Federal polling - April 15

123 posts / 0 new
Last post
Stockholm

This poll is totally meaningless. They ask people who they would bet ONE THOUSAND DOLLARS on to win the election. Well, guess what? any SANE person would bet on the Conservatives "winning" the election (assuming we define "winning" as getting more seats than any other party - as opposed to what happens post-election on a vote on a hypothetical throne speech). If someone was willing to bet $1,000 with me on which party would end up with the most seats and I could bet on the Tories at EVEN odds - i would already be making a list of all the things I would buy myself with the $1,000 I would be 99.999999999999999% certain to have in my pocket on May 3rd! There is still the huge question of whether it will be a majority or not etc...but seriously - who cares who people think will get the most seats - people in BC have been voting CCF/NDP in large numbers fo the past 80 years without having the slightest expectation that the NDP would win the election nationally. I'm actually amazed that only 56% of people in BC expect the Tories to win the election. Can I have the names and numbers of the delusional other 44% so I can bet $1,000 with each of them and become a a millionaire?

Seriously, there have probably been about 200 polls in the past years and half that have all put the Tories in first place by a significant margin - any person paying the slightest bit of attention would know that the Tories will get the most seats - surprise, surprise.

Is there ANYONE on babble who seriously thinks that the Tories won't be the largest party on May 3rd? If so please PM me and I'd be happy to bet ANY AMOUNT of money with you!

melovesproles

I've never understood why the NDP doesn't go negative against the Cons here.  Polarization works in their favour in BC.  The slick anti-Harper ads they ran in Quebec during the last election would have been great out here.  The fact Harper wants to hand out mandatory mins for growing six pot plants is well out of step with most BCers but you don't hear a peep about it, Harper's been given a total pass on his 1950s era social policies.  In the Ads out here the Cons are running against the Liberals and the Liberals are running against the Cons and the NDP is bashing the Liberals??  The NDP have allowed themselves to be taken out of the conversation in what should be one of their key battlegrounds.

Sean in Ottawa

Actually Stockholm it is 65% not 56% who say the Cons will win.

The remaining 35% could interpret a loss as not governing -- even with the most seats -- and that is still quite possible.

I think even if the Cons get the most seats if they lose a lot and the other parties end up forcing them out then that would be considered a loss.

As I said befoer disregard Ignatieff's promise not to have a coalition -- if he is not leader that promise is gone and if he does not get enough seats to govern on his own he won't be the one to decide on a coalition.

That said it is not clear that the NDP would even want to go in a coalition-- they may prefer to do what has been heppening all along-- individual votes but sit in opposition-- letting the Liberals govern for a while.

 

Sean in Ottawa

Of course that is if the Liberals have enough seats to take a run at it by themselves-- they may not. It could even be the NDP asking the other parties for support. The election is not over.

Northern-54

Stockholm wrote:

Garth Brasseur wrote:

Three nights ago in the Nanos tracking poll, the NDP had the best night ever.  This night will drop off the poll tomorrow.  I expect that there will be a substantial drop as the 22% will be replaced with an 18% or so.  This is the problem with tracking polls.  Movements day to day can be misleading.  

Unless you know a spy who works in data processing at Nanos Research, you have no way of knowing what the NDP polled on an individual night. You can speculate that the NDP must have had a really good night three nights ago because support went up alot of that days results - but we don't know what the results were for the day that got dropped off as well.

 

Actually, I calculated it out with a spreadsheet.  It isn't that hard.  You take the average of the three days from the day before.  Then, you substract 1/3 of that vote.  Then you add on what is necessary to make the new percentage and multiply by 3 to get the daily poll result.  I am a Math teacher and I know this will give me a good estimate of what the night's vote was (certainly within a half percentage).  I do wish I had a spy there...

Aristotleded24

melovesproles, I hear you. Hopefully the NDP has an ace up the sleeve and the selection of a new BCNDP leader will free up some resources to go head-on.

ghoris

melovesproles wrote:

The slick anti-Harper ads they ran in Quebec during the last election would have been great out here.  The fact Harper wants to hand out mandatory mins for growing six pot plants is well out of step with most BCers but you don't hear a peep about it, Harper's been given a total pass on his 1950s era social policies.  In the Ads out here the Cons are running against the Liberals and the Liberals are running against the Cons and the NDP is bashing the Liberals??  The NDP have allowed themselves to be taken out of the conversation in what should be one of their key battlegrounds. 

I find this quite maddening as well. I still do not understand why the NDP spent so much time in the last election attacking Harper in Quebec (when Harper's Quebec supporters were never, ever going to vote NDP anyways) which just drove swing voters into the arms of the Bloc. The Liberals don't have a lot of growth potential here - the only real targets are Vancouver-Kingsway and maybe North Vancouver - meanwhile they're stuck playing defence in Esquimalt, Vancouver South and Newton-North Delta. All of the key battles seem to be NDP-Tory battles - why is the NDP not unloading with both barrels on Harper?  The HST snafu seems like a ready-made gift to the NDP. Am I missing something here?

Northern-54

http://www.angus-reid.com/wp-content/uploads/2011/04/2011.04.16_Debate_E...

 

I've not seen this poll mentioned. It is about the English debate.

Aristotleded24

[url=http://www.youtube.com/watch?v=U-kqPvo0WqM]Here is a little bit they've done on the HST issue[/url]

Anonymouse

Actually, I think Stockholm is right, I wasn't think about the survey question clearly (maybe the surveyors weren't either...haha). Going very neg on the Conservatives in BC in the air war might be good because, as was mentioned, it might polarise things. Harper addressing the HST in his latest campaign stop in Vancouver shows that this issue is coming up at the doorstep. It is clearly the Conservative's biggest gaffe since 2008...it cost Campbell his premiership after all. Also, the Conservative counter is not that great: "look at the NDP in NS, they reduced the HST on home heating then raised it on everything else." This is just an attempt at mis-direction, the main anger for BCers is that it was a tax increase and was lied about (during a provincial election campaign) before being brought in right-away later. Also, contrast the NDP's promises with the Conservatives, not only would they respect BCers decision in a referendum on the HST (and not essentially fine the province $1-2 billion), they would drop taxes on the very small businesses that got kicked in the teeth over the tax, and to a lot of people that live "off-grid" (e.g. rural areas) and say the HST take a hit to home heating fuel budgets.

Stockholm

Garth Brasseur wrote:

Actually, I calculated it out with a spreadsheet.  It isn't that hard.  You take the average of the three days from the day before.  Then, you substract 1/3 of that vote.  Then you add on what is necessary to make the new percentage and multiply by 3 to get the daily poll result.  I am a Math teacher and I know this will give me a good estimate of what the night's vote was (certainly within a half percentage).  I do wish I had a spy there...

The only problem is that you are taking the AVERAGE from the three days before when in fact we don't know what the numbers might have been on each individual day - so its a bit of a crude way of trying to estimate what the results were on an individual night.

Winston

Garth Brasseur wrote:

Actually, I calculated it out with a spreadsheet.  It isn't that hard.  You take the average of the three days from the day before.  Then, you substract 1/3 of that vote.  Then you add on what is necessary to make the new percentage and multiply by 3 to get the daily poll result.  I am a Math teacher and I know this will give me a good estimate of what the night's vote was (certainly within a half percentage).  I do wish I had a spy there...

I am currently finishing off my degree in mathematics, and I have to say that you and Stock both are a little correct, and both a little wrong.

While your procedure should provide a satisfactory estimate of the numbers, you must remember that the third you are removing is an average of the prior three days.  But what Nanos actually drops off is the first of the three days (not the aggregate numbers).

Let's say that Nanos samples a number of people each day and let's say that the NDP numbers for each of those days is y0, y1, y2, y3,...  The polling results that Nanos reports are the average of three days' polling.  Now let's say that you have  Nanos' numbers for 4 consecutive days, and denote those 4 days of polling as x2, x2, x3, x4,... We can now set up a system of equations as follows:

x2 = 1/3 (y0 + y1 + y2)

x3 = 1/3 (y1 + y2 + y3)

x4 = 1/3 (y2 + y3 + y4), and so on

It is clear from the above that we are dealing with a set of n equations (with known x values) and n+2 unknowns (all of the y's).  Since you are a math teacher, you will know that is impossible to find a unique solution to a set of n equations over n+2 variables.

Now if Nanos was willing to tell you the results of two of the daily samples, you would be able to solve for the rest of the unknowns. 

 

Northern-54

Winston wrote:

Garth Brasseur wrote:

Actually, I calculated it out with a spreadsheet.  It isn't that hard.  You take the average of the three days from the day before.  Then, you substract 1/3 of that vote.  Then you add on what is necessary to make the new percentage and multiply by 3 to get the daily poll result.  I am a Math teacher and I know this will give me a good estimate of what the night's vote was (certainly within a half percentage).  I do wish I had a spy there...

I am currently finishing off my degree in mathematics, and I have to say that you and Stock both are a little correct, and both a little wrong.

While your procedure should provide a satisfactory estimate of the numbers, you must remember that the third you are removing is an average of the prior three days.  But what Nanos actually drops off is the first of the three days (not the aggregate numbers).

Let's say that Nanos samples a number of people each day and let's say that the NDP numbers for each of those days is y0, y1, y2, y3,...  The polling results that Nanos reports are the average of three days' polling.  Now let's say that you have  Nanos' numbers for 4 consecutive days, and denote those 4 days of polling as x2, x2, x3, x4,... We can now set up a system of equations as follows:

x2 = 1/3 (y0 + y1 + y2)

x3 = 1/3 (y1 + y2 + y3)

x4 = 1/3 (y2 + y3 + y4), and so on

It is clear from the above that we are dealing with a set of n equations (with known x values) and n+1 unknowns (all of the y's).  Since you are a math teacher, you will know that is impossible to find a unique solution to a set of n equations over n+1 variables.  

 

 

Of course, you are right.  I cannot calculate it exactly.  Which is why I said it was an estimate within about 1/2 a percentage.  But, I can do a good estimate.  And, since I've been doing it from the very beginning of the Nanos polling, I can make my guesses a bit more accurate (except in those cases where there is a number so off the wall (an outlier) that it obscures the data - which occurred in the first week).

Unionist

I don't know about all that, but here's a single equation in four variables:

It's my monthly payment on a loan of L dollars borrowed over N months at an interest rate of p. Every month I have difficulty in solving it. Too many variables, I guess...

Winston

Garth Brasseur wrote:

Of course, you are right.  I cannot calculate it exactly.  Which is why I said it was an estimate within about 1/2 a percentage.  But, I can do a good estimate.  And, since I've been doing it from the very beginning of the Nanos polling, I can make my guesses a bit more accurate (except in those cases where there is a number so off the wall (an outlier) that it obscures the data - which occurred in the first week).

I would argue that Nanos got more than a few outliers a little over a week ago (having us at 8% in ON and 13% in BC is preposterous). I would guess (without any real evidence) that the large spike you observed a few days ago was the result of a combination of two factors: the first one being the addition of a fairly good day to the average, and the second one being the removal of a spectacularly bad day.  I guess we'll know when tomorrow's numbers are out...

I tend to agree with you that the G&M is grossly misusing the Nanos numbers to provide commentary on the statistically insignificant daily ups and downs, or as a snapshot of where things stand on any given day.  That is not what tracking polls are meant for; they generally only good for measuring longer-term trends (e.g. the sustained NDP climb over the last week).  If you would like a snapshot of where things are right now, you are better off to trust in the numbers of large-sample size polls taken over a short period of time (Ipsos, ARS, Leger, etc).

Winston

Unionist wrote:

I don't know about all that, but here's a single equation in four variables:

It's my monthly payment on a loan of L dollars borrowed over N months at an interest rate of p. Every month I have difficulty in solving it. Too many variables, I guess...

N, p and L are known (but the sarcasm is appreciated).  :)

Basement Dweller

Those Montreal numbers are mind-blowing. If they are true, stop talking about a handful of seats. The NDP will win at least half of Montreal's seats.

Northern-54

 

I would argue that Nanos got more than a few outliers a little over a week ago (having us at 8% in ON and 13% in BC is preposterous). I would guess (without any real evidence) that the large spike you observed a few days ago was the result of a combination of two factors: the first one being the addition of a fairly good day to the average, and the second one being the removal of a spectacularly bad day.  I guess we'll know when tomorrow's numbers are out...

 

My crude method showed a day when the NDP was at 22% on the big jump day.  I could refine it to 22.3% when taking out the previous guesstimate from three days before (as the previous week had no outliers like in the first week or so) rather than dividing it by 3.  I suspect that the 22.3% is within a tenth or so but I cannot prove it so I went with the 22% which is more "conservative".