Facebook Pixel

Originally from http://davidakin.blogware.com/blog/_archives/2010/12/7/4698036.html

There’s a new horse-race poll out this morning from Abacus Data Inc., a new-ish Ottawa-based pollster affiliated with government relations firm Summa Strategies Inc.

Abacus enters a relatively crowded field of pollsters that do political work, joining the likes of Leger Marketing, Nanos Research, Ipsos Reid, The Strategic Counsel, HarrisDecima and Ekos.

The poll from Abacus this morning is its second-ever ‘horse-race’ or ballot question poll and this one is sure to drive the Liberals nuts: When asked,”If a federal election were held today, which party would you vote for in your constituency?”, just 24 per cent of decided voters said they’d vote Liberal and 20 per cent said they’d vote NDP. (NDP staffers are already gleefully tweeting the Liberal number) Given that the poll’s margin of error (1,361 questioned online) is 2.7 per cent 19 times out of 20, one might say that the red team and orange team are almost close enough part to call it a tie. Abacus found Conservative support to be at 35 per cent. (More on the numbers in a minute.)

The brains behind Abacus is the cherub-faced David Coletto (left), who, this year, earned a doctorate in political science from the University of Calgary. Coletto fills a hole for Summa created ever since Nik Nanos left the firm shortly after the 2006 election. Nanos, political junkies will recall, pretty much called the results of the 2006 election while some of his competitors seemed miss by a mile. Nanos’ accomplishment with the 2006 election opened up a lot of new doors business-wise that he couldn’t walk through while still affiliated with Summa. So Nanos headed off to start up Nanos Research and Summa has been looking for a market research partner ever since.

Summa, by the estimation of its own competitors, is one of a handful of powerhouse government relations firms on the Hill with a judicious mix of players who with both Liberal and Conservative pedigrees. Its chairman is Doug Young, the former Liberal cabinet minister from New Brunswick. The firm’s President is the affable Tracy Hubley, a Prince Edward Islander who cut her political teeth on the staff of several Liberal cabinet ministers.  On the Conservative side of the ledger, the firm has vice-presidents Tim Powers, a frequent spear-carrier for the Conservatives on political talk shows, and Jim Armour, a former communications director for Stephen Harper when he was in opposition.

I point out these political pedigrees as one factor in which one might consider the independence, if you will, of this poll given its rather bleak results for the Liberals. My sense is that Abacus and Summa are trying to build a business that will be attractive to its corporate clients, clients who tend to avoid paying lots of money for polls or market research that might be skewed by a pollster’s own political or corporate agenda.

While the specific methodology — order of questions, phrasing of questions, data cleanup — used by each pollster is often a closely guarded “secret sauce”, Abacus discloses that it draws its polling sample from an online panel of about 100,000. Other pollsters do this as well and, in the trade, there is often an interesting debate about the validity of online polls versus old-fashioned telephone polls. (On the one hand: computer access is closely correlated to higher income and education levels, which could skew results; on the other hand, an increasing number of Canadians are cutting landline telephones in favour of wireless-only).

So, for journalists and others who want to assess the reliability of a given pollster, what to do? Most journalists use their own judgement — a gut feeling, if you will — in combination with two other methods to check up on things. The first: We phone up the parties themselves. They do their own polls and, while their own numbers are jealously guarded, they’ll often tell you if one poll or another is in the ballpark. Secondly; we look at other pollsters are doing. So, by that second measure, where does Abacus stand against a couple of other recent polls?

Abacus (Dec 7/2010):

  • CPC: 35%
  • LPC: 24%
  • NDP: 20%
  • BQ: 10%
  • GPC: 10%
  • Quote: “It is difficult to say whether these numbers translate into a majority government for the Conservative Party. Since the Liberals and NDP are stronger in different regions of the country it will be difficult for the Conservatives to make enough gains to counter the BQ’s dominance in Quebec.”
  • Methodology: Between December 3rd and 6th, 2010, Abacus Data Inc. conducted an online survey among 1,361 randomly selected Canadian adults from an online panel of over 100,000 Canadians. The margin of error—which measures sampling variability—is comparable to +/- 2.7%, 19 times out of 20.

Nanos (Dec 6/2010)

  • CPC: 38.1%
  • LPC: 31.2%
  • NDP: 17.2%
  • BQ: 10.2%
  • GPC: 3.2%
  • Quote: “The current configuration of national support for the Conservatives suggests that numerically a Tory majority government can be formed without a significant breakthrough in the province of Quebec.”
  • Methodology: Nanos conducted a random telephone survey of 1,002 Canadians, 18 years of age and older, between November 29th and December 2nd, 2010. A survey of 1,002 Canadians is accurate to within 3.1 percentage points, plus or minus, 19 times out of 20. For 747 committed voters, it is accurate to within 3.6 percentage points, plus or minus, 19 times out of 20.”

Ekos (Nov. 25/2010)

  • CPC: 33.3%
  • LPC: 27.1 %
  • NDP: 16.6%
  • BQ: 9.5%
  • GPC: 9.5%
  • Quote: “The Conservatives are once again establishing a significant lead over the faltering Liberal party who now appear to be chronically rebuffed by the electorate as they try to move past the Conservatives…The Conservatives are now very close to their moving average over the past year, which is still well short of their achievement in the last election.”
  • Methodology:  EKOS’ weekly tracking polls are conducted using Interactive Voice Recognition (IVR) technology, which allows respondents to enter their preferences by punching the keypad on their phone, rather than telling them to an operator. In an effort to reduce the coverage bias of landline only RDD, we created a dual landline/cell phone RDD sampling frame for this research. As a result, we are able to reach those with both a landline and cell phone, as well as cell phone only households and landline only households. This dual frame yields a near perfect unweighted distribution on age group and gender, something almost never seen with traditional landline RDD sample or interviewer-administered surveys. The field dates for this survey are November 17-23, 20101. In total, a random sample of 1,973 Canadians aged 18 and over responded to the survey (including a sub-sample of 1,696 decided voters). The margin of error associated with the total sample is +/-2.2 percentage points, 19 times out of 20.

So what do you think? Smoke and mirrors or a true pulse-taking?

Good Decisions Require Good Data.