Thursday, November 19, 2009

EKOS Schedule Change

As reported here, EKOS will be changing their schedule of the release of polls. Rather than reporting on the national voting intention every week, they will be reporting every second week, but with larger sample sizes. They'll also be reporting on questions requested by CBC viewers, which could be interesting some of the time.

Unfortunately, since this seems tied to CBC News Network's Power and Politics, I imagine that in the future the info won't be released until the end of whatever day they're released on.

So, this reduces what is already becoming a thin roster of active pollsters. The larger sample sizes, however, will be helpful.

I had based my projection-update schedule on EKOS, so I'm not sure what to do at this point. I'll probably continue to update at the end of every week, but there was only one Nanos poll released over the last week so I'm not sure if it is worth it.

Thoughts?

23 comments:

  1. I guess you might as well do an update even if there is just one poll since you do drop off older polls etc...

    Its funny how in late September and early October when there was such a flurry of election speculation, it seemed like there were 4 or 5 polls a week coming out. Now not so much is going on politically so there is less reason to poll so frequently. I'm sure that if we get into another very politically charged period, Ekos will find that being in field over a two week period is too long etc...

    Its interesting how like heroin addicts we now expect our almost daily fix of polls, but it wasn't so many years ago that only three or four polls would come out during an entire election campaign! Now we get that many PER DAY!

    ReplyDelete
  2. Maybe we can convince ekos to at least give us a little poll ... you know, just to tide us over ...

    I need a fix, man!

    ReplyDelete
  3. It's true, I was inputing last week's PDF address with the date change to see if they had uploaded their latest poll. I was dying to see who had gained 0.6 points.

    I'll probably update tomorrow.

    ReplyDelete
  4. I know what you mean DL. Like a junkie, I check about five different political blogs every day plus two or three newspaper sites a day ... while I should be doing homework (and I don't even study anything related to Canadian politics!)

    ReplyDelete
  5. The existence of the consistent Ekos polls has been something new in Canada, it has been rare to have this many polls, it is definitely and upswing from the 2006 - 08 period or the 2004 - 06 period.

    If there is not significant danger of an election any time soon, it does not make sense to put a lot of resources in to publicly released opinion polls.

    It will be interesting to see how large the Ekos poll numbers will be, will they simply be aggregating all the nights for 14 nights into one release? Will they do be calling a lot one week and not at all the next and give us better daily numbers?

    ReplyDelete
  6. Eric,

    I'm highly annoyed by this development.

    And unless they're actually calling more people over a one week time period then its NOT an increased sample size.

    Jeez, if I wanted to I could meld two weeks worth of results from them myself.

    ReplyDelete
  7. Obviously, CBC didn't want to pay for a survey every week.

    ReplyDelete
  8. I kind of expected the number of polls to drop off after all the election speculation disappeared. I would say do the update as often as you feel like it. If it takes a lot of effort to do it every week, then do it every other week or whatever. What I might suggest is picking a topics or scenarios you're interested in (like those 4 alternate reality scenarios) and producing more content like that until the polls pick up. But that might be a great deal of effort.

    ReplyDelete
  9. BC Voice of Reason19 November, 2009 14:15

    "Obviously, CBC didn't want to pay for a survey every week."

    ----------------

    OR CBC was tired of seeing the polls showing the CPC close to majority.

    The pollsters and the media that pay them are really messing with your methodology.. your long term seat predicitions are tied to having consistent polls made availabe over time.

    It seems that now that ALL the pollsters are all showing the CPC and Harper doing so well in the eyes of Canadians the media is finally following through with the concept that polls should reflect opinion rather than drive it.

    This was not the case when the polls supported the media's position that Mr. Ignatieff was the next great Liberal PM.

    There likely will likely be a lot less poll results published / discussed until the pollsters get the results that the media wants to talk about.

    As this means that with the pollsters are in their own mini-recession, how long will it be before they can come up with poll results that the media paying them wants?


    Are you going to adjust your model and reduce the impact of the Liberal / Ignatieff favourable polls that were being published 6 tiems a week?

    ReplyDelete
  10. --- "OR CBC was tired of seeing the polls showing the CPC close to majority."

    Give me a break.

    --- "The pollsters and the media that pay them are really messing with your methodology.. your long term seat predicitions are tied to having consistent polls made availabe over time."

    Well, I can't assume to know what is going on between polls. I can only use what we have. And each poll is only a data point, not canonical truth.

    --- "This was not the case when the polls supported the media's position that Mr. Ignatieff was the next great Liberal PM."

    Speculation.

    --- "There likely will likely be a lot less poll results published / discussed until the pollsters get the results that the media wants to talk about."

    You're half-right. Polls showing the same thing week after week aren't interesting for the media, especially when it is the government in power that is leading. It's like printing a story that the sky is still blue. They're paying for the polls to sell papers, after all.

    When a poll shows change, or a poll shows that the opposition is way ahead (which creates conflict), that sells papers. It has nothing to do with which party is where in the results.

    --- "Are you going to adjust your model and reduce the impact of the Liberal / Ignatieff favourable polls that were being published 6 tiems a week?"

    No, unless you think I should lessen the polls of the last few weeks which showed the Tories in the lead too.

    ReplyDelete
  11. Eric,

    You can't argue that this is GOOD news for Iggy though.

    He needed a breather. Every poll that came out caused a prolonged discussion about what he was doing wrong.

    Now, i'm not going to go as far as suggesting that Peter Donolo has connections from his media/pollster days and made an ask. Maybe this is a happy coincidence for him.

    Donolo definetly gets a break, he doesn't have to face questions like "he's not working! He hasn't turned around the numbers! Fire him! Fire the leader!" every week.

    ReplyDelete
  12. --- "You can't argue that this is GOOD news for Iggy though."

    I assume you mean BAD, in which case I'd agree. Although, if next week's EKOS poll shows the Liberal up two or three points, he could've used that ego boost a week earlier.

    --- "Now, i'm not going to go as far as suggesting that Peter Donolo has connections from his media/pollster days and made an ask. Maybe this is a happy coincidence for him."

    SC worked with CTV, while EKOS works with CBC. Weren't people telling us here how the pollsters hate each other?

    -- "Donolo definetly gets a break, he doesn't have to face questions like "he's not working! He hasn't turned around the numbers! Fire him! Fire the leader!" every week."

    One wonders, though, how much of the doom and gloom had to do with a lack of a better story for the pundits.

    ReplyDelete
  13. BC Voice of Reason19 November, 2009 15:12

    No, unless you think I should lessen the polls of the last few weeks which showed the Tories in the lead too.

    -------------

    That might make sense... at a sacrifice for MoE and sample size. I think that the sample size over time needs to be normalized... you have a choice in giving more weight to the limited sample size that we will be getting over the next month or so by reducing the weight of the heavy sampled period.

    Either option is not ideal... but is better than limiting the impact of public opinion for this 2 month period on your overall seat projections.

    You may have already chosen to go with giving the small number of samples in November equal weight (by month) to the heavily sample September??

    ------------

    I find it interesting that you dismiss my CBC argument out of hand at the onset "Give me a break" but by the end of your reply agree that the media does not want polls that do not provide what they want to talk about.

    As a former consultant I was able to gain considerable credibility by giving my initial situation analysis with the caveat that would be able to document 20 reasons why it was wrong given sufficient time.

    However there was immense pressure to give the client information that they wanted to pay for.

    The pollsters will be looking for the 20 ways to slice the info to make it saleable. Especially as there is no way to validate their results without a general election it is quite simple to slant/bias the polls to provide saleable material.

    ReplyDelete
  14. --- "you have a choice in giving more weight to the limited sample size that we will be getting over the next month or so by reducing the weight of the heavy sampled period."

    It will be reduced in weight as time goes on.

    --- "Either option is not ideal... but is better than limiting the impact of public opinion for this 2 month period on"

    The more polls I have, the more accurate my information. That we don't have a lot of polls now means I have to rely on older polls. If I give a lot more weight to the odd polls now, their MOE will make my model inaccurate.

    --- "I find it interesting that you dismiss my CBC argument out of hand at the onset "Give me a break" but by the end of your reply agree that the media does not want polls that do not provide what they want to talk about."

    That's not what I said at all. I said the polls were boring in their monotony, so that is why media outlets aren't paying for them anymore. It isn't because the Tories in particular are ahead.

    --- "The pollsters will be looking for the 20 ways to slice the info to make it saleable. Especially as there is no way to validate their results without a general election it is quite simple to slant/bias the polls to provide saleable material."

    If that's true, they'll all cancel each other out in the end since slanted polls for either side have a market.

    ReplyDelete
  15. Polling companies don't do surveys as a public service and they are not charities. They have to charge for what they do. if you want more polls, call any polling company and offer to pay personally for a survey. they will be more than happy to do it!

    ReplyDelete
  16. Indeed. EKOS is really the last to see a downturn in business. We haven't seen the Toronto Star, the National Post, or the Globe and Mail pay for a poll in a little while, which is what they used to do.

    Probably has more to do with money than politics (as in all things).

    ReplyDelete
  17. Eric,

    "I assume you mean BAD, in which case I'd agree. Although, if next week's EKOS poll shows the Liberal up two or three points, he could've used that ego boost a week earlier."

    OK, assuming that Ignatieff is going to turn things around and Donolo really does work out then a weekly progression of good news would be GOOD for Ignatieff.


    So I can put you down as making that prediction ?

    Otherwise stagnant or declining numbers on a week to week basis is BAD news for Iggy. And therefore switching to every two weeks is most certainly a breather.

    ReplyDelete
  18. Huh? I merely said that if the polls continue as they do, the fact that they come out less frequently is good news for Ignatieff. But if the polls don't continue and change to his benefit, then it be bad news.

    In other words, it's only good or bad news based on what the numbers are. You're assuming they will continue to bad numbers. We don't know whether that is the case or not.

    ReplyDelete
  19. I have to wonder how much cost really matters here.

    There are very real costs for most pollsters since they have to pay people to do the telephone questioning.

    But, given that ekos uses an automated system, their system may in fact be very inexpensive.

    ReplyDelete
  20. Operating costs would still be significant. The fact that they may be more inexpensive might be the reason they are the only ones who poll so frequently.

    ReplyDelete
  21. We also don't know whether Ekos even charges the CBC anything. They may have done it for free for a period of time just to get publicity and now they just don't feel like giving away so much research so frequently.

    Anyways, once every two weeks strikes me as more than enough, its not as if anyone else is polling more often than that.

    ReplyDelete
  22. Eric,

    Sorry, I went back and re-read my statement.

    "You can't argue that this is GOOD news for Iggy."

    Should be "you can't argue [with the fact] that this is GOOD news for Iggy."

    I took your correction as disagreement - that you were arguing the two week gap somehow harmed Iggy.

    ReplyDelete
  23. Hi Eric:

    I'd do my predictions every two weeks now as that's the polling data your getting. Thanks again for a great site and the work you put into it. I'm a political junkie in my late forties and have been since I was 8 or 9 years old. I love the site to debate others.

    ReplyDelete

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.