Tony Clement makes a statement - perhaps not the one he intended.

There have been occasions over the past couple of weeks where I've regretted the somewhat combative tone of my first census post. One of those times was this morning, reading Bill Robson's mild rebuke of the decision to make the long form voluntary. He suggests that critics might do better to try to engage the government, and he makes a convincing case.

But then I read the statement released yesterday by the Minister of Industry, and I despaired:

To meet the need for additional information, and to respect the privacy wishes of Canadians, the government has introduced the voluntary National Household Survey (NHS). 

Statistics Canada will conduct and release this survey, applying the same standards used for its surveys and past censuses.

StatsCan will most certainly not apply the same standards used for its surveys and past censuses, for the simple reason that the census data will be generated using an entirely different sampling methodology. Worse, it turns out that even after two weeks of public debate, in which the issue of sample selection has been spelled out by many people in many fora, the federal government still doesn't acknowledge the problem.

To promote data accuracy, this voluntary survey will be sent to a larger cross-section of households than the old long-form census.

I must confess that my immediate response to this was not one of forbearance. The problem of a biased sampling methodology cannot be fixed by sending out more voluntary questionnaires: all you get is a bigger, biased sample. What's particularly annoying is that Minister Clement is well aware of this point, because I explained it to him personally.

This decision was made without consulting anyone who understands statistics. After two weeks of criticism by everyone who does, the government's strategy is not to explain why the material we teach in statistics courses is wrong. Instead, it chooses to pretend that the material we teach in statistics courses doesn't exist.

I'd very much like to take Bill Robson's advice, but what point is there in appealing to logic and reason if the government is manifestly immune to such arguments?

Comments

[Jonathan]

It would be nice if they would just come and say, "we don't care about selection bias - we have other priorities." At least it'd be honest.

[Kevin Milligan]

Stephen,

I thought your twitter exchange with the Minister was wonderful. Calm, forceful, and factual.

In my view, we'll get this reversed not by convincing the Minister of the merits of our argument, but by convincing voters. I don't think setting your hair on fire is very convincing to voters--histrionics likely turns many off, as it feeds into stereotypes about academics.

[Jim Rootham]

At this point voters are too late. Go to the opposition, get them to move in Parliament. I think they may already be doing so.

[edeast]

It's too bad about minister clement being purposefully obstinate. He was better on copyright then james moore. I still feel this is someone else in cabinet's pressure. But then if clement can't stand up for good policy he doesn't deserve his responsibility. This is a rude example of politics.

I was hoping he and Chong would be around to rebuild when Harper gets knocked off.

Since he is doubling down, I assume they think, his seat is safe, and they're doing if for some base elsewhere? He's my mp. A lot of money has come in,but I know there are weak points.

Your first post was too harsh. Sun Tzu art of war, make sure your opponent has a path for retreat. You had started to build a rapport, with the copyright question, and complement, then f----- him with drunken monkey blog post.

So ya, either he concedes or going to have to kill his career... or a myriad of other outcomes: binary choices are more dramatic.

[Kelsey]

What Tony Clement is not saying is that the CONS are concerned about the PRIVACY of their kin and kith and of course themselves. The Cons do not have Privacy Concerns of Canadians at heart.

Stop Reasoning with people who have only personal Interests. This is not about Census at all, it is just that with the Mandatory Long Form Census Scrapped these people and their close ones will have closed themselves and will not figure in the Census at all. Not hard to figure out why they are worried about their own issues - have they done something they are Afraid of? Are they seeking to protect their own future and of their kids?

[Mike Moffatt]

Speaking of people who are missing the point, I'm really enjoying the twitter feed of the Fraser Institute.

[Frances Woolley]

Does anyone else think that perhaps this is all just some really subtle plot to eliminate a whole lot of questions that the Conservatives aren't particularly keen about, especially ones used for programs such as employment equity, from the census?

So the next thing we'll hear is "the long form is back but to lessen the burden we'll scrap X and Y question"

[Simon van Norden]

I'm guessing Tony Clement relies on advisors to decide on the statistical disadvantages of voluntary surveys. If he's tweeting back answers about reweighting, I'd further guess that sample-selection bias was well-discussed prior to his decision and someone assured him that there was a reasonable workaround.

Ministers have to rely on the advice of technical experts. The best way to change their minds is to change the minds of their advisors. Finding out who assured the minister that sample selection was not a problem is difficult; but change their minds and the minister will follow suit. Until then, the minister will just be frustrated that "our message is not getting out."

[Patrick]

I think you did the right thing. Don't worry too much about having come on too strong. If a tenured prof can't call a spade a spade, then who can? Isn't that kinda you're role? I wish more smart people would call BS when they see it.

The way to fight this is for the opposition or an enterprising journalist to figure out what is really motivating the conservatives. So far, they've offered nothing but the privacy red herring to divert the argument - "Favour the long form? You hate privacy!". Keep in mind that the order came out of cabinet, which probably means the policy originated in the PMO. Why? Why would the cabinet or the PMO mess around with the census? What are they trying to accomplish? Who stands to gain? Who stands to loose? I suspect it's something they don't want to reveal, and revealing it will likely torpedo the policy.

[Adam P]

"If a tenured prof can't call a spade a spade, then who can?"

Presumably that's why tenure exists in the first place.

[crf]

The Conservatives (and Clement) have a history of standing tall against logic and reason and the experts who employ those tools ... and "winning".

Recall this?

[SirWilfridL]

"Ministers have to rely on the advice of technical experts. The best way to change their minds is to change the minds of their advisors. Finding out who assured the minister that sample selection was not a problem is difficult; but change their minds and the minister will follow suit. Until then, the minister will just be frustrated that 'our message is not getting out.'"

To be honest, I'm pretty sure this decsion went the other way. That is, Clement got his marching orders from the PMO, and Clement went to his deputy minister and advisers and said, "We're getting rid of the mandatory long form, how can I justify this?"

So basically, we should be screaming at the people in the PMO who are determined to appeal to the libertarian wing of the part with this dumb stunt. Maybe one of them will listen. Or not.

[BillBell]

I wouldn't worry. Judging by the thicknesses of their skins, and of their skulls, the real Rhinoceros Party is already in power in Ottawa.

[learning james]

"what point is there in appealing to logic and reason if the government is manifestly immune to such arguments?"

Always keep fighting for what you believe in.

"There have been occasions over the past couple of weeks that I've regretted the somewhat combative tone" Please, this is your house!

Be as forceful as your passion dictates. Isn't that the point of a blog?

[Peter]

If you accept the axiom "there is too much government", then any action which reduces the efficiency of government is good. It's not sabotage or incompetence, it's policy.

[edeast]

To confirm my intuition. I suppose you've probably already seen this, but Siddiqui has a source who says Clement and Flaherty originally opposed it, before being overruled by Harper. But the point still stands, don't deserve to be a minister of the crown if you can't stand up to the first minister.

[edeast]

My reasoning being, if you can't out argue someone staffed with interns, idealogues and fanboys, when you have a government department at your disposal... The pmo must be unreasonably powerful, they overruled him on the Copyright bill as well.

[Hal Jam]

@ Kevin Milligan

"I don't think setting your hair on fire is very convincing to voters--histrionics likely turns many off, as it feeds into stereotypes about academics."

Exactly what this non-academic has been commenting since that first census post ("An incredibly stupid decision on the 2011 census"): the arguments presented hereabouts came across as far too self-serving and detached from valid real world concerns ... which is not a criticism of the actual motivation or value of said arguments.

As hard as it may be to believe, one CAN accept the need for proper sampling techniques and the value of the data obtained by the long-form census ... while ALSO finding the long-form excessively intrusive, having legitimate concerns about privacy (where local people are used to verify data), and resenting the coercion applied by the state in the face of those concerns.

And one can also dislike the long-form census while having no truck with the Harper government or their friends an allies ...

[edeast]

He's really digging in, exception to farmers. In my opinion the liberals have the best chance at taking his seat. I wish I new what the best party in calgary southwest would be.

[Another civil servant]

If I was ordered to go out and defend a decision I disagree with and strongly believe to be a bad one, maybe I'd be a good trooper loyal to her chief/team and do my best to convince people of the soundness of the decision. Or maybe I would *look* like I'm doing my best to defend the decision, but in fact sabotage its defense by using weak or stupid arguments, making technical mistakes, providing ammunition to opponents - in the hope the decision is overturned. Sure, I'd look like an idiot in the process, but 1) this might be less painful than resigning and 2) it actually improve the chances that the end result is what I wanted from the beginning.

In other words, having read a lot of what has been published in the last few days, I'm starting to wonder if Clement isn't smarter than it appears in how he handles this.

[Wendy]

Starting to agree with "Another Civil Servant" -- but with a twist. Clement may be the Conservatives' way out of this. He's sounded tone deaf and ignorant during this discussion, so the PM could blame the whole mess on him, ask for his resignation, and then back out of the whole census form change thing. We haven't heard a peep out of the PM on this.

Then at the next cabinent shuffle, Clement can be rewarded for his service here with a prominent new role.

PM could also say that Clement jumped the gun, and that a committee will discuss modification to the census act or the long form to make it less intrusive. A face save.

[Mark Szekely]

Solid decision making demands reliable data.

The Long Form Census questionnaire is Statistics Canada's "flagship" data product, and is relied on extensively by economists, policy researchers, historians, and a broad range of other professionals who assist Canada's decision-makers.

The only "winners" from this decision are the paranoids who hold the belief that their personal information is being jeopardized. However there is absolutely NO evidence to support such a contention.

The fact that this change was made with absolutely no consultation or advance warning is even more disturbing.