“The big thing is loss of information,” Prof. Woolley told The Hill Times. “As this policy [a non-mandatory survey] continues, we’re going to be getting further and further away from a point when we ever did have good information about what society looked like. The effect of bad information is cumulative, and it shows up in all kinds of policies.”
She said that making the census voluntary skews data because there are simply just certain people who will respond more than others, therefore a truly accurate picture cannot be formed. “It’s like anything else, whether it’s picking up dog poop in the park or helping you out if you need somebody to help you move: there are some people who are more likely to volunteer than others, those people are going to be different in other ways,” Prof. Woolley said. “If you see something and say, ‘Oh look, there’s more people going to church or fewer people going to church than we expected,’ well you don’t know whether that’s a real trend or whether it’s just some people were more likely to fill out the census than others.”
While the short-form census, a basic 10-question survey, remains mandatory in Canada, in 2010 Prime Minister Stephen Harper’s (Calgary Southwest, Alta.) Conservative government scrapped the mandatory long-form census and replaced it with a similar but voluntary survey. At the time, then-Industry minister Tony Clement (Parry Sound-Muskoka, Ont.) said scrapping the mandatory long-form was a question of privacy. “My position is we are standing on the side of those Canadians who have an objection to divulging very personal information to an arm of government and are subsequently threatened with jail time when they do not do so. So we are, in my view, speaking to those Canadians while at the same time doing everything in our power to ensure that the results that are received from the long-form questionnaire are valid and defensible,” he told the Globe and Mail in 2010.
Statistics Canada was allocated $30-million more in 2011 than in 2006 in order to conduct the voluntary National Household Survey because the agency sought to offset its lack of mandatory status by increasing the population sample that would receive it. In all, Statistics Canada used $22-million of those additional funds, and the 2011 census program cost a total of $652-million.
Previously, the mandatory long form census was sent out to about one-fifth of Canadian households, but the 2011 voluntary national household survey was sent out to about one third of households in Canada instead. So while the 68.6 per cent response rate to this year’s voluntary survey is lower than the 94 per cent response rate in 2006, Statistics Canada actually overall received 211,430 more household responses this year. However, because responses were voluntary, there was no way to ensure equal distribution or representative sample of responses, meaning it could be more or less information available for different pockets of the country.
“Although the response rate was lower we started with a larger sample so for all the processing operations we basically had a very similar amount of work to do. The content of the questionnaire was also basically the same as the census long form,” said Mark Hamel, director general of the census management office. “The part that was different is the portion that we call estimation. In estimation what we do is we basically inflate [or weight] the responses from the sample to represent the entire Canadian population. … When you have a smaller response rate this task is increased in complexity, and that’s what we’re looking at in 2011.”
Mr. Hamel said Statistics Canada is able to use the findings from the mandatory short form census to weight the responses from the voluntary long form census.
Statistics Canada itself has cautioned users over the quality of the findings of the 2011 voluntary national household survey when it comes to smaller communities in Canada. Mr. Hamel said there is “more volatility in the estimates” when it comes to smaller areas.
“The smaller you get the more risky it might be to produce very reliable results, but the prevalence at which it happened in 2011 was higher than in 2006, [there were] more communities for which we did not release [information],” said Mr. Hamel.
“We conduct extensive evaluation of the [national household survey] data before we put it out. … What we’ve found in our outputs up to this point is the data at higher levels of geography, so I’m talking provinces, larger municipalities, and so on, is of very good quality. It aligns very well pretty much with all other sources that we can look at. There were some little tidbits here and there where there are small issues, those have been documented in our reference material,” he said.
For example, Mr. Hamel said the level of people who reported the Philippines as their country of origin “seemed a bit overrepresented in their result compared with immigration databases,” and Statistics Canada documented this observation when the information was released.
“When you go to lower levels of geography … we have released information for less communities than we have in the past, than in 2006, about 1,100 less, and this is a direct effect of lower response, basically,” he said.
These were communities where Statistics Canada didn’t receive enough responses and Mr. Hamel said the agency deemed the information to not be of “good enough quality for release.”
The third and final release on the findings of the 2011 National Household Survey related to income, earnings and housing and shelter costs was set to come out on Aug. 14, but has now been delayed until Sept. 11 after a “last minute discovery” of an error that was spotted in the results’ data processing. Mr. Hamel said the final data set will for sure be published on Sept. 11.
Claude Dauphin, president of the Federation of Canadian Municipalities who is also borough mayor for Lachine in Montreal, Que., said the results of the 2011 voluntary census were “less reliable” for rural Canada.
“In practice, if it cannot be reliable as it used to be, it means that municipalities across the country, it means that FCM, have to organize themselves in another way,” he said. “We’re using Statistics Canada to know exactly how to serve our citizens, where to put, for example, bus routes, where to install more affordable housing, where to provide new programs for new Canadians and all that.”
The FCM previously spoke out against the government’s plan to make the mandatory long form census voluntary. Mr. Dauphin said now the FCM wants to sit down to discuss with the government what alternatives there are and what else can be done if the information coming from Statistics Canada on rural Canada is not reliable.
“Municipalities rely on Stats Canada to understand the changing needs of our communities and to make a wide range of decisions,” said Mr. Dauphin.
Prof. Woolley said she’s still waiting to get her hands on the micro-data from the 2011 census to make an ultimate judgment, but said overall the information is “not as good” and “there has been a loss of quality.”
“When you start feeling the loss in quality it’s when you start really drilling down to specific population subgroups,” said Prof. Woolley, adding that in general, people just trust the voluntary and more weighted census results less.
“One thing I was looking at with the previous releases was the number of people who have migrated from Newfoundland to Alberta and the number of young people who have migrated. I was interested, ‘Are there more men or women migrating from Newfoundland to Alberta?’ I mean, that’s an interesting question. When you look at the numbers, it’s just kind of weird. You think, ‘what’s really going on,’ because it looks like there are more women leaving to Alberta, and that seemed weird to me. So you get something that’s weird and you don’t know, ‘is it really something that’s unexpected here, or is it just the data?’” she said.
Prof. Woolley said some information from 2011 is better than others. Findings on language, for example, she said are “still pretty good,” as is “stuff on family composition,” because such questions are included as part of the mandatory short-form census, which only covers household size, sex, age, marital status, and language.
And while Statistics Canada said it weights information, Prof. Woolley pointed out that “you can only weigh on what you can observe, and the things you can observe are the things in the short form census,” which doesn’t include, for example, whether someone is a recent immigrant to Canada.
Prof. Woolley said in Europe, there are examples of countries that don’t have censuses where the government has instead created an administrative dataset about its population by linking administrative records such as property tax records or income tax records, something she said provides better data than a census. Canada has “decided as a country that we don’t want to go down that route,” she said.
In terms of alternatives outside of government, Prof. Woolley said while companies like Google and Facebook obviously collect and possess data on people, it raises the question of whether people really want those corporations to be collecting, and to be subsequently encouraged to collect personal information on users.
“It’s not obvious that this [voluntary census] is better. It’s not obvious that it’s better in terms of respecting people’s privacy. It’s not obvious that it’s better in providing the information that people want, and it’s not obvious that it’s better in terms of cost,” said Prof. Woolley, noting that when there is a change in government, she hopes the mandatory long-form census returns. “I think it has done irreparable harm because it has politicized something—I mean we’ve had censuses in Canada for over 100 years. … We’ve taken something that was just something people did as their contribution to knowing about this country, as building knowledge. They could fill out the census form and leave a historical record for their children or grandchildren, and we’ve politicized that.”
Original Article
Source: hilltimes.com/
Author: LAURA RYCKEWAERT
No comments:
Post a Comment