Science and Technology
Aspiring poet turned statistician counts on reason over rhyme
Ivan Fellegi, Chief Statistician of Canada 1985-2008, chose the practical realm, mathematics, over his passion for poetry and literature when confronted with an academic crossroads in his native Hungary over 50 years ago.
The biggest decision in Ivan Fellegi's life was to pick reason over rhyme.
As a student in his native Hungary, Canada's chief statistician for the past 22 years was passionate about poetry and literature. But in order to win a place in a university under the Communist regime then, he entered a national academic competition. Placing within the top five would guarantee him a spot.
"My teachers assumed I'd enter in the literature category because I liked to write poetry. But I decided to focus on the one area where there could be no dispute from the judges, no subjectivity," recalls the head of Statistics Canada, who is now 72. "I was good in math, but not exceptional."
He placed third.
Before Dr. Fellegi completed his studies in mathematics, however, the Hungarian Revolution took place and he fled his home for Canada. Arriving in January 1957, it was too late for him to enrol in university for that academic year, so he took a temporary job at the Dominion Bureau of Statistics. (It became Statistics Canada in 1971.)
While working at the Bureau, he finished his bachelor's degree and subsequently pursued both a master's and a doctorate through night courses, earning the first post-graduate degrees from Carleton University.
Fifty years later, he's still there, at the helm of an organization that is not only recognized as one of the best statistical agencies in the world, but has become a fundamental, organic part of shaping public policy and business strategy in Canada.
While there have been profound changes in the way data are gathered, analysed and processed since he began his career, Dr. Fellegi says that perhaps the most important element in the evolution of the agency has been its success in establishing a level of credibility that has allowed its data to transcend partisan politics.
"In the 1960s, the Opposition would use one set of numbers to debate something like unemployment in Parliament and the government would use another set," explains Dr. Fellegi. "We've reached the point now where we are viewed as trusted brokers and our numbers are accepted by all politicians. Which isn't to say they don't dispute what to make of them."
The importance of that impartial status is reflected, for example, in the equalization payments from the federal government to the provinces.
"The payments are based on a negotiated formula, but it's applied blindly based on our statistics," he explains. "The moment the numbers are doubted, the whole system falls apart."
(By contrast, in the United States there is no centralized statistical gathering or analysis. Dr. Fellegi spearheaded a group that had a mandate to recommend changes to that system under president Jimmy Carter. Although the initiative died with the Carter presidency, Dr. Fellegi says problems related to lack of consistent standards are still a problem when calculating key statistics like GDP growth or productivity.)
That commitment to scrupulous objectivity is also reflected in the design of the questions StatsCan asks -- especially when it comes to the ever-more personal and sensitive matters addressed in a census. Privacy is also an issue of increasing concern and complexity.
Directly involved in the national census since 1961, Dr. Fellegi notes that the 2006 census was the first to include some Internet responses. The next one in 2011 will be the first to offer an Internet response option to everyone. But the introduction of that new platform also requires a careful recalibration of the questions to suit it.
"When presenting the questions differently in a physical sense, we have to be sure to avoid any bias based on the medium," notes Dr. Fellegi. "It's both a science and an art."
He adds that a "huge effort" goes into framing and testing questions to "get as close to the truth as possible," a process that takes about four years of consultation and testing. Statistics Canada also solicits input from the public and the deadline for content suggestions for the next census is at the end of this month.
Among the reports still to come from the 2006 census results are language, immigration, mobility and migration in December, labour, workplace, commuting and education in March 2008 and in May 2008, income and earnings, housing and shelter costs.
To prevent bias from creeping into the design of questions, Dr. Fellegi has deliberately fostered an internal culture of questioning, as well as an elaborate system of grooming staffers who come from such disciplines as sociology, information technology, economics and mathematics.
While there is a roster of experts in very specific areas such as price indexing, seasonal adjustment or productivity analysis, the generalists have carefully-managed careers.
Upon joining the department, recruits are assigned a senior mentor and, over two years, get three rotations through the department. When that's done, they're given a job where they're expected to spend about five years.
The next step in promotion is to move into a new area where, says Dr. Fellegi, they have to apply their experience in new areas and in new ways.
Having worked in the agency for so long, he insists the job never gets stale -- and that he's continually surprised by the insights the statistics yield.
With the latest data gathered, he says he was surprised by just how well Canada is doing at integrating -- and educating -- immigrants. Within two years of moving to this country, 8o per cent of immigrants identify themselves as Canadians. And in many ethnic groups, 60 to 65 per cent of second generation immigrants complete university.
"It's the most interesting job in Canada," he insists. "I haven't once been bored."
And Hungarian poetry's loss is Canadian statistics gain.