professional life.
SESSION 3D: Teaching heterogeneous groups
3D1: TEACHING STATISTICS AND RESEARCH METHODS TO HETEROGENEOUS GROUPS: THE WESTMINSTER EXPERIENCE
Alan Porter, Tina Cartwright, Rosemary Snelgar
University of Westminster, United Kingdom
Teachers of statistics are often faced with the task of teaching their subject to heterogeneous groups of students. At the University of Westminster we are faced with groups of students who are all studying psychology but have very different academic and social backgrounds. To teach statistics and research methods successfully to such heterogeneous groups we have developed teaching and learning strategies to enhance student experience of the course. These strategies encompass curriculum design and student support and include interventions using blended learning, study groups, reflective learning journals and the management of statistics anxiety.
3D2: TEACHING ACADEMICALLY DIVERSE GROUPS
Mark L. Berenson
Montclair State University, United States
Imagine entering the classroom the very first day of the term and realizing how academically diverse a particular student body is. You would likely be caught unprepared for this, as I was this past September, and a first thought might be to quickly evaluate how much the prepared syllabus must be reengineered. Practitioners of the “Management-by-Process” philosophy attributed to Deming would immediately opine that understanding and managing variation is fundamental and then view a heterogeneous classroom as both an opportunity and a challenge to explore and develop pedagogy that enhances overall student performance. Others, perhaps more pragmatic or perhaps less risk assertive in the classroom, might immediately get a sinking feeling, and then try to contemplate how the course might be salvaged. This paper will explore various possibilities for making the best out of a situation in which a difficult constraint has been imposed.
3D3: CREATING STATISTICAL RESOURCES FROM REAL DATASETS – THE STARS PROJECT
Penelope Bidgood
Kingston University, United Kingdom
The aims of the STARS (Statistical Resources from Real Datasets) project are to make available real datasets and associated scenarios applicable to a range of disciplines and to develop learning and assessment materials to accompany these datasets for use with various packages. The project team, based in 4 universities in England, have developed worksheets in Psychology, Health and Business, using mainly Excel, MINITAB, SPSS in both pdf format and Word. The worksheets are designed to be used in introductory statistics courses in service teaching and cater for a range of student abilities, backgrounds and needs. Further, resources for individualised datasets and assignments with solutions to be generated from the datasets have been produced. The materials developed and the concepts behind them have a far greater potential and use throughout the statistics teaching community.
SESSION 3E: Multivariate Statistics
3E1: ANALYZING DNA MICROARRAYS WITH UNDERGRADUATE STATISTICIANS
Johanna Hardin, Laura Hoopes, Ryan Murphy
Pomona College, United States
With advances in technology, biologists have been saddled with high dimensional data that need modern statistical methodology for analysis. DNA microarrays are able to simultaneously measure thousands of genes (and the activity of those genes) in a single sample. Biologists use microarrays to trace connections between pathways or to identify all genes that respond to a signal. The statistical tools we usually teach our undergraduates are inadequate for analyzing thousands of measurements on tens of samples. The project materials include readings on microarrays as well as computer lab activities. The topics covered include image analysis, filtering and normalization techniques, and statistical methods. The course materials are designed for someone with little or no statistical background, but due to the novel concepts covered, they could easily be adjusted to accommodate students with practically any background.
3E2: SUCCESSFUL STRATEGIES FOR TEACHING MULTIVARIATE STATISTICS
Joseph F. Hair, Jr.
Kennesaw State University, United States
Perhaps among the most challenging course I have taught is multivariate data analysis. The topic not only is difficult, but most students are ill prepared to take the course. In such situations, one must first overcome the initial fear factor of students and then get them engaged enough to keep them motivated. To accomplish this, I utilize the following teaching strategies: (1) represent yourself as a real person to your students; (2) use lectures sparingly; (3) encourage active learning through discussion and hands-on exercises; (4) include lots of examples; and (5) constantly reinforcement previous concepts. These represent the strategies I have found to be the most successful.
3E3: BASIC MULTIVARIATE THEMES AND METHODS
Lisa L. Harlow
University of Rhode Island, United States
Much of science is concerned with finding latent order in a seemingly complex array of variables. Multivariate methods help uncover this order, revealing a meaningful pattern of relationships. To help illuminate several multivariate methods, a number of questions and themes are presented, asking: How is it similar to other methods? When to use? How much multiplicity is encompassed? What is the model? How are variance, covariance, ratios and linear combinations involved? How to assess at a macro- and micro-level? and How to consider an application? Multivariate methods are described as an extension of univariate methods. Three basic models are presented: Correlational (e.g., canonical correlation, factor analysis, structural equation modeling), Prediction (e.g., multiple regression, discriminant function analysis, logistic regression), and Group Difference (e.g., ANCOVA, MANOVA), with brief applications.
3E4: LUDOVIC LEBART’S APPROACH: A WAY FOR TEACHING APPLIED MULTIVARIATE STATISTICS IN GRADUATE COURSES WITH A HETEROGENEOUS AUDIENCE
Hebe Goldenhersch
Universidad Nacional de Córdoba, Argentina
First of all we describe different kinds of audiences in a graduate multivariate course. We talk about how to start a Multivariate Statistics Course. We show a survey as an example of Multivariate Data. Factorial Methods and Adjustment Criteria are discussed, within a general analysis. The importance of the “illustrative variables” according to Lebart´s approach are discussed, and finally we discuss the complementary applications of factorial methods with cluster methods to analyze the surveys when the data are quantitative and when they are qualitative.
SESSION 3F: Teaching nonparametric methods
3F1: SMOOTHING TECHNIQUES IN SPATIAL STATISTICS
Wenceslao González Manteiga, Manuel Febrero Bande
Universidade de Santiago de Compostela, Spain
The variogram is one of the most important tools in the assessment of spatial variability of a spatial statistical model. Estimation and testing on this function is a crucial problem in random processes inference, with several applications in a broad spectrum of areas such as geostatistics, hydrology, atmospheric sciences, etc. We show in this work how a generalized family of variogram estimators can be built based on the classical ideas of smoothing techniques in nonparametric regression. Some examples will be given in order to compare the performance of Nadaraya-Watson and Local Linear estimators with the empirical variogram. The proper choice of the bandwidth for these methods will be discussed. Some applications to atmospheric and/or environmental data will also be provided. Finally, some extensions to the space-time setting will be considered. Special emphasis will be placed on teaching aspects in this paper.
3F2: SMOOTHING SEQUENCES OF DATA BY EXTREME SELECTORS
Tertius de Wet, Willie Conradie
University of Stellenbosch, South Africa
Non-linear smoothers based on the extreme selectors have been developed as a class with very powerful properties and ideally suited for application to data having impulsive noise, the type of data that often occur in the engineering and financial fields. Some of their properties make them ideally suited as a basis for teaching students about the art and science of data smoothing. These include inter alia their treatment of blockpulses of particular lengths as either signal or noise, its idempotency properties, which powerfully and visually demonstrate the mathematical concept of idempotence (which is often difficult for students to grasp) and the way that they systematically, measurably and monotonically “peel off” variation until one has a sufficiently smooth result. In this paper we define and discuss members of this class of smoothers and illustrate how their properties make them attractive aids in teaching aspects of nonparametric smoothing as well as aspects of Extreme Value Theory. A Standard and Poor 500 financial data set will be used for illustration purposes.
3F3: TEACHING NON-PARAMETRIC STATISTICS TO STUDENTS IN HEALTH SCIENCES
Michael J. Campbell
University of Sheffield, United Kingdom
There are advantages and disadvantages to teaching non-parametric statistics. On the plus side is the fact that minimal assumptions have to be made, and many of the tests (such as the sign test) are intuitive. On the minus side are the difficulties in teaching statistics purely through significance tests and the difficulty of getting estimates and confidence intervals. My experience is that some disciplines, especially psychology, seem to teach almost exclusively non-parametric statistics, and I have been trying to steer them towards a more parametric approach, since, in my opinion, unless you can write a model, you don't really understand what to infer from a p-value.
SESSION 3G: Teaching consultancy skills to statisticians
3G1: TEACHING STUDENTS AND STAFF CONSULTANCY SKILLS
Edward D. Rothman
University of Michigan, United States
Consultants are well aware of the skills we must impart. These include: an appreciation for the role of the work one does in the context of the research effort; an understanding of what we can and cannot do as statisticians; a general appreciation for modeling and visualization; and an understanding of the nuclear elements of modeling tools. All of these skills contribute to one’s ability to communicate ideas to a client and, ultimately, to the reader of any findings. Changes in the toolkit for statisticians require that we continually learn. In this talk I describe a process that creates a continual improvement cycle for the mastery of these skills. By mastery, I mean that the student has demonstrated understanding through the application of a principle and taught what they have done to others, including non-professionals. The principles, based on modeling and visualization efforts, give rise to the nuclear elements that are the simple and communicable features of the tool.
3G2: ANALYSING DATA FROM A CLASS TAUGHT WITH CLICKERS
Enrique E. Alvarez
University of Connecticut, United States
During the fall 2005 the author taught a large introductory Statistics class at the University of Connecticut with the aid of a student personal response system (also known as “clickers"). Apart from its pedagogical value, the system has the secondary advantage of building vast day to day data on students’ performance. This article presents an analysis of the data gathered from the course, together with conclusions and suggestions on how to collect and organize data for further educational studies.
3G3: PRACTICE IMPROVES PRESENTATION
Flavia Jolliffe
University of Kent, United Kingdom
Practice in doing consultancy, and in teaching consultancy skills to statisticians, improves that teaching, regardless of the teaching method used. Teaching consultancy skills involves emphasis on communication skills, making statisticians aware of the range of problems they might meet, and showing them ways of dealing with non-standard problems. Examples based on actual consultancy sessions are discussed in this paper, and suggestions are made as to how to integrate these into a course intended to equip students for work as a practical statistician.
SESSION 3H: Statistics learning with cases/projects
3H1: TEACHING STATISTICS AND RESEARCH METHODS: AN INTEGRATED APPROACH
Hans van Buuren
Open Universiteit Nederland, The Netherlands
Traditional curricula in the social sciences result in students having statistical knowledge that is inert and consequently of low transferability. This is in part because these curricula separate mathematical and probabilistic content (present in statistics service courses) from the context in which the collection of observational and experimental data is designed (present in courses about research methods). This paper proposes a curriculum that removes this separation by merging the two domains into the research competency, in line with emergent pedagogical insights. This study describes the new curriculum and compares some preliminary learning outcomes of students following the proposed integrated competency-based curriculum with that of students following the traditional curriculum. The results suggest a higher level of understanding is achieved through the integrated approach.
3H2: ASSESSING AND EDUCATING PRESCHOOL TEACHERS ON PROBABILITY CONCEPTS IN THE CLASSROOM
Jenny Pange
University of Ioannina, Greece
This study refers to an experiment on teaching probabilities, conducted in Greece at preschools in Athens and Ioannina. The aim of this study was to assess teachers on how they introduced common statistical concepts to children throughout the academic year of 2004-2005. Moreover, this study presents a new didactical model on the way we tried to educate preschool teachers on how to introduce to preschool children probabilistic concepts that are not contained in the official national curriculum. The majority of the teachers agreed that the study was interesting and that their intentions were positive, but they lacked the ability and specification to include those concepts in their everyday class curriculum.
3H3: MAKING STATISTICS REAL: WORKING WITH STATISTICS NEW ZEALAND
Jennifer Brown
University of Canterbury, New Zealand
Richard Penny
Statistics New Zealand, New Zealand
Marco Reale
University of Canterbury, New Zealand
At the University of Canterbury the statistics group has developed a teaching program that has a high level of involvement by the official statistics agency, Statistics New Zealand. Statistics New Zealand’s involvement in the teaching programme includes participating in the teaching of courses, providing statistics examples for student assignments, industry placement of student and student financial support. Their involvement is more in-depth than these tangible contributions. By having Statistics New Zealand staff in our department, and joint research projects between the department and the statistics group at the University, students come in contact with "statistics in action." The students have regular interaction with practicing statisticians and have the opportunity to formally, or informally, discuss trends and developments in statistical methods.
SESSION 3I: Teaching Bayesian Statistics
3I1: TEACHING INDEPENDENCE AND EXCHANGEABILITY
Lisbeth K. Cordani
Instituto Mauá de Tecnologia, Brazil
Sergio Wechsler
Universidade de São Paulo, Brazil
Most of the statistical curricula, mainly that written at the elementary level, is based on the classical (frequentist) approach. The Bayesian school, even if originated in the 18th century, has only recently seen a strong development of its tools. This development, however, has not been seen in a basic level. The discipline, as well as the teachers, reflect the classical dominance, which reinforces the current paradigm. Although they have different starting points, both approaches, classical and Bayesian, have tools to analyze data, and we should offer the choice to the student. This article deals with two important concepts, one very useful from the classical point of view, which is the concept of independence, and the other related to the Bayesian thought, the concept of exchangeability. Definitions and simple examples are presented to relate both approaches, from an elementary point of view.
3I2: A BAYESIAN MATHEMATICAL STATISTICS PRIMER
Jose M. Bernardo
University of Valencia, Spain
Bayesian Statistics is typically taught, if at all, after a prior exposure to frequentist statistics. It is argued that it may be appropriate to reverse this procedure. Indeed, the emergence of powerful
objective Bayesian methods (where the result, as in frequentist statistics, only depends on the assumed model and the observed data), provides a new unifying perspective on most established methods, and may be used in situations (e.g., hierarchical structures) where frequentist methods cannot. On the other hand, frequentist procedures provide mechanisms to evaluate and calibrate any procedure. Hence, it may be the right time to consider an integrated approach to mathematical statistics, where objective Bayesian methods are first used to provide the building elements, and frequentist methods are then used to provide the necessary evaluation.
3I3: UNPREDICTABILITY, PROBABILITY UPDATING AND THE THREE PRISONERS PARADOX
Rosangela H. Loschi
Universidade Federal de Minas Gerais, Brazil
Pilar L. Iglesias
Pontificia Universidad Católica, Chile
Sergio Wechsler
Universidade de São Paulo, Brazil
This paper discusses the Three Prisoners paradox in the light of three different procedures for the updating of probabilities – Bayesian conditioning, superconditioning and Jeffrey’s rule – as well as assuming the unpredictability of receipt of information by prisoner A. The formulation of the paradox in this temporal setting brings new insight to the problem and, on the other hand, the paradox is a good way to explain the different updating probability procedures and the difference between conditional probabilities and posterior distributions.
3I4: STANDARD STATISTICAL CONCEPTS: CAN THEY PRODUCE INCOHERENCE?
Carlos Alberto de Bragança Pereira
University of São Paulo, Brazil
The present article concerns statistical concepts that are usually presented in the statistical classroom. Examples are presented in a way such that simple applications of these concepts produce incoherent conclusions. The examples illustrate that:
iid random variables are in fact strongly dependent; conditional probabilities may depend on how the conditioning arguments were learned; confidence intervals may have the property of diminished precision when information is increasing; and significance tests may not reject impossible hypotheses.
SESSION 3J: Sampling for surveys
3J1: CAPTURE-RECAPTURE SAMPLING: A STUDENT PROJECT
Bruno C. de Sousa
Universidade do Minho, Portugal
Capture-Recapture Sampling is used when one is interested in estimating the size N of a certain population. With the help of EXCEL, we will illustrate how a student can estimate the size of a population using this technique. Different estimators for N will be considered, and the sampling distribution of the estimated values will be studied. Confidence intervals for N will be proposed and its interpretation will be presented. Some of the problems that we may encounter with this methodology will be briefly discussed, and a simulation in which the first captured individuals remain in a certain region of the population will be shown as an example of the effect that such behavior has on the estimated values.
3J2: THE ROLE OF SCALES IN TEACHING STATISTICS FOR SOCIAL SCIENCE STUDENTS
Peter Sedlmeier
Chemnitz University of Technology, Germany
Statistical analysis for social scientists very often means statistical analysis of some questionnaire data. The meaning of the numbers obtained in such analyses depends very much on the kind of scales used. In this paper it is shown that the meaning of numbers can also depend on how exactly the scales are constructed. First, some background information about how scales of the same type (e.g., interval scales) can considerably differ in meaning is given and then a series of study results with interval, ordinal, and nominal scales that demonstrate these differential effects are reported. It is argued that such results can easily be replicated in statistics classes. It is further argued that due to the preponderance of scales in social science research, statistics courses should put an emphasis on teaching the correct use and interpretation of scales. Here, demonstrations such as the ones described in this paper can play a helpful role.
3J3: CLUSTER OPTIMAL SAMPLE SIZE FOR DEMOGRAPHIC AND HEALTH SURVEYS
Alfredo Aliaga, Ruilin Ren
ORC Macro, Inc., United States
For practical purposes and simplicity, the sample design used in the Demographic and Health surveys is a two-stages clustered sample. In general, the sampling frame is a complete list of enumeration areas (EAs) created in a recent population census (around 100 households per EA). In a second stage, a prefixed number of households is selected from each EA. All household members (all women ages 15-49 in particular) are selected for interviewing. This paper looks at nearly optimum sample sizes and compares them with different situations. The results show that for most of the Demographic and Health surveys conducted, the sample size per cluster met the optimum size with a tolerable precision loss.
SESSION 3K: PANEL DISCUSSION
3K1: TEACHING STATISTICS IN CONTEXT
Johan Anton Van Buuren
~ Open Universiteit Nederland, The Netherlands
Neville Davies
~ Nottingham Trent University, United Kingdom
Peter Petocz
~ Macquarie University, Australia
John Harraway
~ University of Otago, New Zealand
There are two trade-offs within the world of data based scientific research. One is between a teacher-centred approach which transmits knowledge and a student-centred approach which may facilitate learning within a research design context. A second trade-off is between learning research skills and transmitting or teaching statistics. This session will take the form of a debate with audience participation over the teaching of statistics in context. It will involve four panelists, all experienced teachers of statistics, who will debate the two positions:
* The “Statistics Teacher” is an obsolete phenomenon and will be replaced by the “Research Facilitator”
* Learning research skills is more effective than learning statistics in a classical service course.
At stages during the presentations the audience will vote on the pro’s and con’s of the persuasiveness of the arguments in what promises to be a lively, interesting and controversial exchange.