Edinburgh RSS local group
Meetings 2012



Tuesday 24th January 2012
The role of uncertainty in adapting to the impacts of climate change within Scotland
ICMS, 15 South College Street, Edinburgh EH8 9AA
Meeting starts 6pm, tea and coffee from 5.30pm
Annual general meeting follows at 7pm - all are welcome

This meeting includes three speakers:

Dr. Joseph Hagg (Adaptation Scotland / Scottish Environment Protection Agency)
Probabilistic climate projections (UKCP09) - more than two years on...
The UK Climate Projections (UKCP09) are the latest generation of climate information for the United Kingdom, and for the first time provide a measure of uncertainty in the range of possible climate outcomes. However, more than two years after their release there remain significant challenges in using these probabilistic projections to both communicate climate change and inform decision-makers. I'll introduce UKCP09 and provide a few examples of how they challenge us to think about uncertainty in our future climate.

Dr. David Jenkins (Urban Energy Research Group, School of the Built Environment, Heriot-Watt University)
The probability of building 'failure' due to future climate change
The latest UK Climate Projections (UKCP09) provide an indication of future climates in a probabilistic format; that is, they allow the inherent uncertainty of climate models to be expressed in the output of a weather generator. This weather generator provides the user with a range of climate parameters (temperature, solar radiation, humidity etc) for a range of probabilities across different future climate scenarios. The data is therefore vast and requires appropriate management if focussing this information on a specific application. One application of climate information is for building simulation. Typically, someone simulating the thermal performance of a building will use a single reference climate to assess that building. UKCP09 requires a different approach that is potentially time-consuming, namely multiple simulations of hundreds (or even thousands) of climate files. To reduce the computational time of such an exercise, the Low Carbon Futures project (funded by the Adaptation and Resilience in a Changing Climate Programme) has developed an algorithm that, once calibrated on a single simulation, can emulate multiple simulations in an efficient way for a given building. This raises the possibility that the quantified uncertainty provided in UKCP09 can be translated into uncertainty in building performance. Specifically, for the Low Carbon Futures project, this uncertainty is applied to a future overheating risk analysis, indicating the probability that a building might exceed certain overheating criteria as a result of a warmer climate.

Prof. David Elston (Director of Biomathematics and Statistics Scotland; leader of the Signficance, Risk and Uncertainty Workstream of ClimateXChange)
Sources of variation and uncertainty in models of the impacts of climate change
The UKCP09 web interfaces allows users to create sequences of simulated weather data for a 5km by 5km square for a specified 30-year period. Sequences can be obtained that correspond to one of 10,000 climate samples for each of three emissions scenarios. We have investigated the use of these weather simulations to drive process-based, dynamic crop-environment models for silage production, spring barley and short-rotation coppice. I will describe some of the statistical aspects of this project, including the use of strata in the selection of climate samples and the partitioning of variance in the simulated yields into components associated with the strata, with the climate samples within strata, and with year-to-year variations in weather within climate samples.


Tuesday 27th March 2012
How much does statistical illiteracy cost?
David Walker, Director getStats and John Pullinger, Chair, getStats Campaign Board and RSS President-elect
ICMS, 15 South College Street, Edinburgh EH8 9AA
Meeting starts 6pm, tea and coffee from 5.30pm

Abstract

getStats is the Royal Statistical Society's numeracy campaign. The talk will draw on evidence on the misunderstanding of frequency, risk and probability and will chart the potential cost of such misunderstanding - in the consumption of collective goods such as health and education and in commercial markets. It will then turn to modeling the benefits of higher levels of statistical literacy on productivity and resource allocation. The talk will outline the ambitions of the getStats campaign, noting challenges to, and opportunities for, professional statisticians in widening public understanding of the application of statistical techniques. It will be followed by a discussion to identify priorities in Scotland for improving statistical literacy.

Biographies

David Walker is a communications professional, with extensive experience in journalism, research, public affairs and marketing. He is Director of getstats, the Royal Statistical Society's numeracy campaign and contributing editor of the Guardian's Public Leaders Network. Until 2010 he was Managing Director, Communications and Public Reporting, at the Audit Commission. He is a Council Member of the Economic and Social Research Council, visiting professor at King's College London, and was formerly a trustee of the Nuffield Trust for health services research.

John Pullinger is the Chair of the getStats Campaign Board and President-elect of the RSS. He is an RSS Council member, former chair of the RSS National Statistics Working Party and member of the RSS Long-Term Strategy Group. His day job is House of Commons Librarian, and prior to taking up that post in 2004 he was an Executive Director at the Office for National Statistics. He has also chaired Advisory Boards for departments and programmes at the Universities of Manchester, Essex and Leeds.


Royal Statistical Society Interdisciplinary Debate
Academics and clinicians from all disciplines and specialities are warmly invited to attend.

Tuesday 12 June 2012 (17.45 - 19.15)
Refreshments available from 17.15
Randomized controlled trials in medical research - gold standard or unhealthy fixation?

Is the randomized controlled trial (RCT) always the optimal choice in seeking an evidence base for causal relationships in clinical research? What role do the target population and ethical issues have to play in evaluating the external validity of an RCT? Indeed, are RCTs even necessary? These questions comprise a taster for the complex issues that will be raised during this lively informal debate involving experts from the fields of philosophy and medical statistics. Opportunities will be provided to ask questions and cast a final vote!

Poster: Download poster in PDF or RTF format.

Venue: Sydney Smith Lecture Theatre, Floor 2 (accessed via *Doorway 1*), University of Edinburgh Medical School, Teviot Place (http://www.ed.ac.uk/maps/buildings/medical-school)

Chair: Margaret MacDougall (email: Margaret.MacDougall@ed.ac.uk)


Wednesday 31st October
Louisa Blackwell and Nicola Rogers (2011 Census Administrative Data Matching Team, Office for National Statistics)
will be talking about work with Andrew Charlesworth on
"Matching of administrative data to validate the 2011 Census in England and Wales"
ICMS, 15 South College Street, Edinburgh EH8 9AA
Meeting starts 6pm, tea and coffee from 5.30pm
Download the poster for this meeting in PDF or RTF format

Abstract:

This presentation describes the role of administrative data matching in the quality assurance process for the 2011 Census in England and Wales. Data sharing across government allowed ONS to use administrative microdata to check and understand census counts and estimates. The sources will be described. The innovative data matching methods and systems designed for this task will be presented. A flexible, reactive approach applied analytic methods that were appropriate to the research questions that arose during the census quality assurance process. On their own and in combination, the administrative sources provided new insights, both empirical and theoretical, into population enumeration.

Data extracts from the NHS Patient Register, Valuation Office Agency, Electoral Registers, School Census and Higher Education Statistics Authority were matched at record level to create interacting matrices of linked information for dwellings and people. These were enriched with evidence from the Address Register History File. Matching ahead of census quality assurance focused on 37 local authorities representing a range of issues in administrative and census data. These data were mined and analysed at varying levels of detail, as required.

The data architecture, systems and software used to support this matching and analysis had to be developed and built as data availability and requirements became clearer ahead of Census. The tension between a desire for hard-coded, rigorously tested and robust systems and the reality of operational imperatives and the need for early results was carefully reconciled through bespoke, ad-hoc solutions for urgent requests that then required careful quality scrutiny. Address matching was confounded by the different reporting of addresses to different administrative bodies. Name capture and differential recording posed additional challenges. Through matching people we learned about the behaviours of different types of administrative data in different types of area. There were characteristic lags that impacted the currency of the data for different age/sex groups. Comparability and match rates between administrative sources and census were highly geographically specific, but also typical.


Tuesday 13th November
Dr. Lesley Graham MBChB MSc DRCOG MRCGP DMFPH (Associated Specialist, Public Health, ISD, NHS NSS)
Alcohol in Scotland: from evidence to action
ICMS, 15 South College Street, Edinburgh EH8 9AA
Meeting starts 6pm, tea and coffee from 5.30pm

Abstract:

Alcohol problems in Scotland are a major public health problem. Trends of alcohol-related harm rose dramatically in the past few decades in contrast to falling trends in other Western European countries. The lecture will describe the epidemiology of alcohol-related health and social harm in Scotland to the present day. It will also explain the crucial role that evidence, statistics in particular, played in raising the awareness of alcohol-related harm in Scotland with the public, politicians and policy makers. Evidence was also central to the development of the national alcohol strategy, the key elements of which will be outlined. The strategy adopts a whole population approach and includes the ground-breaking policy of Minimum Unit Pricing, the development and progress of which will also be described.