Seppo S. Laaksonen, ISI Elected member and Professor Emeritus of Social Statistics in the Faculty of Social Sciences at the University of Helsinki, passed away on 20 December 2020.
SSA Chair of the Official Statistics Section, Stephen Horn, reflects on an Australian and a personal link to Laaksonen, which may be of interest to the Official Statistics community.
In 1993 I travelled to the 4th international workshop on survey nonresponse - a rather informal gathering, then of about 30 people organised by a core of north Atlantic methods people, in the wake of the 1982 multivolume review of response in survey by Hansen, Hurwitz and Madow and the alarming collapse in response rates in represented agencies. I used this rare opportunity to seek out experience on weighting, particularly for periodic household expenditure surveys, notorious for layers of missed data. Seppo, then working for Statistics Finland, had recently completed his thesis on the topic with an offprint reaching me in Canberra.
This motivated a whistle stop call on him and colleague Kari Djerf at SF. We crossed paths a few times in next years at the increasingly grand successors to the workshop, and the increasing interest at European level in survey inference under missingness. Seppo attended the 2006 ISI gathering hosted by ABS in Sydney and agreed to travel to Canberra afterwards to give a seminar at my then workplace the Department of Family and Community Services - host of a stable of longitudinal surveys exploring the social condition of Australians.
The subject of his talk was imputation in panel surveys, bringing up to date his thinking on survey adjustment, and timely for us as we were sorting out quality issues with HILDA. Statistics Finland had been collaborating with the Finnish DFaCS equivalent on welfare series and inference binding survey results with administrative data. Their experience in the mid-1990s would come in handy with the big push for harmonised statistics for the European Union over the next decade, and the launch of a European panel on income.
The informal workshop on nonresponse had evolved in the meantime to a full-blown European Conference on Quality in Official Statistics; with the 2010 round being held in Helsinki. This pooled the gamut of methods challenges of individual agencies, transforming an internal concern with declining response to household surveys into a continental push for new techniques and technologies for collecting information from businesses and households. Agencies became incubators for both practice and theoretical advance, with some levening from academics. By the end of the decade small teams from across Europe, drawn from agencies and universities were commissioned to reinvigorate statistics collection. It was an exhilarating prospect from an agency perspective, given the wide remit, resources and determination. We may not yet have seen its full flowering, although I suspect that the current fascination with data analysis and statistical integration at the expense of a good inferential base may spell the end of that wave.
Seppo, like me, had a career in the shadow of ‘The yellow book'.
Särndal, Swensson and Wretman's text Model Assisted Survey Sampling, appearing in 1992 was instrumental in the long reforms in survey design and adjustment. Li-Chun Zhang has been invited to give this year’s Foreman Lecture at the 2021 ANZSC. This will be an important marker in the next chapter for survey inference applied to official collections. I look forward to Professor Zhang's lecture. Like Seppo Laaksonen Li-Chun is a hybrid, with a solid agency background (Statistics Norway) and dedication to the foundations of inference through an academic career. Seppo's 2006 Canberra seminar, a friendly gesture at the time, reminds me of the peculiar blend among agency methods people of borderless deep thinking, and pragmatic application.
It is a small and loose convocation, but no signs of diminishing; it was personified for me by Seppo Laaksonen.
Chair, SSA Official Statistics Section