President
	Stephen M. Stigler (U.S.A.)
	President-Elect
	Niels Keiding (Denmark)
	Vice Presidents
Nicholas I. Fisher (Australia)	Jae Chang Lee (Korea)	Pilar Martín-Guzmán (Spain)
Newly elected ISI Council members (to serve from 2003-2007 period) with the ongoing Council members
			Ben Kiregyera (Uganda) No photo available
Carol Carson (USA)	Chihiro Hirotsu (Japan)	Graham Kalton (USA)
Maria Gabriella Ottaviani (Italy)	Jon N.K. Rao (Canada)	Simon Schwartzman (Brazil)	Klaus Trutzel (Germany)
Present Council members who will be continuing their service for a further two years are:
Jayant V. Deshpande (India)	Frederick W.H. Ho (Hong Kong SAR, China)	David S. Moore (USA)	Sylvia Richardson (France)
Paul Switzer (USA)	Ib O. Thomsen (Norway)	Norbert Victor (Germany)	Nicholas I. Fisher(Australia)

Select “On line access” to retrieve data from multiple countries and areas. According to the IDB, this data set "is a computerized data bank containing statistical tables of demographic and socio-economic data for 227 countries and areas of the world". There are data for 191 independent sovereign states and for 36 dependencies. The data base contains multi-year data on several basic demographic variables. For population, vital rates (births, deaths, net migration), and infant mortality, the data set is fairly complete for the years 1990 to 2000, and age distribution is almost complete for 2000. Completeness for other years vary. Population estimates and projections are available for all countries for time periods from 1950 to 2050. On the other hand, infant mortality is available for only 110 countries for 1980, and for fewer countries for earlier time periods. The IDB also has other variables, such as population by ethnic group, and by religious group, and prevalence of contraceptive use by age, however data for these variables are very limited. For example, many countries have no data for many of these variables and many other countries only have single years available. 

The IDB data is based on data from country sources, such as censuses and surveys, and on estimates from the Census Bureau’s International Programs Center. The Census Bureau's data set is public domain. The data at the Census web site is somewhat cumbersome to use, so we prepared a data set containing several variables (population, births, deaths, net migration and infant mortality), in a spreadsheet, available at http://gsociology.icaap.org .

UNICEF’s End of Decade Database, at http://childinfo.org/eddb/index2.htm , includes data for infant and child mortality rates from 1960 to 2000, in the Child Survival and Health section. The infant mortality rate table is at http://childinfo.org/cmr/revis/db1.htm  and the child mortality rate table is at http://childinfo.org/cmr/revis/db2.htm . This database has infant mortality data or estimates for most of the world. For 2000, there is infant mortality data for 186 countries. For 1980, there is infant mortality data for 171 countries. For 1960, there is infant mortality data for 159 countries. For all three of these years, there are data for over 95 percent of the world population, comparing the population of countries included in this data set to the total world population in the IDB data set. There is child mortality data for similar numbers of countries. The UNICEF site also presents data for "industrialized", "developing" and "least developed" groupings, but doesn't seem to indicate how those categories are constructed. 

Documentation on how data are obtained or estimated is available at http://childinfo.org/cmr/revis/kh98meth.html . According to chapter 2, infant and child mortality rate are usually calculated using civil registration data. Where registry data is less reliable or complete, UNICEF uses various survey data for estimations. 

WHO presents percent of population age 60 and above, and fertility data, for 1990 and 2000, at http://www.who.int/whr2001/2001/main/en/annex/annex1.htm . Most of the other data presented at the WHO site is only for 2000. The data set has aging and fertility data for 191 countries, which includes more than 90% of the world population, comparing the population total in this data set to the total world population in the IDB data set.

The WHO notes page http://www.who.int/whr/2001/main/en/annex/annexNotes.htm  says that they present data on countries, based on "a systematic review of all available evidence from surveys, censuses, sample registration systems, population laboratories and vital registration on levels and trends in child mortality and adult mortality". They especially cite UNICEF, the United States Census Bureau and the UN Population Division 2000 demographic assessment. 

The United Nations Development Program, Human Development Report 2001, at http://www.undp.org/hdr2001/indicator/index.html  has urbanization data for 1975 and 1999. Two tables (tables 48 and 49) show percent and number urban by country, and two tables (tables 43 and 44) show total population. These tables can be used to calculate urbanization for the world or for any region desired. This data set has urbanization data for both years for 162 countries and for over 90 percent of the world population, comparing the population total in this data set to the total world population in the IDB data set. 

An appendix, “Human Development Indicators” includes “Notes on statistics in the Human Development Report” at http://www.undp.org/hdr2001/indicator/index.html  which indicates that where possible, the data tables are “based on internationally standardized data, collected and processed by sister agencies in the international system, or, in a few cases, by other organizations” (p. 133). Specific tables include references for data sources. Urbanization data is “The mid-year population of areas defined as urban in each country, as reported to the United Nations” (definition as given in the urbanization tables).

The UNESCO Institute for Statistics, http://www.uis.UNESCO.org  has literacy data for 1970 to 2000 for every 5th year. Click "literacy", then click "statistical tables". According to the UNESCO Institute for Statistics, "a person is literate who can, with understanding, both read and write a short simple statement on his everyday life." (Facts and Figures 2000, page 23. Report available at http://www.uis.UNESCO.org  then click Facts and Figures, then click Publications). Literacy data was obtained using methods described in “Methodology used in the 1994 estimation and projection of adult illiteracy”, available at http://www.uis.UNESCO.org  then click “literacy” then click “technical guidelines”. The data in the statistical table seems to exclude many advanced countries, such as the United States, United Kingdom, France, Germany, Australia, Austria, Belgium, Canada, Denmark, Finland, Japan, Netherlands, Norway, Sweden, and Switzerland. The data set does, however, include data for 138 countries, which includes almost 85 percent of the world population, comparing the population total for the countries in this data set to the total world population in the IDB data set. 

Barro and Lee, at the Center for International Development also present a data set on literacy, the International Data on Educational Attainment, 1960 to 2000, at http://www.cid.harvard.edu/ciddata/ciddata.html . The variables in this data set include percent of population with no schooling, percent at level of schooling attained (i.e., secondary, post secondary) and average years of school. This data set has data for, depending on the year and variable, 110 to 120 countries, and almost 90 percent of the world population, comparing the population of countries included in this data set to the total world population in the IDB data set. 

The data are explained in an April 2000 working paper of Barro and Lee, International Data on Educational Attainment: Updates and Implications, at http://www.cid.harvard.edu/cidwp/ . 

Barro and Lee indicate their data was from census/survey information complied by UNESCO and other sources, with statistical estimations for missing data and for average years of schooling. 

Ethnolinguistic Fractionalization is from: Philip G. Roeder. 2001. "Ethnolinguistic Fractionalization (ELF) Indices, 1961 and 1985" available at http://weber.ucsd.edu/~proeder/data.htm . Dr. Roeder defines ELF as "the probability that two individuals chosen at random from the population (country) will be from different ethnic groups". Further, "It increases as the number of groups increases and their proportionate sizes decrease. So a country with many ethnic groups, each of which has about the same number of members, would have the highest ELF. As one ethnic group comes to occupy a larger share of the population, the ELF declines" (personal communication). In this data set, there is data for 138 countries in 1961, about 86 percent of the world population, comparing the population of countries included in this data set to the total world population in the IDB data set. Almost all countries with 1961 data also have data for 1985, so have data for two time periods. There are a few more countries that have data for 1985 but not 1961.

The data sources and some notes on defining ethnic groups are described at http://weber.ucsd.edu/~proeder/elf.htm . The sources appear to be statistical yearbooks published mainly in Moscow. 

While all of these data sets include over 100 countries, the same countries are not always in all of the data sets. For example, there are only 75 countries in common among the US Census IDB data set, UNESCO’s literacy data set, UNESCO’s urbanization data set, the ethnolinguistic data set and Barro and Lee’s literacy data set. However, these data sets can be used to tell a consistent story. For example, several of our reports (available at http://gsociology.icaap.org/reports.html ) use these data sets to show a relatively coherent picture among developing countries of declines in population growth, illiteracy, and infant mortality rates, and increases in urbanization and percent elderly, patterns which are consistent, for example, with a demographic transition.

Gene Shackman*
Social Change Website Manager
http://gsociology.icaap.org/ 

* Neither Dr. Shackman nor ISI endorse any of the sites listed here, and do not assume responsibility for content of the Websites listed in this article. This article is solely presented for educational and reference purposes.

Top

---

PARIS21 annual meeting (17-18 0ctober 2002) - 

Report of the ISI Representative

PARIS21 held its third annual meeting in the OECD premises in Paris, from 17-18 October 2002. The Steering Committee had decided that this year’s meeting was to be focussed on the work programme with those directly involved. A full Consortium meeting will thus take place next year.

Altogether, 50 countries and international organisations were represented: 29 countries (8 donor countries; 21 beneficiary countries); 20 international agencies and organisations, plus 1 international consultant.

Several reports were available: PARIS21 progress report; Andean Community Countries meeting report; outline of the first Central America PARIS21 workshop, to be held in Salvador in 2003; report of the Task Team (TT) on Statistical Capacity Building Indicators; Agriculture Statistics TT report; “Governance of national statistical systems”, by W. de Vries (UNSD); “Supporting policy with statistics”, B. Kiregyera’s report for the PARIS21 sub-regional workshop held in Kampala on 23-26 July 2001; a Eurostat note on their new SWOT (Strengths, Weaknesses, Opportunities, Threats) analysis (for complete information, see www.paris21.org ).

This year, three topics were high on the agenda: the relationship between PARIS21 on the one hand and developing countries’ statistical regional and national levels on the other hand; as well as the relationship between donors and PARIS21, and finally PARIS21 perspectives.

1- For the time being, PARIS21 is emphasising its supportive approach to developing countries basically through regional levels. This approach has proved to be quite useful and is largely recognised as one of PARIS21’s best assets. However, it may also be felt that, maintaining the supporting focus only at this level could, in time, lead to some frustration as the actual decision making process and implementation practice lie obviously at national levels.

From the views expressed during the discussion, mostly by donors but also by some beneficiary country representatives, it became apparent that PARIS21 was “more a philosophy than a tool”, a “body to bring reflection on statistics, not to make statistics”, whose “first priority was not to design statistical projects”. “PARIS21 stakeholders were indeed confronted with a contradiction: on the one hand, no one would deny that it is only at national levels that the upgrading of national statistics could be successfully completed, however, on the other hand, PARIS21 was not intended to become a new instrumental tool in the international statistical field”. 

In conclusion, it was generally agreed that the PARIS21 Secretariat should continue working at regional levels, supplying expertise for the design of general outlines of statistical master plans and projects (the example of Afristat was mentioned) and invest at national levels only in cases of emergency (emerging from a political crisis, for instance). Therefore, national workshops and seminars should be kept under the aegis and responsibility of regional statistical authorities. 

2- PARIS21 has gained substantial support from donors working in the field of statistical technical assistance. Therefore, one of its roles could be to assist developing countries in avoiding any overlap between their own statistical programmes and those of their donors. In this respect, mention has been made of possible overlap between the Poverty Reduction Strategy Papers (PRSP) and the national Strategic Statistical Development Plans (SSDP). Of course, PRSPs are not only utilised to introduce statistical indicators suited to the purpose of fighting poverty. As they are progressively built up, they highlight the limits, deficiencies, inadequacies, etc., of national statistical systems, hence pleading for their upgrading. 

However, discussions are not always easy between national statisticians and their foreign partners: generally speaking, donors are mostly interested in indicators related to the projects they promote; also, they may not have a clear view of statistical systems as a whole Furthermore, on the donors’ side, people who are involved in the follow up of projects are primarily administrators, not statisticians and a mutual understanding between these two categories of professionals is not easy to achieve. Last but not least, as one participant emphasised, even if, as far as principles are concerned, consultation and co-ordination between donors are satisfactory at HQ level, much remains to be done at the field level where the specificities of each international organisation are enhanced and defended by their local representatives. On this specific point, a suggestion was made to hold fora for these local representatives aiming to train people within the co-ordinating spirit of PARIS21, keeping in mind the goal of building up national strategies for statistics.

3- Fighting poverty and improving national statistical systems are nowadays donors’ priorities. For this two-pronged purpose, quite a number of projects and/or programmes have been proposed: PRSP and CWIQ (World Bank), GDDS, DQAF, ROSC (IMF), MISP and SWOT (Eurostat), PROSMIC (Afristat), etc. The first and most obvious risk of this flood of ideas and proposals is that they may overwhelm statisticians from developing countries under a huge burden of tasks (not accounting for the numerous international questionnaires they are already asked to answer) should they immediately be required to conform to the requirements that these programmes stipulate. In this respect, a simple, obvious and quite sensible remark has been made by one developing country representative: never forget the necessary national human resources that are to be dedicated to the implementation and follow-up of internationally-proposed programmes or projects. This is a simple, obvious and sensible remark, but in practice it is rarely, if ever, taken into account. 

The participants of the meeting agreed that PARIS21 was doing excellent work, and several suggestions were made to make some further steps forward; for instance:

- If we should always remember that Statistics exist for people to use them, this basic “philosophy” should not allow us to forget that statistical producers also exist, and that they have their own specificities and constraints. In other words, the correct balance has to be found, in any statistical plan or programme, between the shortsightedness of the demand and the statistician’s risk of taking the long-term perspective.

- There is a risk to limit PARIS21 debates to statisticians only. Therefore, consideration should be given to the possibility of a wider definition of the Consortium (up to what point remains to be discussed). Does it include policy makers from developing countries? Does it include all professionals, such as teachers, the media, etc., using statistics in the normal course of their professional activities?

4- Personal comments by way of conclusion

Last year, these personal comments focussed on the challenge that PARIS21 had to tackle for the future; how to draw a line between statistical projects and statistical master plans. In the first case, answers were technical and usually translated in terms of indicators, whereas the second orientation resulted in the definition and follow up of a policy in the field of statistics. 

To-day, the feeling is that the last Consortium meeting has moved forward in this respect: international organisations seem to be less reluctant to admit that they are not the only ones investing in the statistical field and that “national voices” have also to be heard. Finally, the PARIS21 Secretariat has decided to launch a specific reflection on the topic “Synergy between international and national statistical initiatives”.

Alain Azouvi

PRSP: Poverty Reduction Strategy Paper
CWIQ: Core Welfare Indicator Questionnaire
GDDS: General Data Dissemination System
DQAF: Data Quality Assessment Framework
ROSC: Report on Observance of Standards and Codes
MISP: Multi-annual Integrated Statistical Programme
SWOT: Strengths, Weaknesses, Opportunities, Threats
PROSMIC: PROgramme Statistique MInimum Commun.

Top

---

Director of the ISI Permanent Office

The position of Director, ISI Permanent Office, will become vacant in August/September 2003 due to the retirement of Dr. Marcel Van den Broecke. A Search committee composed of Messrs. Dennis Trewin (Chairman), Steve Stigler, Jef Teugels, and Bill van Zwet has been established. 

Among other qualifications, the following are required:

• completed university education

• managerial experience and qualifications and experience as a senior administrator of a self-contained business unit is highly desirable

• fluency in English, some knowledge of French and Dutch

• knowledge of Dutch labour legislation is highly desirable

• some knowledge of Internet-based publishing

• diplomatic qualities in performing in an international environment

• good management skills concerning personnel and finances, and judgement to make appropriate decisions whilst working in a largely isolated environment

• ability to provide leadership to the work of the ISI both at the level of the ISI Permanent Office and contributing to the Executive Committee’s strategic planning work.

Arrangements will depend upon the qualifications and experience of the new Director. Salary commensurate with the responsibilities of the position.

Interested candidates can direct enquiries to the chair of the Search Committee (dennis.trewin@abs.gov.au ). Formal application with full biographical data should be sent as soon as possible and not later than April 1, 2003 to:

Dr. Marcel Van den Broecke

Director of the Permanent Office

International Statistical Institute, 

P.O. Box 950

2270 AZ Voorburg

The Netherlands.

Fax: +31 70 3860025

 @cbs.nl

 

Top

---

Examining the Essential Functions of Statistical Organisations

GUS/ISI Satellite Conference - Szczecin, Poland, 8 – 9 August 2003

 

Managers of national statistical offices must view their responsibility for organisational design in a way that takes into consideration the constantly evolving society that these organisations measure. Yesterday’s organisational paradigms are today’s dinosours. With this in mind the Polish Central Statistical Office (GUS) together with the International Statistical Institute (ISI), have organised a conference examining the essential functions of official statistical organisations in anticipation of recent and forthcoming changes.

There will be eight sub-topics included in this conference, each one organised by respected officials in the statistical community: 

• Identifying changes: New Environments for Official Statisticians (Organiser: Marcel Van den Broecke / Daniel Berze, ISI Permanent Office)
• Impact of Technologies on Statistical Operations and organisations (Organiser: Dennis Trewin, Australian Bureau of Statistics)
• Functions and structures of statistical organisations (Organiser: Willem de vries, UNSD)
• Personnel in statistical organisations (Organiser: Jean-Louis Bodin, ADETEF)
• Quality issues (Organiser: Ivan Fellegi, Statistics Canada)
• Dissemination of Statistical Information in a Knowledge Based Society (Organiser: Janusz Witkowski, GUS)
• User considerations (Organiser: Joris Nobel, Statistics Netherlands)
• Strategic Planning & Management in Statistical Organisations (Organiser: Photis Nanopoulos, Eurostat)

 

Conference speakers will include individuals who are recognized and respected authorities on the issues they will discuss, including Geert Bruinooge, Heinrich Brüngger, Len Cook, Ovidio Crocicchi, Donal Garvey, Heli Jeskanen-Sundström, Joachim Lamel, Denise Lievesley, Josef Olenski, Zoltan Papp, Jan Plovsing, Peter Pupkli, Marc Roure, Jacob Ryten, Algirdas Semeta, Tadeusz Toczynski, Tadeusz Walczak and Aija Zigure.  Author abstracts will appear on the conference website shortly. This will be a satellite conference to the Berlin ISI Session. Participants at the Berlin Session will be able to reach the conference, to be held in Szczecin, Poland, via the regular train connection between the two cities (2 hours 15 minutes travel time), via a scheduled shuttle bus that is operated by the hotel in which the conference is situated (1.5 hours travel time) or by car. The regional airport in Goleniow, at a distance of 40 kilometres from the city of Szczecin, is also an option.  The conference will be launched with a cocktail reception on August 7 (at which time participants can also register). The scientific programme will take place on August 8 and 9. The conference will take place at the five star Radisson SAS Hotel in Szczecin.

 

To make this visit a memorable visit, the Polish organisers have taken special efforts to arrange a social programme that will be quite special, including a welcome reception at the National Museum in Szczecin, an official dinner at the Palace in Maciejewo (a 100 year old Palace which once served as the summer residence of the former von Flemming family and which is now a luxury hotel surrounded by lakes and forests), and a Concert at the Szczecin Pomeranian Dukes Castle.  Due to the size limitations of the venue (150 persons maximum), registration applications will be accepted on a first come first served basis. Registration fees (125 Euro before May 31 and 150 Euros after, with an accompanying persons fee of 100 Euro) will include participation in all conference scientific and social events and lunch on August 8 and 9.

For more information, please see the conference website at ( http://www.stat.gov.pl/ISI.htm ) 

or contact: 

• Barbara Domaszewicz, Central Statistical Office of Poland

Phone: + 48 22 608 30 50; Fax: + 48 22 608 38 70; E-mail: B.Domaszewicz@stat.gov.pl 

Top

---

Programme of Short Courses on Survey Methodology in Berlin 2003

August 11-12, 2003

The International Statistical Institute (ISI) will sponsor a programme of short courses to take place before the 54th ISI Session to be held in Berlin. The courses will be led by international high-level experts in their respective fields and will offer statisticians and students of statistics the opportunity to improve their skills on the topic being presented and discuss issues of common interest. All courses will be presented in English and participants should have the ability to work in this language. There are some opportunities to sponsor travel and accommodation costs for participants from developing and transition countries. 

The list of courses is attached. The deadlines for registration and the course fees are attached. Information will be available via the conference website at ( http://www.isi-2003.de/ ). Please look for possible changes and additional information on these websites. Any questions, requests, sponsoring opportunity inquiries and comments are welcome, and can be sent by e-mail to Daniel Berze at @cbs.nl. 

Course 8. - Analysis of Longitudinal Data (cancelled)

 

Course 9. – Recent Developments in Graphical Markov Models (cancelled)

Course 10. – Analysis of Categorical Data with Overdispersion

Course Organiser: Nanny Wermuth (Johannes-Gutenberg University, Mainz)

Instructors: Jorge G. Morel (Department of Biometrics and Statistical Sciences, P&G Pharmaceuticals, Cincinnati, USA) and Professor Nagaraj K. Neerchal

Date: 12 August, 2003 (four hours in duration)

Audience: The course in intended for people with at least a Master’s degree in statistics. Basic knowledge of the binomial and multinomial distributions, logistic regression, maximum likelihood estimation and Fisher’s information matrix is needed.

Venue: Freie Universität Berlin

Short description: The aim of the course is to present a general overview of the phenomenon of extravariation relative to the binomial/multinomial distribution, as well as to provide different methods (quasi-likelihood functions, likelihood models, and generalized estimating equations) to cope with this problem. Several practical examples will be shown to illustrate the available methodology to model categorical data with extravariation.

Topics to be covered:

1. Introduction. Characterization of extravariation (overdispersion) relative to the binomial/multinomial distribution. Relevant examples. Clusters, the intra cluster correlation and the design effect. Quasi-likelihood functions. Iterated reweighed least squares. An example with ossification data.

2. Likelihood models for binary responses. The beta-binomial distribution (Skellam, 1948), a binomial distribution with extra zeros (Cohen, 1966), two generalizations of the binomial distribution (Altham, 1978), a correlated binomial model (Haseman and Kupper, 1979), a finite mixture model (Morel and Neerchal, 1997). An example with secondary association of chromosomes data. 

3. A goodness-of-fit test for testing the adequacy of specific likelihood models when cluster sizes vary (Neerchal and Morel, 1998). Illustration using data from an experiment designed to study induced mutagenic effect (Haseman & Soares, 1979).

4. Likelihood models for multinomial responses. The Dirichlet-Multinomial (Compound-Multinomial) distribution (Mosimann, 1962). The multinomial finite-mixture “clumped” distribution (Morel and Nagaraj, 1993). Asymptotic properties for large cluster size. A forest count example.

5. Generalized estimating equations techniques (GEE). Fundamental concepts (Liang and Zeger, 1986, and Zeger and Liang, 1986). Exponential family, link functions, and “working” correlation matrices. Taylor’s series expansion, asymptotic properties and the robustness of GEE. Binary responses and different link functions (logit, probit, complementary log-log). Examples using the SAS^® procedure “GENMOD.” Small sample size corrections (Morel, 1989, Mancl and DeRouen, 2001). Generalized Estimating Equations techniques for multinomial data. The generalized logistic and the proportional-odds models. Working correlation matrices. An example on a shoulder pain study (Lumley, 1996).

Top

---

Back to Home Page

Other Newsletters

News from ISI Sections in Volume 27, No. 1 (79) 2003

---