Posted on 10/16/2006 5:13:34 PM PDT by AntiGuv
I thought this one was looking pretty good.
I"m saying that tongue-in-cheek. I haven't seen too many FReepers actually involved in polling and prior to realizing VeritatisSplendor was involved in such activity my opinion of pollers and those involved was rather low.
That being said I still believe we're going to surprise the rats. Many conservatives absolutely refuse to be polled per my feelings about it above. I've asked several pollers not to disturb my family during dinner time ever again. Gallup called one evening and I stated this and the guy said "You don't want your opinion to count?" I stated that I wanted my opinion to reflect WHEN AND WHERE IT COUNTED basically on election day. I don't trust Gallup WAD anyways not to slant it their way.
So it's nice to see the good guys have their hands in this, maybe once in a while we'll get untainted results that's actually useful and not simply a propaganda tool. I just didn't want to type all this last night.
Have a good one, coffee is ready and the need is a might powerful this early AM. : )
Judging by the Excel spreadsheets at the SurveyUSA website, you do damn good work. Thanks for that analysis. Now I've got something to judge other polls by.
Such as why compare a weekend poll with a more accurate weekday poll and also what is the response rate of the poll(i.e how many phone call have to be made to get 1 response).
Please provide a reference showing that "weekday polling" is more accurate than "weekend polling".
Our polls have widely varying response rates, depending on the geography being polled and the source of phone numbers.
You can see documentation on our website here:
http://www.surveyusa.com/respondentcoop.html
A cooperation rate of 50% (50% of the people we reach agree to take the poll) is as high as we ever get for RDD polling, but it typically averages in the 20's, as you can see from the data I linked.
If response is poor, we have the capability to improve it by using other methods than RDD polling (lists of registered voters, for example).
Theoretically, the low cooperation rate is by far the most serious problem with telephone polling, and it affects all pollsters. The polling industry constantly worries that nonresponse will introduce systematic biases, but the accurate track record of the major pollsters shows that, fortunately, this usually doesn't happen, though it is always a possibility.
JMO, that is a Clintonian statement if I ever heard one.
Also JMO, you have a vested monetary interest in providing and selling polling to such outlets such as old line commerical local television stations and in that transaction it is not out of the realm of possibilty that the management of that TV station wishes to have a poll that makes good "idealogical copy" for them.
Look I am not saying that you should be banned from selling your polling, but I do think that I should also not be tarred and feathered when I question news shaping outlets on their polls and the outlets they contract to do those polls, given their previous biaes.
That's the new media world and it seems that the old media is getting miffed at the changes.
Be careful, you're getting close to the line of impugning my integrity.
Sometimes some of the TV stations we work for exhibit bias in their news coverage, sure, but I *never* repeat *never* change the way I analyze polling data in order to make the numbers come out the way station wants them to. I am graded on the accuracy of our polls as compared with actual election returns. Furthermore, the stations, by contract, are not supposed to "spike" polls they don't like, and we post the polls to our website the same night anyway so they can't keep the data from becoming public.
And what is "Clintonian" about my statement? Go to our website, there's a huge amount of data there on the accuracy of polls (both ours and everyone else's). The emprical fact is that the theoretical problems related to nonresponse have not prevented pollsters from being pretty darn accurate. This may change in the future, but don't pretend the evidence doesn't exist.
Thanks for your reply.
I think GOTV is overrated by pollsters, your criteria for "likely voters" really includes the effects of the historical GOTV efforts.
It's just that the electorate is unformed until the election. Until then you're trying to predict the group you're sampling, instead of sampling a group. That's why polls get more accurate ( and usually more favorable to Republicans...) as the time of the election nears and the electorate forms.
I think this will be a breakthrough year for automated polls- one way or the other!
I'm not impugning your integreity, just giving my opinion about polling companies and the media outlets that hire them. It's no different than giving my opinion on the difference between a GM and Ford or their faults, so don't get so defensive.
Thatr said I find fault that you did this poll on a weekend when more people are out and those who actually answer the poll, by your roundabout answer it is 1 in 5 or about a 20% response rate, will be more likely "democrat" voters(i.e 1 land line, no caller id, or voice mail, etc.etc.)
I also take objection to your characterization that there is a big shift, especially since this was a weekend poll.
You can say I am impugning your integrity if you wish, but I find you "findings" hubristic and arrogant and trying to shape opinion of an election.
Albuquerque Journal "Bernalillo County Clerk Mary Herrera was unaware that a man hired as elections coordinator by her office was under investigation for embezzling money at the Grant County Clerk's Office.......
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.