Thursday, August 12, 2010

Issue Management

In all the pages and pages of Census e-mails and documents released Tuesday, the excerpt above is probably the most telling. Just 2 days before the Census changes were made public, an internal Communications Plan was circulated proclaiming "These changes will not have a negative impact on the quality of the Census. Response rates and data quality will be comparable to previous censuses."

In the margin was a lone voice of reason - "Really? Won't some stakeholders argue the opposition?".

I'd say that's the understatement of the year.

This document reveals so much about what went wrong for the government and how one of the unlikeliest fiascos in Canadian political history was conceived. Just two days before the move was announced, they solicited "initial comments" on the plan...and it occurred to the government some people might oppose it.

Presumably this "grid on reactions" was heavy on passive aggressive tweeting.

Most alarming was the party line - that the response rate would be comparable and these changes would not have any negative impact on the quality of the data. This, despite warnings to the contrary from Statistics Canada three months earlier:


In short, Clement had been warned by the experts but decided to forge ahead. Rather than mount a reasoned defense of these changes, his strategy was to hope no one paid attention - after all, the Queen was in town, the World Cup was in full swing, and it was summertime! And really, who enjoys talking about statistics in the summertime? (besides me)

So Clement's office employed a "communications strategy" which removed all mentions of their involvement in this decision, in the hopes they wouldn't have to do any communicating:



They were not prepared, and it shows in the aftermath. Three days after the change was announced, Clement's Communications Director e-mailed StatsCan asking if any other G8 countries used voluntary censuses. This is a fact so basic you'd need it for any sort of communications strategy - clearly, there was no expectation they'd actually have to defend the move.

Two weeks later, his office was asking questions about what this change meant when it came to releasing data 92 years later (as is done for Census data) - to me, this isn't so much a communications problem, as a sign they hadn't at all considered the ramifications of their actions.

Clement himself had his facts wrong on his first public statement about the changes (on Twitter, 3 days later), and he did not issue a real defence until 9 days after the change was announced, when he first mentioned the "coercion" argument (on Twitter). He answered questions for the first time two days later - again on Twitter, and again it showed he did not fully understand the issue.

At this point, the government was still hoping the issue would die. Sure, statisticians were angry and there had been newspaper editorials but the story was just too boring to last longer than a week or two. After all, it wasn't like Helena Guergis was involved in it.

But the story didn't die. So sometime around July 15th, the government decided to fight back. Clement's office asked StatsCan for information about "Jedi responses" and prosecutions (yes, 3 weeks into this story, they still weren't aware no one has ever gone to jail over not filling out a Census form). Three days later, Dmitri Soudas sent out a letter on the topic. The first mention of the PMO in any of the released documents falls on this same weekend, so it seems likely this was when they first started taking it seriously (how could they not after watching Clement flail around?). That was also when Maxime Bernier started pontificating, either by coincidence or PMO design.

Clement also began doing more interviews on the topic.

In retrospect, he should have stuck to Twitter.

Despite knowing very well that StatsCan had spoken against the change (and had begun saying things in e-mails like "with the utmost respect, the answer you provided is factually incorrect"), Clement made it sound like everything was hunky dory at StatsCan.

This, despite the Head Statistician making it perfectly clear he was not cool with this. Notice how he suggests editing the following statement:


To this:


Despite all evidence to the contrary, Clement kept up the idea that StatsCan supported the move. At McGill. In a Globe & Mail interview. His office kept pushing for Sheikh to validate the government's decision - something they had made it clear they could not do. As StatsCan e-mails show, they believed Clement had "decided to put the ball in [their] court".

So on July 21st, Munir Sheikh resigned. By putting the ball in Sheikh's court, Clement turned a bad news story into the Great Census Crisis of 2010.

And now, thanks to these documents, the story has morphed into one of whether or not Tony Clement lied. Generally speaking, stories about lies and cover ups are easier for the general public to digest than ones on survey sample methodology.

This can all be traced back to that first "won't some stakeholders argue the opposite" comment I posted off the top. Simply put, Clement's office didn't understand what they were doing and didn't expect anyone to care.

As a result, we're now into week 7 of a story that simply refuses to die. So if Tony Clement feels lonely, he really has no one to blame but himself.


For background on Census issue, you can read "Everything You Always Wanted To Know About The Census (But Were Afraid To Ask)"

Labels: , ,

33 Comments:

  • http://parliamint.blogspot.com/2010/08/vic-toews-mr-hypocrisy-visits-gay.html

    By Anonymous Anonymous, at 9:26 AM  

  • "Generally speaking, stories about lies and cover ups are easier for the general public to digest than ones on survey sample methodology."

    Well, it's even better than that, in terms of having people relate. How many Canadians, do you think, have had some idiot boss come up with an obviously bad idea and then blame them, or one of their coworkers, when the inevitable happens? 33%? 50%? 75%?

    By Anonymous Anonymous, at 10:10 AM  

  • I think they don't understand stats. When they talked to them before, stats-can mentioned that voluntary surveys have roughly 2/3 response rate. Whoever decided on this likely thought, "Oh, in that case, let's send it to more households." So instead of 100% of 20% of households, we now get 66% of 30% of households. And since they don't understand math, they probably assumed that all that mattered was the total number of data points.

    By Anonymous Matt, at 10:48 AM  

  • Here's a warning about what life would be like under a harper majority.

    Bad! Atrocious! Terrible!

    By Blogger JimTan, at 11:32 AM  

  • Well, it's even better than that, in terms of having people relate. How many Canadians, do you think, have had some idiot boss come up with an obviously bad idea and then blame them, or one of their coworkers, when the inevitable happens? 33%? 50%? 75%?

    Ha Ha. Love it. That should be the Liberal campaign narrative next election.

    By Blogger calgarygrit, at 12:20 PM  

  • Nice job, and thanks for incl. the rel. links & screenshots of some of the most damning passages.

    Q, tho: a commentor on one of the media stories said that somewhere in the doc's it said they weren't going to count anyone with no fixed address this time. Have you seen that?

    By Anonymous Anonymous, at 12:22 PM  

  • So now the detractors (including our gracious Dan) are saying that the problem is with the overall response rate.

    Can't they at least get their own story straight?

    By Anonymous Anonymous, at 2:26 PM  

  • yeah, re: the "idiot boss ... {w.] obviously bad idea ...blam[ing] them, or ...coworkers, when ...inevitable happens"

    ...are there any photo-shoppers out there who can work up a (Dilbert's) Pointy-Haired Boss ft. the hapless Tony & forward it to Rick Mercer?

    By Anonymous Anonymous, at 2:40 PM  

  • @ Anon 2:26's lame attempt to turn the tables:

    yeah, the story _is_ straight: for the census q. data this is trying to continue, they need an overall response rate of, like, 99% of the random sample for them not to be concerned about selection bias among the respondents. Nimrod.

    By Anonymous Anonymous, at 2:45 PM  

  • It's nice to know government works the same way as the private sector - they communicate only when it's too late to act on any responses.

    Great summary, by the way Dan.

    After all these years it amazes me how the Tories keep stumbling into these things. They always act so shocked that they're being so carefully scrutinized.

    By Blogger Robert Vollman, at 8:38 PM  

  • And really, who enjoys talking about statistics in the summertime? (besides me)

    Me!

    Generally speaking, stories about lies and cover ups are easier for the general public to digest than ones on survey sample methodology.

    I actually chortled while reading that on the bus tonight. Chortled!

    By Blogger WJM, at 11:30 PM  

  • Two things stand out for me. Firstly, Clement appears to have believed the changes would have a minimal impact. I wonder to what extent he read what Stats Canada was telling him. Perhaps he skimmed through some emails and focused in on parts that supported his move (eg. "although it is [acceptable] for some social surveys of a recurring nature").

    For politicians with strong cognitive biases, the neutrality of civil servants may be problematic. Wishful thinkers may only pay attention to corroborating data in reports, and sometimes key objections will be buried deep into a memo.

    For instance, the problems leading to the Challenger disaster were mentioned in a memo sent to management. However, the revelations were not given centre-stage under the heading: "RE: OMFG WE'RE ALL GONNA DIE"

    Secondly, it is interesting that Clement only asked questions when faced with a communications crisis. That may speak to his motives (politics vs. policy), but I also think it is revealing of something else. Clement IS a smart guy, but one of the problems with smart people is that they don't always ask questions.

    So while some people may paint this as the peril of putting country bumpkins in office, this is not supported by the fact that Harper's least bumpkiny minister is responsible for the blunder at hand.

    By Blogger hosertohoosier, at 12:20 AM  

  • 1. WTF. This is easily the best coverage of the issue I've found.

    2. Do you buy the Paul Wells / Stephen Taylor conspiracy theory of undercutting the long form census as a way to dismantle the welfare state? I don't after reading your post (and htoh's comment above).

    3. Is there any usefulness at all in the long form census if it is voluntary? That has remained unclear to me.

    By Blogger matt, at 12:34 AM  

  • @ Matt: mind if I (aka Namesake, on Kinsella's blog, and a couple of the anon's above) field this in the meantime:

    on 2: Yes, I think they're half right; but they leave out the part that drove Kinsella to break ranks & cry out for the coalition in the first place: viz., this is probably also tied to the plan to kill the vote subsidy, since the small area data is an invaluable fundraising & campaigning resource that the other Parties are soon going to need like never before, & it'll be half-worthless; whereas the Cons don't need it anymore now that they've got the best Canadian political database ever.

    (and I noted that 6 weeks ago at http://worthwhile.typepad.com/worthwhile_canadian_initi/2010/06/an-incredibly-stupid-decision-on-the-2011-census.html?cid=6a00d83451688169e201348529b887970c#comment-6a00d83451688169e201348529b887970c

    re: 3. Sure, the new National Household Survey will have value as the largest General Social Survey ever. The GSS's run every year on different themes like Time Use, & Criminal Victimization, and used to have sample sizes of just 10,000, which have now been bumped up to 22,000. But even with what'll probably now only be a 40% response rate (thanks to all the abuse its taken by the ConBots), the NHS will have MILLIONS of respondents, permitting all kinds of detailed cross-tabs & analyses comparing the influence of various variables (occupation, ethnicity, education etc.) But only if we weight it using -- you guessed it, the _last_ long form census. So it won't be good for what StatCan & all the planners etc. want which is to use it to weight future surveys & make accurate assessments of what's needed today. So it might as well just be cancelled, or just be sent to, say, half a million households, instead of 6.5-M or whatever it is, and save either $100 or $75-M.

    By Anonymous Anonymous, at 1:41 AM  

  • "But only if we weight it using -- you guessed it, the _last_ long form census"

    Goodness.

    There is every reason to believe that respondents lied outright on the last longform census, and you're basing your weighting on it?

    Have you no statistical training at all?

    By Anonymous Anonymous, at 5:01 AM  

  • Terrific coverage of a complicated story, Dan.

    By Anonymous Jacques Beau Verte, at 10:08 AM  

  • @ the "Goodness" comment: twerp. Yes, StatCan -- the preeminent Stat'l agency in the world -- does use the long form to weight their other pop. surveys, & they have plenty of training, thank you very much, and plenty of error & outlier (& yes, BS replies) correction procedures, incl. filtering out the silly 0.07% "Jedi" religion responses you ConBots try to wave around like truthiness swords.

    By Anonymous Anonymous, at 11:13 AM  

  • @ hosertohoosier: I really don't think this was Clement's decision at all: it was Harper who pushed it (goaded on by Bernier). In fact, the media reported Clement tried to dissuade Harper, but gamely soldiered on as best he could (i.e., badly) when overruled, and the correspondence makes clear that the PMO intervened when things started falling apart on the messaging & media relations.

    By Anonymous Anonymous, at 11:18 AM  

  • and hey Con-bot: before you go to the trouble of countering that it was the far more serious (because expensive, to provide more services) only-speak-French language q. you had in mind, not the 'Jedi's, consider this:

    It's now well-known that some gas pumps give false readings, too. Tell you
    what, let's stop using that mandatory measurement system to gauge what's owed: let's let people give a voluntary estimate and what they think they should pay, as with shareware. (See how long your blue-eyed Sheik pals in Alberta will like or permit them apples.)

    Or better yet, let's charge everyone the full capacity of their vehicle's gas tanks every time regardless of what they put in, because after all, either they or the pumps would be lying (but not the manufacturers of the vehicles, no way). (And see how many hours it takes to get the Timmies crowd you lot have been trying to lead by the nose to become a lynching mob.)

    That's the only rational way to do it, or have you no economic training at all?!

    (brought to you by the unfriendly neighbourhood Namesake)

    By Anonymous Anonymous, at 12:40 PM  

  • Annoying thing is CPC doesn't show ability to learn beyond Social Credit GOP ideology. A government beauracracy is sometime more efficient than pri-beau, pub-market, pri-mar; for some things assuming goal is quality-of-living. In other countries a market forces census is intrusive and expensive. Here, CPC voluntary census undercounts poor.

    You get highest quality-of-living gains from 1st $10000/yr; money really does buy happiness. And when a third of our largest companies are petro and the externalities of AGW aren't costed/taxed, cutting highest-tier corporate taxes is horribly inefficient. This is a holocaust in the making we can maybe still prevent.

    By Anonymous no more holocausts, at 7:48 PM  

  • Another thing, our government public service has been ranked #1 in the world by a trust index, I forget which one. Privacy concerns may be an issue one day; if anything we should be setting the bar here for other nations with corrupt civil services instead of deconstructing a very efficent programme. Very simply, if the 1st $10000/yr in income really means so much for quality-of-living, and if this government gives everything to the rich, forcing social net cuts (that aid individuals reaching 1st $10000/yr in income)....Alberta is hostile to Canadian quality-of-living.

    By Anonymous SS Calgary oil execs, at 7:55 PM  

  • For an example of info and market forces efficiency: Thomson. The company sold its media holdings and bought medical and educational specialized data. It sells this data to professionals. You can't self diagnose yourself accurately online anymore than you can print out a degree. Probably the actual medical equipment can be amateur made on the cheap in an exceptional circumstance like a pandemic stockpile (fossil fuel energy is inferior here to Ignatieff's hydro east-west grid). But gov of Canada would be inefficient to infringe on Thomson.

    By Anonymous unleashing Revelations, at 10:46 PM  

  • Well, it's even better than that, in terms of having people relate.

    By Anonymous Zenegra, at 1:42 PM  

  • Find away how with the intention of get amazing done a reorder cell phone lookup otherwise a telephone search or else still a reverse phone lookup number search lookup with the intention of uncover elsewhere who is losing with the intention of individual to facilitate has been calling you this entire moment ok.

    By Anonymous Anonymous, at 10:25 AM  

  • Undo phone search internet pages resembling this one motivation give you all kinds of telephone search specifics so that you can ultimately come across out their name, address, where they exist, reverse phone lookup their current age along with greatly other. So whilst you want with the purpose of pay out a mobile phone number lookup today, simply complete persuaded so after to run through this caring tool to locate not in both kinds of reverse telephone search specifics down that one at present.

    By Anonymous Anonymous, at 8:00 PM  

  • I enjoy this nice undo telephone lookup or phone search or cellular phone number lookup spot. I be able to hear to a extreme extent of phone number lookup material on the subject matter of repeal telephone lookup here nowadays once I check away commencing home this guide here.

    By Anonymous Anonymous, at 9:29 PM  

  • By Anonymous Cure ear infection, at 3:32 PM  

  • Great website, looks very clean and organized.

    By Anonymous free bondage porn video, at 2:55 AM  

  • Nice info and articles .Thank you.

    By Anonymous Data recovery software, at 3:21 AM  

  • Doxycycline
    Doxycycline is a tetracycline antibiotic characterized by bacteriostatic pharmacological action. Buy Doxycycline

    By Anonymous doxycycline 100mg, at 11:30 PM  

  • Download From Youtube
    Fast speed and easy-to-use downloader is the revolution in downloading and converting video files. Download Video From Youtube.com

    By Anonymous Tube Downloader, at 11:39 PM  

  • I suggest this site to my friends so it could be useful & informative for them also. Thanks!!!!

    By Anonymous Colon Clenz, at 5:13 AM  

  • Dude I have to say I'm so happy to find your blog, even I did it by a fluke. The information posted here is very very useful for so many people in the whole world. In short, a really nice blog.

    By Anonymous viagra online, at 3:49 PM  

Post a Comment

Links to this post:

Create a Link

<< Home