Posted on 09/18/2008 5:42:27 AM PDT by sunmars
2 new polls in 2 days have tried to boost Obama, however when you look deeper, theres a better answer here, huge oversampling of Democrats.
http://www.surveyusa.com/client/PollReport.aspx?g=edc4d070-fb0f-43a8-b19c-6b32c3ad36f7
However you look at the sample.
Its 48% Dem, 37% Republican and 13% Indie.
So Obama didnt much to swing him to 52% did it. Theres no way that sample is a good sample.
Again also look here at the Poll posted 2 days ago for Ohio
http://www.surveyusa.com/client/PollReport.aspx?g=43154ee9-1bca-4042-84df-131e90cc13ad
It has McCain leading 49 to Obama 45 however i suspect the lead is larger.
The sample was 44% Dem and 35% Republican and 17% indie and McCain still way ahead, so either a huge amount of dems defected or McCain took 14% of those indies.
It would however seem there is a pattern here among SUSA polls that are they are boosting Obama by way oversampling Dems.
"Show me just what Mohammed brought that was new, and there you will find things only evil and inhuman, such as his command to spread by the sword the faith he preached." - Manuel II Palelologus
But isn’t SUSA a Republican pollster?
Rush called this 10 days ago when he said the polls will show McCain with an early lead, then Obama will go ahead so the MSM can say Obama is “surging”.
Note to MSM: Americans are “on” to you. You have become more than irrelevant.
no
Doesn’t make sense. Pollsters certainly understand this issue and adjust for it! On the other hand, what they cannot adjust for is that 17% or some high number of people do not have home phones (only cell phones), and do not get polled. These folks are young Dems. Tell my why this is not true?
Gee, what a surprise.
Do poll results really influence others? Did it work in 2000 and 2004?
Most polls are weighted to ensure a certain result...The ONLY poll that really matters is the one on NOV 4th....
Its good that we know and can diseminate the information that pollsters are trying to influence with dodgy numbers.
I’ve read that pollsters take that into account (people without hard wired phones).
Benjamin Disraeli
Good.
It keeps the Obama supporters placid and the Republicans lean, hungry and alert right to the end, or, at least, until the debates...
Let the Democrats think they are winning up until 7 pm EST November 4th.
Think you have to look to see if they’re weighting or artificially changing the ratios of calls.
If the calls are random and the people are self identifying then this may be the case.
Unlikely the spread is anywhere near 12-13% more Dems.
Particularly in Ohio or NM.
I’ve noticed before, the MSM/polls usually wind up saying it’s too close to call. If their favorite loses they can scream fraud or bias. And they keep selling papers or keeping viewers glued right up to election day and beyond.
I am pretty sure that New Mexico has been since at least Kennedy, if not further back, a heavily Democrat state.
Correct; that is what happened in the last two elections. Do you remember the MSM calling the race for Kerry at around noon. They said Kerry was winning FL by large numbers. Fl wasn’t even that close in 04 and I bet with the large Jewish vote against BO it will be even better for McCain.
Sampling error is a big issue in polling. The only way to account for it is by using a strict quota system based on demographics and party affiliation. This costs more and takes longer. The results can be weighted, a common technique in business marketing research.
The challenge is to determine the correct weights. In business, researchers refer to census data. Even though it’s out of date, it changes little from year to year.
In politics, party affiliation and preference can shift enough in a few years to impact the results. So what do they do? Therein lies the rub.
You state that phone surveys undersample younger voters who lack land lines and rely on mobile phones. That’s true, but there’s historically less of a tendency for younger voters to actually turn out, no matter what their stated intention.
Phone surveys also undersample more affluent voters. Upper middle class voters use caller ID and answering machines to screen calls. They are less likely to answer calls from unknown numbers, such as research companies. They are also more likely to be away from home in the evening, eating out or traveling on business. These voters skew Republican and have a high turn out.
Bottom line, any one poll can be off more than the statistical margin of error if sampling bias creeps in. Look at multiple polls and watch for trends. This is problematic in state polls, such as New Mexico.
The candidates’ internal polling, which we don’t get to see, is probably the most accurate. You can tell which states are in play according to their polling based on the time, energy, and investment from the campaigns.
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.