On September 21, 2021, Canadian statistics lost an emblematic figure, Constance van Eeden, une grande dame de la statistique. Constance has made an enormous contribution to statistical science as an educator and researcher, through a very active career that spanned more than half a century.
Constance was born in Delft, the Netherlands, on April 6, 1927; the family spent the trying war years in Bergen op Zoom. A happy memory was the day when the town was liberated by the Canadian troops, in October 1944.
Constance received her BSc in 1949 from the University of Amsterdam, and worked on her MSc and PhD while being employed by the Mathematical Center (now CWI) in Amsterdam. She earned her PhD cum laude in 1958 as one of the four PhD students of the renowned mathematician David van Dantzig. Her thesis topic stemmed from consulting work under the supervision of Jan Hemelrijk.
In 1960 Constance was invited by Hermann Rubin to Michigan State University where she met her husband, the gifted statistician Charles Kraft, who died untimely in 1985. After a short stay at the University of Minnesota, van Eeden and Kraft moved to Montréal in 1965, invited by several people, in particular Jacques St-Pierre and Maurice L’Abbé. The arrival of Kraft and van Eeden gave a big impetus to scientific statistics in Québec.
Constance’s career as a researcher spanned over 50 years, with the years after retirement as active as ever. Her main coauthors were Charles Kraft and Jim Zidek, and she contributed more than 70 peer-reviewed papers and two books.
Her earliest contribution, in the 1950s, revealed her great intellectual courage, mathematical skill, and originality. For, in her PhD thesis, she proposed a whole new subarea of statistical science, namely inference for restricted parameter spaces. (Dan Brunk proposed the same thing independently at about the same time.)
Its full impact can be truly appreciated by considering its context, a time when Fisherian statistics was in its full glory and Wald’s theory of decision was in its ascendance. Both paradigms made the unrealistic assumption of an unrestricted parameter space, while Constance argued for a theory that embraced restrictions.
Being well ahead of its time, development of the field of restricted parameter spaces remained quite dormant until Casella and Strawderman published their paper on this topic in the early 1980s. Their paper stimulated an interest that continues unabated. Moreover, throughout, Constance contributed to a strong research program in this area; in 2006 she published a Springer monograph on estimation in restricted-parameter-space. The importance of this field cannot be overestimated as statistical procedures that exploit parameter restrictions can achieve huge improvements in performance.
In another direction, Constance, often in collaboration with Charles Kraft, produced pioneering work in nonparametric statistical inference. Their many seminal contributions were published in first-class journals, like the Annals of (Mathematical) Statistics and JASA. For many years, “Kraft and van Eeden” were household names in that discipline. Their visibility was likely enhanced by their publication of a highly innovative book on that subject in 1968. Actually, Constance kept publishing in the field beyond 2000, while her first papers in nonparametrics date from 1958; her early contributions are quoted in Hájek-Šidák's classic book on rank tests.
Her reputation as the “wizard of nonparametrics” was reinforced by the fact that many of her students and their students (her ‘‘children'' and ‘‘grandchildren’’ in her own words) worked in nonparametric testing and estimation; as such, she is the founder of the Québec school in nonparametrics.
As for her contributions in other fields we mention her long-term collaboration with Jim Zidek that stemmed from his work on group Bayesian theory, where one cannot restrict oneself to the posterior case. However, in the pre-posterior case the role of the unknown parameter in Wald’s theory is played by hyperparameters that index either the prior or the utility function. Since in that context, group members would be expected to share general knowledge, hyperparameter restrictions are unavoidable. This inspired Jim to engage Constance in what turned out to be a very productive collaboration. Some of it has spilled over to the topic of weighted likelihood estimation, where restricted parameter spaces also arise. And in a totally new direction, came their work on Bayesian subset selection.
All her collaborators appreciated her extremely high standards and her keen sense of precision. Most and foremost, she would never rush into publishing: drafts of papers had to sit in a drawer for a while. Working with Constance, Jim learned firsthand of her passion for research, her originality, and her love of the discipline.
Constance’s sharpness as a reviewer was proverbial; she has been associate editor of the Annals of Statistics (1974–1977), the Canadian Journal of Statistics (1980–1994), and Annales des Sciences Mathématiques du Québec (1986–1998). Finally, her long appointment as an editor for the ISI’s abstracting journal (1994–2004) is another of her major contributions to our discipline (rewarded in 1999 with the ‘‘Henri Willem Methorst Medal’’). She helped maintain its high standards while being instrumental in having it put on an efficient electronic platform.
Constance was also a great educator. Those who took a course with Constance van Eeden remember her lectures, clear and easy to follow, the result of very careful preparation. Some of these former students are still referring to her course notes, no matter their present specialization (mathematics, computer science, statistics). Constance’s talent as a teacher explains why so many students wanted to pursue graduate studies under her supervision. In all she had 14 PhD and 19 MSc students. She was an exemplary and quite unique advisor and mentor, who instilled in her students her professionalism and sense of rigour. Actually, in Canada, her PhD students held academic positions from Vancouver to St. John's, Newfoundland, in compliance with Canada's motto: A Mari usque ad Mare!
Constance chose to retire in 1988 and became professeure émérite at Université de Montréal and professeure associée at UQAM. Besides, 1989 marked the beginning of her long-term involvement with UBC: appointed as an adjunct, her position was soon converted to Honorary Professor of Statistics. That appointment of an icon in statistical science gave the stature of the fledgling department a big boost and gave Constance another academic home away from home, as she visited UBC every year through 2012. Moreover, Constance created an endowment for the department, which funds a variety of programs for visitors and graduate students.
Her extraordinary career as a researcher, educator, and contributor were well acknowledged and she received numerous honours. In recognition of her merits, Université de Montréal established, in 1998, the ‘‘prix Constance van Eeden’’ to be awarded to the best BSc graduate in statistics. She was awarded an SSC Gold Medal (1990), fellowships in the IMS (1972), the ASA (1972), and an elected membership in the ISI (1979). At the time of her passing, she was an honorary member of both the Canadian and the Dutch statistical societies.
During her 75th birthday fest at the CRM in May 2002, there were numerous references to the generosity and patience of “Madame van Eeden.” Whether in Montréal or Vancouver, where she made herself a member of the academic family, she was in her office punctually every day, always willing to listen both to faculty and students and to give advice where appropriate. She insisted on having lunch with faculty and students, would not miss a holiday party. This was Constance the person and she was much loved.
Constance was always praiseful of the younger lot, her students, her daughter Kari, her stepchildren (Kraft's daughters and son). She viewed herself as a problem solver: two of her favourite pastimes were crossword puzzles and knitting. But in order to know Constance, one had to visit the place of her retirement, the bucolic village Broek in Waterland, north of Amsterdam. Only 15 minutes by bus from the busy Amsterdam Centraal, Broek lives at a slower pace, surrounded by green pastures where cows graze.
Both Constance the person and Constance the icon will be greatly missed.