Edinburgh RSS local group
Meetings 2013 - Archive

Monday 18th February 2013
Using statistics to study wildlife
ICMS, 15 South College Street, Edinburgh EH8 9AA
Meeting starts 6pm, tea and coffee from 5.30pm

This meeting will be followed by the Annual General Meeting of the Edinburgh local RSS group, which will include elections for committee members and office bearers. All are welcome to attend.

This event will include three speakers:

Kate Searle (Centre for Ecology and Hydrology)
Analysis of ecological data: "ecology isn't rocket science, it's harder"
Drawing inference in ecology is inherently difficult. Ecological processes and systems are multi-faceted and multi-scaled, such that an understanding of any individual part of the system requires recognition of drivers and constraints resulting from many interconnected processes. Moreover, states and variables within ecological systems are often not able to be measured directly, but must be inferred from surrogate observations. This means that ecological data typically confounds simple statistical approaches due to factors such as detectability, sampling error, overdispersion, zero-inflation, and unequal sampling effort over space and time. In this presentation I will give a brief overview of statistical problems commonly encountered by wildlife ecologists, and present examples of empirical approaches for overcoming, or at least mitigating, some of these issues.

Megan Towers (Scottish Natural Heritage)
Experiences as a statistician within Scottish Natural Heritage
This talk will include work to look at power analyses for monitoring otter activity within the Scottish Beaver Trial and revegation within the Monadhliath mountains. The former is an example of estimating the sample size necessary to detect a given effect size and the latter is an example of estimating what effect size it is reasonable to be able to detect given the current monitoring effort.

Ruth King (University of St. Andrews)
Incorporating individual time-varying covariates within the analysis of capture-recapture data
Capture-recapture studies are often undertaken in order to obtain data on wildlife populations, where individuals are repeatedly sampled over a period of time. Within such studies interest often lies in the relationship between survival probabilities and individual level covariates (such as weight, breeding status etc.). However, individuals are typically not observed at every capture event, leading to unobserved (or missing) time-varying individual covariates at these times. Recent approaches for dealing with such missing covariate information within capture-recapture data will be presented and discussed.

Tuesday 30th April 2013
Professor Susan McVie (University of Edinburgh)
The mysterious case of the disappearing crime in Scotland
ICMS, 15 South College Street, Edinburgh EH8 9AA
Meeting starts 6pm, tea and coffee from 5.30pm

Scotland, like many other countries worldwide, has seen an unprecedented drop in its crime rates over the last two decades. And yet crime as a topic of debate continues to generate concern, both politically and socially. Public perceptions of crime as a problem have not changed as dramatically as the crime rate, and they play an important role in driving criminal justice policy. So how do we make sense of the crime problem in Scotland?

This paper will focus on the what we know about patterns of youth crime in Scotland and, using longitudinal data from the Edinburgh Study of Youth Transitions and Crime, will explore offending from the perspective of young people themselves. A unique aspect of this study is that it allows us to compare young people's criminal justice histories with their own reports of offending, which often show two very different sides of the same coin. There are still many gaps in what we know about patterns of offending and the lives of offenders. This paper will conclude that a national survey of offenders is essential if we are to fully understand the phenomenon of youth crime in the future.

Thursday 9th May 2013
Professor Sheila Bird (MRC Biostatistics Unit)
Trial versus Before/After Policy Evaluation: effectiveness of take-home naloxone in reducing opiate overdose deaths
ICMS, 15 South College Street, Edinburgh EH8 9AA
Meeting starts 6pm, tea and coffee from 5.30pm

After a brief introduction to the epidemiology of Scotland's opiate overdose deaths (currently about 400 per annum) and the very high risk of overdose death soon after prison-release, international research and policy on take-home naloxone (opiate antidote) are briefly reviewed to set the scene for the design of the prison-based pilot N-ALIVE trial of naloxone-on-release in England and for the before/after evaluation of Scotland's public health policy (2011-2013) on take-home naloxone. Ethical issues, informed consent, record-linkage, likely cost per QALY, and efficacy versus effectiveness are discussed.

Tuesday 11th June 2013
Event to celebrate the International Year of Statistics 2013
Dr David A. Lagnado (University College London)
Stories and Statistics: what can the Sally Clark case tell us about the psychology of evidential reasoning?
ICMS, 15 South College Street, Edinburgh EH8 9AA
Meeting starts 6pm, tea and coffee from 5.30pm

Sally Clark was convicted of murdering her two children, but was eventually released after a lengthy legal process. This case has had substantial repercussions in the legal domain. In this talk I argue that the case also highlights many issues in the psychology of evidential reasoning. These include: the key role of causal reasoning; issues of witness reliability; the interpretation of probabilistic evidence; the role of stories and the attribution of blame. I will discuss how these issues can be elucidated within a general framework for evidential reasoning based on causal models.

Thursday 19th September 2013
Bayesian computing with INLA
ICMS, 15 South College Street, Edinburgh EH8 9AA
Meeting starts 6pm, tea and coffee from 5.30pm

The meeting will feature two talks:

Professor Andrew B Lawson (Department of Public Health Sciences, Medical University of South Carolina)
Bayesian Disease Mapping with INLA: an overview

Bayesian Disease Mapping often focusses on the hierarchical modeling of health outcomes in predefined small areas (e.g. postcodes, counties). The data level outcome is usually a count of disease and a risk parameter (relative risk) is to be estimated in small areas. A hierarchical model can be derived that includes a parsimonious description of the risk variation. When spatio-temporal (ST) variation is considered a variety of models can be conceived. In particular, separable effects are often assumed, with the addition of a ST interaction. The integrated nested Laplace Approximation has recently been efficiently implemented in R for a range of likelihoods and associated prior distributions (Rue et al, 2009). The package INLA can be used to fit a range of spatial and ST models to small area data. In this talk I will present a number of examples of the possible analyses for the famous Ohio county level respiratory cancer data set and, if time permits, parish level FMD data from the 2001 Cumbrian outbreak. Some example INLA code for spatial and ST models can be found in Appendix D of Lawson (2013) as well as the INLA website (http://www.r-inla.org/).

Rue, H., Martino, S., and Chopin, N. (2009) Approximate Bayesian inferences for latent Gaussian models by using integrated nested Laplace approximations. JRSS B, 71, 319-392
Lawson, A. B. (2013) Bayesian Disease Mapping: Hierarchical Modeling in Spatial Epidemiology. CRC Press 2nd Ed.

Dr. Janine B Illian (University of St. Andrews)
Fitting Complex Spatial Models in INLA: developments and extensions

Integrated nested Laplace approximation (INLA) may be used to fit a large class of (complex) statistical models. While MCMC methods use stochastic simulations for estimation, integrated nested Laplace approximation (INLA) is based on deterministic approximations where there are no convergence issues. INLA is a very accurate and computationally superior alternative to MCMC and may be used to fit a large class of models, latent Gaussian models.

Since INLA is fast, complex modelling has become greatly facilitated and has also become more accessible to non-specialists. In addition, due to the fact that the fitting approach is embedded in a large and general class of statistical models, very general types of models may be considered. This allows us a lot more flexibility in the choice of model than previously and hence the models to capture interesting aspects of the data and consequently the system they are relevant for. In the context of spatial statistics, for example, we can now fit models to spatial point patterns of high dimensionality, replicated point patterns, hierarchically marked point patterns etc. In many cases, analysing these data sets with MCMC approaches would be very cumbersome and computationally prohibitive.

The INLA-methodology has been implemented in C, and the associated numerical calculations and algorithms rely on an efficient implementation of numerical procedures for Gaussian Markov random fields (GMRF), in particular the algorithms in the C-library GMRFLib. However, most users do not need to worry about this, as the INLA-methodology has been made accessible through a user-friendly R-library, R-INLA, described and available for download at www.r-inla.org. Specifying and fitting models using R-INLA is just as easy as applying standard routines in R, for example fitting generalised linear models, and it also provides great flexibility with regard to the models that may be fitted. In order to illustrate INLA's versatility I will discuss a range of spatial and non-spatial examples and present a number of recent developments. This concerns generalisations of the methodology as well as new functionality within the R-INLA library.

Tuesday 12th November 2013
Freeing the power of administrative data: can Scotland become a global leader?
Dr. Stephen Pavis, Programme Director, Farr Institute (Scotland) ICMS, 15 South College Street, Edinburgh EH8 9AA
Meeting starts 6pm, tea and coffee from 5.30pm

Scotland has some of the best administrative data in the world (eg health, social care, housing, education and criminal justice), However, these data are not being fully exploited: to ensure our social policies are designed to respond to multiple disadvantage; our public services are as efficient as possible; and that these data are mobilised to support economic growth. This lecture will explore some of the reasons for the under utilisation of administrative data and chart some of the new initiatives which aim to release these data's potential.