The Census under attack

In their continuing effort to give precedence to ideological considerations over evidence based decision making, the government has decided that for the 2011 Census, the census long form will be replaced by a voluntary survey covering the same content. Why? The point men on this issue are Tony Clement, the Minister currently responsible for Statistics Canada, and Maxime Bernier, who was in a previous life the minister responsible. Let's hear what they have to say. As usual with this government, the stated reasons evolve and change over time but the narrative seems to have settled on the following:

Clement tweet: "It's important to free Cdns from this unwarranted intrusion & coercion!"
Bernier bleat: "We're not there to please all special-interest groups; we're there for the silent majority of Canadians," Bernier said. Bernier has said the decision to go from a mandatory long census to a voluntary survey was a response to the "silent majority" of Canadians who don't like the intrusive, coercive nature of the process.
Intrusiveness:

The most sensitive question in the long form is the income question. It always gets the poorest response rate. Because of this, Statistics Canada asked in the last Census if it could use the respondents' income tax data instead. A large majority of Canadians agreed to this and so the most intrusive question on the long form can be handled this way, rather than by creating a new, voluntary survey. Problem solved. The next most intrusive questions on the Census are the list of persons living at this address, the relationship between these persons, their age and marital status, including common law and same sex. These are truly personal questions and in exchange for asking them, Statistics Canada guarantees that the responses will be held in the utmost confidentiality. All of these questions are on the short form, however, and will continue to be asked. So how has this decision dealt with the issue of intrusiveness?

Coercion:

There is no difference in response rates between the short form and the long form. Regardless of the form, 80% of Canadians fill it out promptly, most considering it to be a civic duty. This is the real silent majority. There is no better or more accurate poll than this. (Minister Clement says they are speaking to the 168,000 who did not respond to the last Census - is this Bernier's silent majority?) Getting the last 20% requires more reminders, more follow up. Only the very hard core refusals are escalated, with letters from the Chief Statistician, and then lawyers. A miniscule number of refusals are taken to court. The conviction rate is infinitesimal. The short form will still be mandatory. 100% of Canadians will still need to fill it out. There will still be follow ups for the 20% that are slower to respond, just as before. Hard core refusals will continue to be escalated, with letters from the Chief Statistician, and then lawyers. A miniscule number of refusals will still be taken to court. So how has this decision dealt with the issue of coercion?

"You try to limit the amount of state coercion that you have, you try to limit the intrusiveness of government activities, and that's the balance that we've struck."

STC conducts 378 surveys and statistical programs. Most are business surveys, or statistical programs that use existing data as their source. Less than 20 are regular household surveys that require Canadians to fill out questionnaires. Of these, only 2 are mandatory: the Labour Force Survey (a monthly sample survey used to produce the unemployment rate), and the Census of Population, held once every five years. The balance is already heavily tilted towards voluntary participation, with only the most critical programs retaining the mandatory designation.

Cost/Response burden/ paperwork reduction

There is no need to repeat the cost implications of this decision. Suffice it to say that this new approach will be more expensive than the traditional Census. That is truly unbelievable! Statistics Canada has been under intense pressure to reduce the cost of each Census since 1986, ever striving to produce more for less. Now, it will produce less for more! Lost in the cost issue is the increase in response and paperwork burden that Canadians will face because of this decision. One third of households will get this new questionnaire instead of the one fifth of households that would have received the long form. That represents an increase of over 50% in the response burden and paperwork burden placed on Canadians by the Census. Yes, it will be voluntary, but it will still end up in your mailbox, there will still be an expectation that it will be completed, it will still be the subject of follow up, in person or by phone. In fact, voluntary surveys require even more follow up calls to get satisfactory response rates than mandatory surveys. So instead of one fifth of households getting those annoying supper time phone calls asking them to complete their long form census questionnaire, one third of Canadian households will be subjected to the vigorous follow up required to get an acceptable response rate for the new voluntary questionnaire.

So how is this decision getting the government off the back of Canadians?

"That if we went to a voluntary census and if we did the measures that they recommended, that we could mitigate and/or eliminate the legitimate concern [about] going from a mandatory to a voluntary census."

Legality

The Minister cannot make the census voluntary. Under the Statistics Act, responding to all Statistics Canada surveys is mandatory. Refusing to do so can entail fines or imprisonment. However, section 8 of the Act allows the Minister (although in practice, it is delegated to the Chief Statistician of Canada), to make response voluntary, except for a census of population or agriculture.

So how can this decision be legal? To achieve his aim, the only way is to split the set of questions that for a century have been considered part of the Census of Population into a part called the census (the short form questions) and a new separate survey, containing the questions of the long form. He can then exercise his discretion under the Act to prescribe this new survey as voluntary. So just by calling what was the census a survey, he can get around the obligation under the Act for mandatory response to these questions. But then, just what exactly is the Census? Has not a century of practice in Canada established a precedent regarding the broad content categories that it includes? Is there a basis for a legal challenge here?

So what will this change have achieved? It costs more, increases the response burden of Canadians, degrades the data, destroys the time series, hardly reduces intrusiveness and is still coercive. Huh? I'm not getting it.