The International Association of Statistical Computing - European Regional Section (IASC-ERS) and the Institute of Statistics, University of Neuchâtel organize the summer school
ADVANCED METHODS IN SAMPLING AND ESTIMATION IN FINITE POPULATIONS
30 August - 1 September 2017, Neuchâtel, Switzerland
About the IASC-ERS summer school
The IASC-ERS Summer school is intended to provide training in special areas of statistics for PhD students, junior researchers and lecturers at universities. Professionals working in industry who are interested in the application of new statistical methods are also welcome to participate. Participants are expected to have good background in statistics at the M.Sc. level although not necessarily related to the subject of the course.
Aims of the summer school
Probabilistic sampling methods from finite populations are used in many fields, from demographics, economics, public health to official statistics, but also in environmetrics or forestry. In this summer course we will start by introducing basic ideas of sampling from finite populations, as well as usual estimation methods. Advanced sampling methods will also be covered, showing recently proposed methods, such as balanced sampling, balanced spatial sampling, etc., as well as advanced estimation methods based on bootstrap in finite populations. The summer course will bring together the theoretical and practical aspects of sampling and estimating in finite populations. Exercises will be provided using the R software. Knowledge of R is not required (an introduction to R will be given), but knowledge of a statistical software is a must. The course is given at the M.Sc. level, and does not require preliminary knowledge of sampling and estimation in finite populations.
Speakers
Guillaume Chauvet, ENSAI - National School for Statistics and Information Analysis, Rennes, France,
Anton Grafström, Swedish University of Agricultural Studies, Umea, Sweden,
Alina Matei, University of Neuchâtel, Switzerland,
Yves Tillé, University of Neuchâtel, Switzerland.
Organizing committee
Alina Matei, University of Neuchâtel, Switzerland
Yves Tillé, University of Neuchâtel, Switzerland
Adminstrative coordinator, Corine Diacon, University of Neuchâtel, Switzerland
Location
Faculty of Science, University of Neuchâtel, UniMail Building, Rue Emile-Argand 11, 2000, Neuchâtel
****************************************************************************************************************************************
Registration Fee: 250 Swiss francs (free for students and researchers of the University of Neuchâtel).
Attendance : Participation to the Summer school is limited to maximum 30 participants,
on a first-come, first-served basis.
Registration : By e-mail at Institute of Statistics, University of Neuchâtel, Switzerland :
Deadline 10 July 2017 (possible registration after this date only if places are still available).
****************************************************************************************************************************************
SWIFT/BIC : POFICHBEXX
SWIFT/BIC : UBSWCHZH80A
ADVANCED METHODS IN SAMPLING AND ESTIMATION IN FINITE POPULATIONS
30 August - 1 September 2017, Neuchâtel, Switzerland
The summer scholl includes three lectures (in the morning) and three lab sessions using R (in the afternoon).
Lecture 1
Yves Tillé - Survey sampling, main definitions and tools
This first course introduces the main notion, tools and definitions of the theory of survey sampling. After defining a sampling design, we study the different ways of conducting inference in surveys in the design-based and model-based frameworks. The mains sampling designs are presented including performing algorithms for unequal probability sampling and balanced sampling. The different techniques of estimation based on the theory of calibration are presented in details including the question of variance estimation. The introduction concludes by the establishment of general rules to conceive a sampling strategy composed of a sampling design and an estimator.
Lecture 2
Guillaume Chauvet - Bootstrap for complex sampling designs
Survey data are obtained by means of complex sampling designs, which may vary depending on the populations under study. For example, business surveys are typically performed by means of stratified simple random sampling on a register of businesses, while household surveys and epidemiologic surveys are usually carried through multistage sampling designs. The resulting estimators account for the way the samples are selected, and the measures of accuracy must be defined accordingly. Among the possible ways for computing measures of accuracy, bootstrap enables to produce variance estimators or confidence intervals for parameters via resampling in the original sampled data. The resampling scheme is supposed to mimic the way the data were originally sampled, which is challenging in the context of complex survey designs. In this lecture, we will present some bootstrap methods proposed for survey sampling. We will illustrate them on practical examples and/or simulated data.
Lecture 3
Anton Grafström - Spatial sampling
In spatial sampling a goal has been to select probability samples that are well spread over the population. One of the reasons for that is that for populations with spatial trends in the variables of interest, the estimation can be much improved by selecting samples that are well spread over the population. The concept of spatial balance of a sample is introduced as a measure of spread of a sample, and as a distance between the sample and the sampling distribution. The connection between balanced samples, spatially balanced samples and representative samples will be discussed. Novel sampling methods to select probability samples that are well spread in any auxiliary space will be presented. In particular the local pivotal method, spatially correlated Poisson sampling and the local cube method will be covered. Some examples end exercises will be done using the R package Balanced Sampling.
Alina Matei - Some applications in R (lab sessions)
The software environment R is recognized today as an important statistical computing tool, providing a convenient framework to implement functions related to random sampling and estimation in finite populations. Therefore, applications to some of the topics presented during the lectures (such as unequal probability sampling designs, balanced sampling, calibration, generalized calibration, etc.) will be illustrated using the R package 'sampling'. Knowledge in R is not required; a short introduction to R will be given. It is however expected from the participants to have an important interest in statistical computing.
ADVANCED METHODS IN SAMPLING AND ESTIMATION IN FINITE POPULATIONS
30 August - 1 September 2017, Neuchâtel, Switzerland
Summer school schedule
The lecturers will take place in the room B013, while the R lab sessions in the computer room E026 (both rooms are in the UniMail Building, Rue Emile-Argand 11, 2000, Neuchâtel).
Wednesday, 30 August 2017
08.30-9.00 - Registration (UniMail Building, ground floor, north entrance)
09.00-10.30 – Lecture 1 – part I
10.30-11.00 – Coffee break
11.00 -12.30 – Lecture 1 – part II
12.30-14.00 – Lunch break
14.00-15.00 – R session 1 – part I
15.00-15.30 – Coffee break
15.30-16.30 – R session 1 – part II
Lecture 1
Yves Tillé – Survey sampling, main definitions and tools
R session 1
Alina Matei – An introduction to R
(R objects and operators, data structures, basic graphics, packages, etc.).
___________________________________________________________________________
Thursday, 31 August 2017
09.00-10.30 – Lecture 2 – part I
10.30-11.00 – Coffee break
11.00 -12.30 – Lecture 2 – part II
12.30-14.00 – Lunch break
14.00-15.00 – R session 2 – part I
15.00-15.30 – Coffee break
15.30 -16.30 – R session 2 – part II
Lecture 2
Guillaume Chauvet – Bootstrap for complex sampling designs
R session 2
Alina Matei – Applications on unequal probability sampling designs and balanced sampling
___________________________________________________________________________
Friday, 1 September 2017
09.00-10.30 – Lecture 3 – part I
10.30-11.00 – Coffee break
11.00 -12.30 – Lecture 3 – part II
12.30-14.00 - Lunch break
14.00-15.00 – R session 3 – part I
15.00-15.30 – Coffee break
15.30 -16.30 – R session 3 – part II
Lecture 3
Anton Grafström – Spatial sampling
R session 3
Alina Matei – Applications on calibration and generalized calibration
