Aims and history

The International Association of Survey Statisticians (IASS – AISE) was founded in 1973 as a section of the International Statistical Institute. It aims to promote the study and development of the theory and practice of sample surveys and censuses. It also aims to increase the interest in surveys and censuses among statisticians, among governments and the public in the different countries of the world. The IASS carries out its aims by means of:

  • International meetings held every two years in conjunction with the International Statistical Institute. These meetings provide a forum for discussion of survey statistics. They include a number of sessions on recent advances in survey and census methodology, and their applications.
  • Regional meetings or seminars devoted to specific aspects of surveys.
  • Publications distributed to members on a regular basis or on request, and the opportunity for members to subscribe to certain journals at reduced rates.

The launch of IASS

The first large-scale appearance of the IASS took place during the 39th ISI session in the castle of Hofburg, Vienna, Austria, August 18-30 1973. These are some personal memories of this event, based on my report to my employer by that time, the Audience and Research Department of the Swedish Broadcasting Corporation. I was also a student of Tore Dalenius’s, professor at the University of Stockholm, where he taught sampling and survey theory. He was the one who encouraged me to enroll in the Association and attend the session.

Tore was one of the founders of the Association, together with other legendary survey statisticians like P. C. Mahalanobis (who died before the session) Morris Hansen, Leslie Kish, and Ivan Fellegi. The latter had laid down excellent work as chair of the organizing committee of the IASS part of the conference. All invited papers meetings were organized jointly with ISI and published in the proceedings, whereas you had to be present to collect hand-outs for contributed papers. Some of the contributed papers were, however, published in the Indian journal Sankhya afterwards.

Still today, many statisticians working outside of the survey field believe that our association is devoted to survey sampling solely. No doubt, sampling was, and still is, a major subject of ours. However, the most important reason for the creation of the IASS was to address major limitations of sampling theory. Those had been identified in articles written by P C Mahalanobis and W E Deming, the major one being that sampling theory was a theory for true values that did not take into consideration measurement errors.

This was very exciting to me, because in my job with methodology for TV audience surveys based on telephone interviewing, I had realized that sampling was not the only soruce of error; we had also nonresponse and recall bias, to mention the two most serious ones. Of particular interest to me was therefore Gad Nathan’s presentation of experiences of sampling and nonsampling errors in Israel, since one of the examples was the TV audience surveys. Gad reported a study in which a one-day recall period was compared to a one-week period. The longer recall period was found to be better because the increase of recall bias was more than balanced by the decrease of sampling error. This was presented in a meeting on the balance between the two sources of errors, organized by Tore Dalenius. Contributions to this meeting came also from Fellegi & Sunter, Canada, and Jabine & Smith. USA. So, you could design a survey subject to a mean square error criterion rather than a sampling variance criterion, thus incorporating measurement bias into the design. Later on, we used a similar criterion to improve the quality of the Swedish TV audience surveys. I think that there is still a need to bring up similar issues on our conference agendas.

Many invited and contributed errors addressed nonsampling errors, how to control them, and bring them together with the sampling errors in a survey model. The US Bureau of the Census survey model was hot at the time and efforts to estimate its components in different applications were presented by Ivan Fellegi, Canada, and Fred Smith, UK.

Sampling theory was by this time a theory for the estimation of a restricted number of parameters, e g arithmetic means, ratios, and totals. However, an optimally designed stratified sample for the estimation of a mean, is not necessarily a good one when you want to estimate a regression slope. This limitation was addressed in an invited papers’ meeting, Analytical uses of and inferences from sample surveys, organized by J N K Rao, Canada.

Superpopulation models was a new concept to me. It was introduced by Scott & Smith (UK) and others to address a third limitation of sampling theory: that it was a theory for finite populations, whereas we often are more interested in the superpopulations (I think that we rather call them processes today) that generate them. However, in my report, I noted that even though this might be true, It may be a good start to study the finite populations that the processes generate.

Sampling got its share as well. Foundations of survey sampling (organized by C E Särndal, Canada) Sampling in two or three dimensions with respect to ecological problems (A R Sen, Canada) and Large scale multipurpose computations of sampling sampling errors (L Kish, USA) were titles of invited papers’ sessions. In addition a number of contributed papers were devoted to sampling problems. Other topics treated were, Multi-subject surveys, Sample surveys in the private sector, and Dual system estimates.

It was very stimulating to take part in this conference. There were excellent presentations and refreshing discussions on different subjects throughout the IASS program, and I came back home with a number of ideas for future survey methodology research. One reflection that comes to my mind now that I am writing this is: How could they organize it so well without the e-mail? The Austrian hosts excellently organized the whole conference. The social program included tours to the surroundings and a marvelous Farewell Heuriger party. Also the weather was on our side, presenting a warm late summer sun. I remember my favorite place to cool down in the weekend was the pool in Krapfenwald.

Most of all I remember all the new friends that I got. I can still meet them at IASS conferences, remembering that in Vienna, we were all young and handsome.

Anders Christianson

Present at the creation?

In the January, 2005 issue of The Survey Statistician Anders Christianson wrote an article about “The Launching of the IASS in Vienna 1973”. I have subsequently been asked to contribute my recollection of this event which was obviously seminal in the life of IASS and, indeed in that of ISI itself.

The ISI has, for decades now (probably since its beginnings), been searching to find ways to remain relevant. During its 37th Session held in London in 1969 a fundamentally important reappraisal of the mission and modalities of ISI was tabled; the committee which authored it had been chaired by M. G. Kendall (“Report of the Reappraisal Committee”).It contained wide-ranging recommendations which are well worth rereading even today. A segment of the report dealt with the possible creation of new sections within the Institute. It suggested that “if a strong feeling arises that it would be an advantage to create an international association in a new field, we suggest the Institute ought to take the lead and set up such a society as a section of the Institute”.

I don’t know whether the Reappraisal Committee was initiating a new discussion about sections of ISI or whether it was reacting to ideas already circulating then. As it happens, several of us had, in fact, been agitating for a new section, one devoted to survey methodology, a field that was, in most countries in the late 1960s, still quite an undeveloped field. A very influential advocate was P. C. Mahalanobis. As a “young Turk”, indeed a very young one, I was also actively agitating in favour of such a development at every available opportunity, whispering in whatever influential ear was polite enough to let me do so.

It was during the ISI session in London in 1969 that the Bureau of the ISI (chaired at that time by W. G. Cochran) asked me to join them for a discussion about the possibility of forming a section of the ISI devoted to this new and aggressively evolving discipline. They listened to me making a brief pitch, but it was my sense that they already had their minds made up to give it a try – so I cannot claim that silver-tongued oratory, even less my impeccably argued case, convinced them. As is usual, when you talk too much, you end up being asked to do what you have been agitating for. So I was asked identify to chair a small committee whose first task would be to draft terms statutes for the putative new association.

I suggested to the Bureau that the committee consist of J. P M. R. Desabie, Leslie Kish, M. N. Murthy, M. R. Sampford and S. Zarkovich and my recommendations were accepted. The drafting committee was in business. Our task was helped by an early draft of the statutes prepared by P. C. Mahalanobis although we did, in fact, draft the new statutes according to what we thought was needed. We worked by correspondence over a period of many months and I must underline what a particularly valuable member Leslie Kish was: he never failed to respond to correspondence and he was full of good ideas – anyone surprised?

We submitted our work to the Bureau – not for approval of the statutes themselves, since we thought (and the Bureau agreed) that only the new association could adopt its own statutes – but to secure their blessing for the formation of a new Section of the ISI. They did, indeed, put forward our recommendations to the General Assembly of ISI during its 38th Session held in Washington in 1971 and this was unanimously accepted. We were asked to reconstitute ourselves as the new leadership of the formally yet to be created IASS: its Bureau, its Program Committee, its Nominating Committee and whatever else might be needed.

Had we been legalistic, we would have been facing an unsolvable problem. First of all, since there was no IASS, we lacked any legitimacy (we were unelected). Second, the Statutes clearly had to be approved by the members of the new Association, but since there was no Association until the Statutes were accepted (and until people enrolled in the new organisation), who would approve the draft statutes? We were aware of these legal niceties but not unduly worried about them. We received a number of time slots within the ISI program for sessions of the new IASS and we set about creating a program. Anders Christianson is quite right in remarking that the program in Vienna encompassed quite a broad range of topics. This was by conscious design: we wanted to establish a precedent for the broad scope of activities of the new organisation.

We also developed a list of people we wanted to nominate as the new (elected) leadership of the yet to be established IASS and we proceeded to sound out informally some leading lights of our profession.

During the first of “our” scientific meetings we set aside part of the available time for the first General Assembly of what became the IASS. Those present were asked to regard themselves as the founding members and they subsequently approved right then and there both the statutes and our proposed slate for the leadership of the new Association, under the presidency of Morris Hansen.

The rest, as they say, is history; except that in this case the entire narrative is that: the early history of IASS.

Ivan P. Fellegi