John Marriott - Academia.edu (original) (raw)

Papers by John Marriott

In this session • an approach to engaging students who experience introductory statistics as a se... more In this session • an approach to engaging students who experience introductory statistics as a service course will be presented. • a problem solving approach is employed • the students collect their own data via a real-time classroom survey tool. • exemplar resources will be introduced Summary i. Introductory statistics modules

In these two papers we review teaching statistics, statistical education and related outreach act... more In these two papers we review teaching statistics, statistical education and related outreach activities by a range of providers since the beginning of the last century. We discuss the extent and form of relevant published papers, books and conferences and give examples of where these have influenced teaching practice. In this part we show that by learning the lessons that (i) statistical and mathematical thinking are different and (ii) the goal of statistics of getting information from real data in context are both prerequisites for improving statistical literacy in people of all ages.

Journal of the Royal Statistical Society: Series B (Statistical Methodology), 2009

Structured additive regression models are perhaps the most commonly used class of models in stati... more Structured additive regression models are perhaps the most commonly used class of models in statistical applications. It includes, among others, (generalized) linear models, (generalized) additive models, smoothing spline models, state space models, semiparametric regression, spatial and spatiotemporal models, log-Gaussian Cox processes and geostatistical and geoadditive models. We consider approximate Bayesian inference in a popular subset of structured additive regression models, latent Gaussian models, where the latent field is Gaussian, controlled by a few hyperparameters and with non-Gaussian response variables. The posterior marginals are not available in closed form owing to the non-Gaussian response variables. For such models, Markov chain Monte Carlo methods can be implemented, but they are not without problems, in terms of both convergence and computational time. In some practical applications, the extent of these problems is such that Markov chain Monte Carlo sampling is simply not an appropriate tool for routine analysis. We show that, by using an integrated nested Laplace approximation and its simplified version, we can directly compute very accurate approximations to the posterior marginals. The main benefit of these approximations is computational: where Markov chain Monte Carlo algorithms need hours or days to run, our approximations provide more precise estimates in seconds or minutes. Another advantage with our approach is its generality, which makes it possible to perform Bayesian analysis in an automatic, streamlined way, and to compute model comparison criteria and various predictive measures so that models can be compared and the model under study can be challenged.

In C. Reading (Ed.), Data and context in statistics education: Towards an evidence-based society.... more In C. Reading (Ed.), Data and context in statistics education: Towards an evidence-based society. Proceedings of the Eighth International Conference on Teaching Statistics (ICOTS8, July, 2010), Ljubljana, Slovenia. Voorburg, The Netherlands: International Statistical Institute. www.stat.auckland.ac.nz/~iase/publications.php [© 2010 ISI/IASE] ONE HUNDRED YEARS OF PROGRESS – TEACHING STATISTICS 1910 – 2010: WHAT HAVE WE LEARNED? PART II: PROBLEM SOLVING, PEDAGOGY AND EMPLOYEES

Journal of Statistics Education

In this paper we report the results from a major UK government-funded project, started in 2005, t... more In this paper we report the results from a major UK government-funded project, started in 2005, to review statistics and handling data within the school mathematics curriculum for students up to age 16. As a result of a survey of teachers we developed new teaching materials that explicitly use a problem-solving approach for the teaching and learning of statistics through real contexts. We also report the development of a corresponding assessment regime and how this works in the classroom. Controversially, in September 2006 the UK government announced that coursework was to be dropped for mathematics exams sat by 16-year-olds. A consequence of this decision is that areas of the curriculum previously only assessed via this method will no longer be assessed. These include the stages of design, collection of data, analysis and reporting which are essential components of a statistical investigation. The mechanism outlined here could provide some new and useful ways of coupling new teaching methods with learning and doing assessment-in short, they could go some way towards making up for the educational loss of not doing coursework. Also, our findings have implications for teaching, learning and assessing statistics for students of the subject at all ages.

Abstract : For a general stationary and. invertible ARMA (p,q) process, we show how to carry out ... more Abstract : For a general stationary and. invertible ARMA (p,q) process, we show how to carry out a fully Bayesian analysis. Our approach is through the use of sampling based methods involving three novel aspects. First the constraints on the parameter space arising from the stationary and invertibility conditions are handled by a convenient reparametrization to all of Euclidean (p+q)-space. Second, required sampling is facilitated by the introduction of latent variables which, though increasing the dimensionality of the problem, greatly simplifies the evaluation of the likelihood. Third, the particular sampling based approach used is a Markov chain Monte Carlo method which is a hybrid of the Gibbs sampler and the Metropolis algorithm. We also briefly show how straightforwardly the sampling based approach accommodates missing observations, outlier detection, prediction and model determination. Finally we illustrate the approach with two examples.

Statstutor is an on-line statistics learning and teaching resource, being developed by the sigma ... more Statstutor is an on-line statistics learning and teaching resource, being developed by the sigma Centre for Excellence in Teaching and Learning (CETL) at Loughborough and Coventry Universities, in collaboration with the Royal Statistical Society Centre for Statistics Education. It differs from existing on-line resources, focusing on the practical application of statistics using both a topic based approach for learning, as well as opportunities for learning through a problem solving approach using case studies. It combines video, paper-based and other electronic media into one environment. It is a pilot project which will form part of the mathcentre family of on-line mathematics resources (www.mathcentre.ac.uk). This paper describes the project's development, and lists the resources that are available. We hope the paper will motivate discussion of our approach and the content of statstutor and, in turn, help to inform us about directions we should take for post-pilot developments of the resource.

A Bayesian approach to the analysis of AR time series models, which permits the usual stationarit... more A Bayesian approach to the analysis of AR time series models, which permits the usual stationarity assumptions to be relaxed, is presented. Example analyses of published time series data are included, which illustrate some potential dangers in the application of the usual criteria for di erencing.

An International Perspective, 2010

MSOR Connections, 2013

A report on a workshop held at the University of Bristol on the teaching of statistics.

MSOR Connections, 2010

&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;quot;The authors briefly describe an approach fo... more &amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;quot;The authors briefly describe an approach for engaging students in introductory statistics modules with data which the students generate themselves which is returned to them in real-time. We extend an invitation to all to make use of our system and contribute ideas for broadening its use. It has been clear to the authors for some time that more interesting sources of real data for teaching and learning introductory statistics (graphs, descriptive statistics, regression etc) are desirable for motivating students in the subject. Having ‘interesting’ data is key and one approach is to collect data which can be used across disciplines, and which is of interest to all students. Two topics that are often both interesting and useful to them, especially in the first year at university, are accommodation and neighbourhood safety/crime. In any case these are topics students are likely to discuss and can be investigated through the ‘Statistical Problem Solving Approach’ cycle with which many should already be familiar with from studying GCSE mathematics. Our survey tool project extends the LimeSurvey software [www.limesurvey.org] by adding the ability to download participants’ responses to them, in a controlled manner in real time.&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;quot;

MSOR Connections, 2011

Here we report upon an event held by the Network to which, for the first time, undergraduate math... more Here we report upon an event held by the Network to which, for the first time, undergraduate mathematics students from universities in the Midlands were invited. The aim was to explore their views on their undergraduate experience, teaching and learning of mathematics and statistics, and peer support. The report is in two parts: the first deals with aspects of the teaching and learning of mathematics, and the second with statistics. The mathematical elements formed the basis of a presentation at the British Mathematics Colloquium held at the University of Leicester in April 2011.

Teaching Statistics, 1997

Scandinavian Journal of Statistics, 2001

We consider the problem of selecting a regression model from a large class of possible models in ... more We consider the problem of selecting a regression model from a large class of possible models in the case where no true model is believed to exist. In practice few statisticians, or scientists who employ statistical methods, believe that a``true'' model exists, but nonetheless they seek to select a model as a proxy from which they want to predict. Unlike much of the recent work in this area we address this problem explicitly. We develop Bayesian predictive model selection techniques when proper conjugate priors are used and obtain an easily computed expression for the model selection criterion. We also derive expressions for updating the value of the statistic when a predictor is dropped from the model and apply this approach to a large well-known data set.