The politics of people-counting Question on ethnicity haunts census-takers

 By Lynda Hurst
Toronto Star Feature Writer
January 22, 2000

Every five years the census-takers come to find out who we are, and every five years they risk stepping into a minefield.

Alarms already are being sounded over next year's census.

In 1991, when people were asked to give their ethnic origin, they were provided for the first time with a checklist of 15 options (French, English, German, and so on), followed by a blank space. Canadian wasn't listed.

The missing option set off a howl of public indignation.

People whose families have been here for generations resented being asked to come up with a long-lost ethnic' heritage. Three per cent wrote Canadian in the empty space.

Ottawa got the message. In 1996, for the 20 per cent of the population who got the long form containing the question, Statistics Canada dropped the check-off options. Instead, Canadian was fifth in a list of 24 example groups. People were asked to specify as many groups as applicable.

The percentage of those claiming Canadian as their ethnic origin jumped to 19 per cent. The plan for the 2001 census is to list it as the first example, or prompt.

Problem solved. Everybody happy?  Far from it.

If StatsCan had its way, the question would be dropped as demographically tainted. The agency proposed that last summer. Not possible, countered the department of heritage, Ottawa's purveyors of multiculturalism policy and funds. For the past six months, the two have been locked in a tug-of-war. 

Official multiculturalism has been enshrined for 29 years. Just as the federal Employment Equity Act requires the collection of statistical data on visible minorities (in a different census question), so, too, does multicultural policy require it on ethnic groups. The question will stay, but perhaps in a different form.

There are several different formulas we're looking at, says Norman Moyer, the heritage department's assistant deputy minister for Canadian identity. It's a tricky question, but I think we can come up with a deal with StatsCan in the next week or so.

Doug Norris, director of StatsCan's division of housing, family and social sciences, says the agency has provided a number of alternatives. The only one he will discuss is replacing ethnic origin with place of birth of parents or grandparents. Norris knows it's problematic, because country of birth and ethnic origin aren't necessarily the same thing.

What we are trying to get is reliable, meaningful data, he says. Is the question doomed or can we just change it in some way? Whatever we end up with has to get cabinet approval.

When StatsCan did a year-long consultation on the question after the last census, many sociologists and demographers attacked it on several fronts. Chief among them was the fact that Canadian isn't an ethnic origin, it's a self-identity. Listing it as an option defeats the whole purpose of a question designed to measure the different backgrounds of the population.

Ottawa has been asking for ethnic origin in one form or another since 1871. To throw in a non-ethnic prompt renders the modern data useless, critics say, because it can't be compared with the old.

And putting it as the first prompt means even more people will choose it, says Peter Li, a University of Saskatchewan sociologist.

Norris agrees that the Canadian listing confuses people. It muddies the water. They aren't sure if they're being asked about their ethnic background or about how they see themselves.

StatsCan may be in search of consistent, precise data, but the agency is not calling the final shots. With the census, StatsCan is a servant of government departments which are, in turn, subject to public pressures.

It's a tough position to be in, says David Andrews, a University of Toronto statistics professor who has advised StatsCan in the past.

It may be useful for (Heritage Minister) Sheila Copps to know whether 5 or 85 per cent of people identify themselves as Canadians and I can see the argument for including it. But it has an effect many would like to avoid. To demographers, it's useless.

Some of them want the question out altogether; others would settle for ditching the Canadian option.

But that's no mean feat, says Ellen Gee, chair of sociology at Simon Fraser University - not when a question is politically sensitive.

Data collection is politics, she says. When the federal government approves the questions, they are one and the same thing.

Gee, who has served on StatsCan's advisory committee, says she would at least separate the Canadian question from the background query.

I'd favour a second question after ethnic origin that asks if you self-identify as a Canadian. Having both would give a true picture of the evolution of the country.

Gee fervently hopes place of birth isn't substituted. That information might be useful, she says, but only in addition to - not as a replacement for - ethnic derivation. 

It's not a proxy, not these days with global transmigration. You may have an East Indian born in Tanzania who moved to London and then came to Canada. Place of birth isn't telling you anything.

Moreover, in the 20th century, nations came in and out of existence. For many people, where you were born depended on when you were born.

Other critics say that all the example answers - not just Canadian - should be removed, and the question asked without prompts. StatsCan tested that, however, and found people didn't understand what was meant by: To which ethnic/cultural group did your ancestors belong?

We have to give examples, says Norris. In 1996, almost all of those who wrote in Canadian'' turned out to be of French or British background. Next year, if the prompt stays, it's a safe bet many more people of other ethnicities will do the same. 

One suggestion is to preface the question with a clear definition of what's being sought and why. People might be less likely to see it as an insult to their Canadian identity. 

Or maybe not, says Knight: There is a danger that precisely because of the preamble, people will fiddle about with the answer out of spite. 

Another problem is the lack of a time frame in asking to which group a person's ancestors belonged. Is someone's vague memory of a German great-grandfather relevant when more immediate forebears were Italian? 

People don't know how far back they're supposed to go, says Peter Li, or they simply don't know their backgrounds. We should just ask the ethnic origin of mother and father.

There's yet another flaw in the current version of the question, he adds. Asking people to specify as many groups as applicable encourages them to give multiple answers: Wittingly or not, the phrasing elicits a certain response.

Four answer boxes were supplied in 1996, presumably to cover four grandparents. According to StatsCan, however, some people wrote in more.

It's okay if a few do that, Li says, but if a vast majority starts giving multiple answers, it destroys the question. How do you analyze the response?

Most sociologists and demographers want the question to remain. They need the information it provides to examine who Canadians - in the everyday sense - are. It's used by governments at all levels to plan programs; by medical researchers, to trace genetic diseases; by educators, to monitor language-training needs; and by ethnocultural groups.

Others may ask why Canada is still checking into ethnicity when it's the racial makeup of the country that has changed so dramatically in recent years. Li's answer is that visible minorities account for only 11.7 per cent of the population; the majority is still white, of British, French or other European background. 

We've been tracking that majority for over 100 years, he says, and for historical continuity, we must go on.

But in what form - and with what results - is still anyone's guess.