Tuesday, August 24, 2010

Old Léger Poll: 9-pt Conservative lead (down two)

Léger Marketing has finally put up the results of a federal poll taken between August 2 and August 4. The poll is now three weeks old and accordingly less relevant, but it is worth taking a look at nonetheless.

The poll has the Conservatives at 37%, up one from the previous Léger federal poll taken in April 2010. The Liberals are up three to 28% while the New Democrats are down four to 16%. The Bloc Québécois is unchanged at 9%, as are the Greens at 8%.

The Conservatives lead in Ontario with 41%, up two. The Liberals follow with a gain of four points and are at 34%. The NDP is down three to 18%.

In British Columbia the Conservatives are up seven to 48%, followed by the NDP and Liberals at 19% apiece. The NDP dropped five points and the Liberals dropped eight. The Greens picked up three to reach 11%.

In Atlantic Canada, the Liberals gained six points and are at 43%. The Conservatives dropped six to 24% and the NDP dropped nine to 20%.

The Conservatives lead in Alberta with 60%, followed by the Liberals at 20% (up six) and the NDP at 11%.

And finally, the Conservatives lead in the Prairies with 48%. The NDP and Liberals have 23% apiece, marking a six point gain for the NDP.

The Conservatives would have won 76 seats in the West and North, 55 in Ontario, 8 in Quebec, and 6 in Atlantic Canada for a total of 145.

The Liberals would have won 38 in Ontario, 23 in Atlantic Canada, 15 in Quebec, and 14 in the West and North for a total of 90.

The Bloc would have won 50 seats in Quebec.

The NDP would have won 13 seats in Ontario, 5 in the West, 3 in Atlantic Canada, and 2 in Quebec for a total of 23.

Now, what of the polling result for Quebec? This is where things get odd.This poll is available here, as it was conducted for Le Devoir. But the results of this poll, taken between August 16 and August 19, are exactly the same as the one taken between August 2 and August 4. Is that possible? The amount of people polled in the first survey (377 in Quebec, 1,500 Canada-wide) was much smaller than in this one, yet the results are exactly the same, down to the Greens and Other.

While it is certainly possible that Léger would get the same results, it is pretty unlikely. For that reason, I will be treating the Canada-wide poll and the Quebec-only poll as one poll for the projection, rather than two.

In this Quebec poll, the Bloc is down two points from June and leads with 37%. The Liberals are down one to 24% and the Conservatives are up three to 19%. The NDP is down two to 13% and the Greens are up four to 7%.

Demographically, the Bloc dominates among francophones with 44% (down one). Interestingly, the Conservatives are second among francophones with 18%, up three. The Liberals have dropped to 17% (down four) while the NDP is down one to 14%.

Among non-francophones, the Liberals dominate with 47% (up three). The Conservatives follow with 20% (down two) and the NDP and Greens are tied at 11% each. The Bloc is down four to 9%.

Regionally, the Bloc leads in the Montreal region with 32%, down nine. The Liberals are down one to 28% while the Conservatives are up seven to 18%. The NDP is at 13%, down one.

In the Quebec City region, the Conservatives have gained five points and lead with 33%. The Bloc follows with 29%, down three. The Liberals are up one to 18% and the NDP is down three to 15%.

In the rest of Quebec, the Bloc is up four to 43%. The Liberals follow with 20% (down three). The Conservatives are down one to 17% and the NDP is down three to 13%.

As to which leader garners the most confidence from Quebecers, Gilles Duceppe leads with 28%, followed closely by Jack Layton at 27%. Stephen Harper only has 11% while Michael Ignatieff has 9%. This shows that the two main federalist parties are not being helped by their own leaders, while the NDP has room for growth.


  1. "While it is certainly possible that Léger would get the same results, it is pretty unlikely. For that reason, I will be treating the Canada-wide poll and the Quebec-only poll as one poll for the projection, rather than two."

    Unlikely ??

    If a pollster says its poll is valid 19 times out of 20 that means that 95% of the time they replicate their poll the results should be the same.

    Same pollster, same method, pretty much the same time frame.

    Shouldn't it be more than likely they'd get the exact same results ??

    (As for treating it as one poll that seems fair since its essentially the same information. However, don't you weight projections on sample size and decrease the weight of older polls ?

    For that reason shouldn't the newest Quebec numbers be used as they are more recent and have a larger sample for weighting purposes ?)

  2. Yes, I will be using the newer numbers for Quebec in the projection.

  3. Stated confidence intervals for polls are supposed to include the possibility for variation. This margin of error is usually listed as a part of that statement.

    The confidence interval is meant to express easily the most likely range of results.

    If someone running a national poll actually claimed a margin on error under 1% or zero, I'd expect to see a sample size of something like ten thousand people or more.

    Two polls in sequence with different samples getting the exact same results is not impossible, but it is pretty unusual. Either there was no random statistical variation at all, or whatever there was happened to be exactly the same as the last poll's, or the change in the measurements happened to coincidentally compensate for fluxuations in random variation.

  4. http://www.visioncritical.com/2010/08/two-in-five-canadians-would-scrap-long-gun-registry/

    13% of Canadians think the gun registry has helped prevent crime, 29% think no effect and 43% think it has been completely unsuccessful.

    ....Just 13% think it is useful to prevent crime.

    44% want it scrapped, 35% want to keep it.

    To be honest I have trouble reconciling peoples opinions for them... 13% think it is useful for its effect on crime,... but 35% want to keep it.. I don't get it.

    If people don't believe it makes them safer... then what good is keeping it? What is the rational? Is it just controlling what other people do in an area that you just don't care about?

  5. Barcs - People are bad at understanding costs. They don't see any problem with the resistry, and while they don't think it has an effect of crime, they probabaly think it might or it could, so they're keeping it just in case it comes in handy.

    That it costs millions of dollars to run doesn't seem to enter their minds.

    This happens a lot. People don't solve a problem more than one layer deep.

  6. 95% of the time they are within the margin of error of the poll.

    if it was +/-3%, and you had a 4 company poll that had each 25% of the vote. The company would say that 95% of the time we are within the margin of error.

    25-25-25-25 would in effect be 22-28% for each of them. Given 19 more polls, atleast 18 of them should show each of those 4 companys always in the 22-28 range.

    But given random variability, a deviation of 6 points on each of 4 companies... should produce slightly different numbers on every poll since any one of them could be up to 3% higher or lower.drame

  7. "Barcs - People are bad at understanding costs."

    Basic economics should be a highschool level class required to graduate.

    A cost/benifit ratio is at the heart of absolutely everything we do every day.

    Want to go to town? Well how much gas do I have? what am I getting there? Am I going tomorrow anyway? The wife is there already, why doesn't she bring it home?

    What are we having for supper? Does steak cost too much? Or is it macaroni again? IThe wife wants to go out... do we have the $$ to do that?

    I don't understand how so many people around are so bad with making decisions about money, I would think that the basic decision making about their survival would entice them to learn. But then again I guess it is probably the result of too many safety nets and social programs where the providing for you and the thinking for you is done for you.

  8. Kevin thanks for the correction.

    Right so a confidence interval of 95% means that that is the probability of the survey's results being valid anywhere within the range of + or - the MOE.

    But the distribution would be curved would it not ?

    And everything is being rounded off as well, which could hide a fair bit of noise.

    Anyways i'm not a statistician. Maybe we could get Tony Clement in here to settle if this is an unusual event or not ?


    Earl maybe people think the long gun registry helps SOLVE crime and not just prevent it.

    (Not a supporter, just offering an idea as for the discrepancy you pointed out.)

  9. While realizing we are digressing, does anyone understand how the long gun registry costs so much?

    Eg., you can register online with the Department of Foreign Affairs as an expatriate. This information is automatically (I imagine) entered into a database. The whole thing probably costs under a million dollars to administer.

    Granted, there are more Canadian gun owners than expatriates. Nonetheless, how does something that shold be mostly self-administering cost over a billion dollars?

    I have no problem with registering long guns. We registrer cars and lobbyists, which are only moderately more dangerous. But not if the costs outweigh the benefits. How did the costs get out of control? Anyone have insight?

  10. All,

    I favour the long-gun registry (like most chiefs) but it's another he said-she said of an issue: most cops are killed by long-guns (it isn't even close); most firearm seizures are long-guns but, on the other hand, most killings are committed with handguns. So take you pick. Which statistical poison do you prefer?_

  11. Though isn't the point of solving crime ultimately to prevent crime?

  12. Shadow wrote,

    "But the distribution would be curved would it not ? "

    Indeed, at a Confidence Interval of 80%, one could reasonably say that the sampling error in a poll of 1500 is only about 1.6%.

  13. "most cops are killed by long-guns (it isn't even close); most firearm seizures are long-guns but,"

    Did either of those change with the registry? did deaths of officers plummet with the introduction of the registry? or are they generally the same as before? The fact that there is still deaths speaks to its ineffectiveness, its not a comparison on whether or not there would have been fewer, or more deaths with or without the registry.

    Has it resulted in easier tracking? or more work? (my friend received a call from the Vancouver police.... tracking down 177 of a set of 178 guns registered with what people thought was a serial number.... it was a model number, and they had to track every one of them to confirm that the one they had was the only 1 missing. He... among others told them to go away, so uniformed officers spent hours going to each residence. While they were there they discovered that on a second gun, although the form to register he sent in was correct, the data entry person screwed up the serial number. That is 2 screwups on the 3 guns he had. Nice system)

    "In the survey of 2,631 officers nationwide, 2,401 -- or 92% -- agreed the registry didn't help them perform their duties and was a useless crime-fighting tool."

    ... why don't the chief's of police agree with their members??

    GI, The 2 billion cost was in setting up the registry. Computer programming for the system, people to enter the data, equipment to run the system, creation of the databases and how the information was to be collected and accessed. etc etc.

    The yearly cost to run it now is approximately 4 million.

    With an estimated 3 million gun owners, and 9 million guns in Canada (midrange guess from teh department of justice), that is $166 per registration to set up, (except that they are still missing several hundred thousand guns) and 30 cents per registration per year to keep it (on top of the fees every person who registers pays into it... in other words.... nowhere near 30 cents)

  14. You know this whole Long Gun Registry thing is a farce.

    The running cost to each Canadian looks like 8¢ a year.

    Hello?? Aren't you willing to invest a dime to help police??

  15. Almost looks like, Eric, you get what you pay for ??

  16. I'm not willing to invest a dime to help the government track information they don't need.

    Information is power. Look at Stalin.

  17. "Information is power. Look at Stalin."

    Stalin had the world's largest army, nuclear weapons, fanatical party followers and secret police that arbitrarily killed millions of people.

    He ignored some of the most important information that he did have (eg., that Germans were preparing for a surprise attack.)

    Knowledge is power, but in Stalin's case, guns were power.

  18. Ira:

    I'm not willing to invest a dime to help the government

    Thank you Ira, you don't support the troops, glad you made that clear.

  19. I hear people telling me that a new spending program will only cost me a dime a year about a million times a year.

    Do the math and you get the idea.

  20. "Hello?? Aren't you willing to invest a dime to help police??"

    Why do you hate ambulance drivers peter?

    Yes, I know a stupid leap of logic. But one just like your post peter at 19:32,.....

    But to be more serious, Yes I am willing to invest to help the police. To give them the equipment and databases they need. To give them prisons for example.. to get the criminals off the street and make their job easier.

    But I don't think the gun registry does that. It is not worth you 8cents,... let alone mine or everyone else's. (and to be fair,... it is more like 1/3 of the population paying 24 cents, 1/3 paying nothing with the middle 1/3 picking up the rest.... ) I think I can guess which 1/3 you are peter, by your willingness to spend spend spend the monies that the government wrests from its citizens... well most of them.

    And back to the 14 out of 16 deaths of police officers were killed by long guns. Noone seems to have answered, Is that better? or worse then before the registry? Is it better or worse because of the registry?


COMMENT MODERATION POLICY - Please be respectful when commenting. If choosing to remain anonymous, please sign your comment with some sort of pseudonym to avoid confusion. Please do not use any derogatory terms for fellow commenters, parties, or politicians. Inflammatory and overly partisan comments will not be posted. PLEASE KEEP DISCUSSION ON TOPIC.