When the poll numbers don't fit the narrative...

I suspect news editors sometimes think that they have a fix on public opinion, commission a poll to prove it but allow their enthusiasm for the storyline they have concocted in their heads to cause them to write the story before the hard news has arrived.

The Globe published its latest Strategic Counsel poll this week but the numbers didn't fit the narrative that the Globe appeared to be expecting of a Harper government doing well and perhaps headed for a majority. But it pushed that narrative in its coverage anyway. They covered the contradiction by calling it the "Harper Paradox".

The Globe editors appear to be astonished and chagrined that the Conservatives aren't on the verge of a majority. Thus, although the poll showed the Liberals and Conservatives tied overall with 33%, for the Globe, Canadians display "mixed feelings" about Harper as a person as they "grow more comfortable with the direction he is taking the country and as support for his party solidifies."

To prove this point we discover that 57% think the country is on the "right track" (I will come back to this phrase in a moment). Um, except, we then find out that this is four points DOWN from a year ago. Where can we take comfort? Well it is "up more than 10 points since the final days of the Liberals in early 2006."

Now there is an apt comparison. Let's take the tracking number for a government in its final days as it heads to defeat (having been labelled over and over again as corrupt) as the yardstick, rather than public opinion on Harper from a year ago, once the country has had eight months to react to his administration.

Returning to the "right track" statistic, while I have followed it in various polls for years I am not sure what it means to the public. One could think the country is on the right track because the economy is humming, even while one is appalled by the government's inaction on global warming and overseas military adventures. It makes more sense to me to ask a simple approve/disapprove type question to determine how the electorate actually feels about a government. Note that even when the Liberals were headed for a wintry defeat in 2006, getting just 30% of the vote in election 2006, according to the Globe the Grits had a right track number somewhere in the mid-forties.

The poll's details are all available at the Strategic Counsel web site here. However, the narrative problem continues in talking points on the SC web site and in the analysis by the firm's principals, Peter Donolo and Tim Woolstencroft, in the Globe. Thus the story begins:

After a year and a half in office, Prime Minister Stephen Harper has built up some impressive political capital.

Nearly six out of 10 Canadians think the country is on the right track. Even if he's not setting Canadians' hearts aflame, most have a neutral-to-positive impression of Mr. Harper. He's seen as a decisive leader. A majority believe he's kept his promises. And most Canadians trust him to do the right thing for the country.

Before we address this paragraph, I will make two points:

1. On the SC web site (page 5) there is this observation: "While there has been some volatility in party support over the past 18 months, party vote has stabilized in the most recent sampling around the levels of the 2006 election."

Oh really. In the last election the Conservatives nationally had 36% and the Liberals 30% whereas in the poll they are tied at 33%. My estimate of the seat numbers using the SC data is that this distribution of support would actually reverse the net outcome of the election and produce a Liberal minority. Is it really accurate to say "around the levels of the 2006 election"?

2. From the Globe story: "In Quebec, meanwhile, Mr. Harper's Conservatives are in solid second place to the Bloc Québécois outside of Montreal. The percentage of Quebeckers who believe the country is on the right track has nearly doubled in two years."

Um, except the poll data tell us that the Conservatives overall have declined in Quebec including outside Montreal while the Liberals have gained. To be sure the BQ is also down a bit but the gains outside Montreal at the expense of the BQ have seemingly gone to the Liberals, the Greens and the NDP while the Conservatives have dipped. See the table on page 28 of this pdf.

You get the picture.

Much of the poll is composed of mush - meaningless statistics about how many Liberals, NDP etc. are "open" to voting Conservative, feelings about Harper being "partisan" or "controlling". But leadership sentiment in any case is often a trailing indicator of party support not a leading indicator. When the last election campaign began, Harper's leadership numbers were a net -17 and still -10 three weeks into it but up to +12 with a week to go. See here on page 17.

The one agree/disagree statement in the mix that seems fairly straightforward to me is: “The Conservative government under Stephen Harper has accomplished little during its time in office.” (On page 3 here) 53% of Canadians agreed with that while only 41% disagreed.

Re: the paragraphs above regarding Mr. Harper's political capital. Note how the authors don't mention this Q & A, a simple question asking respondents for their overall assessment of the government's performance. Why is this not noted?

In a world dominated by celebrity and gossip journalism, is this the best analysis our national newspaper can produce?