JOINT ST ANDREWS/HIGHLAND RSS GROUP MEETING ON SPATIAL STATISTICS
Wednesday 12 May
Lecture Theatre C, Mathematical Institute, North Haugh, St Andrews
3.00 p.m. Professor Daniel Griffith (University of Miami)
"So what should be the size of your geo-referenced sample?"
4.30 p.m. Dr. Patrick Brown (University of Lancaster)
"Some applications of spatial modelling in the veterinary sciences"
Tea will be available in the Staff Room between the talks.
The meeting will be followed at 6.30 p.m. by a 2-course meal
at the Doll's House Restaurant in Church Square, St Andrews. The cost
of the meal will be £15 (including coffee and a glass of wine or a
soft drink or a bottle of beer). The menu at this restaurant varies
from week to week; it is hoped that by about Friday 7 May the menu
for the relevant week will be available at
http://www-maths.mcs.st-and.ac.uk/StatsSeminars/menu.html.
(Vegetarian dishes are usually available.) If you intend to come,
please inform Peter Jupp, preferably by email ([log in to unmask]),
before noon on Monday 10 May.
-------------------------------------------------------
ABSTRACTS
Professor Daniel Griffith (University of Miami)
"So what should be the size of your geo-referenced sample?"
As spatial autocorrelation latent in geo-referenced data
increases, the amount of duplicate information contained in these
data also increases. This property suggests the research question
"What is the number of observations, say n*, that is equivalent to
the sample size, n, of a data set?". This is the notion of effective
sample size. Intuitively speaking, when zero spatial autocorrelation
prevails, n* = n; when perfect positive spatial autocorrelation
prevails, n* = 1. Equations are presented for estimating n* based on
the distribution of a sample mean or a sample correlation
coefficient, with the goal of obtaining some predetermined level of
precision, using the following spatial statistical model
specifications: (i) simultaneous autoregressive; (ii) geostatistical
semi-variogram; (iii) spatial filter. These equations are evaluated
with simulation experiments, and are illustrated with selected
empirical examples found in the literature.
Dr. Patrick Brown (University of Lancaster)
"Some applications of spatial modelling in the veterinary sciences"
This talk will present a number of problems in veterinary
epidemiology which have been solved with the use of spatial
statistics. The first problem presented deals with the spatial
distribution of Campylobacter Jejuni in cattle faeces. Binary
presence/absence was collected in an area of farmland in Cheshire,
and used to make inference about the spatial transmission of the
bacteria. In the same study the genetic distances between pairs of
samples were calculated, and whether the genetic distances are
related to spatial distances is examined. As the data relate to pairs
of observations, independence cannot be assumed and inference is
accomplished by permutating the observations. A final problem
explored relates to tick-born diseases in Tanzania, where the spatial
and temporal properties of the dataset are outlined.
==================================================================
Please check http://www-maths.mcs.st-and.ac.uk/StatsSeminars/ shortly
before coming, in case there are last-minute changes.
******************************************************************
Dr. P. E. Jupp
School of Mathematics and Statistics
University of St. Andrews
North Haugh, St. Andrews tel: (44) 1334 463704
Fife, KY16 9SS fax: (44) 1334 463748
Scotland e-mail: [log in to unmask]
******************************************************************
|