(Hint) Online Zogby polls are about as authentic as DU polls.

[Clintonista2]

n/t.

[dsc]

1. I am not so sure of that

If you are refering to the ones he does via email, those could be scientific, they many not be too, but it is possible to do a scientific poll via email.

[Clintonista2]

4. How many "older" folk have email?

Probably not alot, at least not enough to consider the sample "scientific." Older dems also tend to support Hillary, and younger dems tend to support Obama.

[dsc]

7. they can adjust for that

They use what is called a stratified random sample. You take x number of old people and y number of young people. Then you use a factor adjust the numbers back to what they should be to reflect the population. For example if they have 100 old people and 500 young people and need an equal number of both they simply multiply the old people by 5.

[jkshaw]

10. Lirwin, I'm about as old as a person can get ...

... and still use my fingers to type, and *I* support Obama. What's more I don't know what I'd do without email. I've had an account since the 90s.

This older-younger thing really aggravates me. I've been using the computer to write since the early seventies, the Internet since 1994, and have had a web site of my own since January of 2000. And there are plenty of "older" people like me, especially women who've earned a living by way of keyboards.

There ARE Progressive 77-year-olds, you know. We all aren't living in the 50s and 60s.

[Clintonista2]

12. I understand that, but statistically, older people don't use the internet/email

as frequently as younger people do.

[CreekDog]

25. they can't be random, sorry, you are wrong n/t

[dsc]

26. that becomes a matter of degree

People who are willing to answer phone surveys also aren't purely random. When are the calls being made? Who has unlisted or do not all numbers? I admit, it would be way more work, and I haven't a clue if Zogby is doing the work, but it would be possible to get a sample as good as the ones we currently use, using email.

[Captain Hilts]

2. Online polls aren't worth a damn. His regular polling is good.

Online polls are self-selecting and the Internet still skews young and male whereas voters skew older, wealthier, whiter and female.

[madrchsod]

3. that was an online zogby poll?

i did`t notice that..even though i detest hillary i found that "poll" was`t right. i think hillary is still in the "lead"

[Clintonista2]

6. People keep posting these online zogby polls as "proof" that Hillary is slipping

Or as "proof" that Hillary can't win a GE. Online polls are not legitimate.

[bluestateguy]

5. Agreed. His online polls from 2006 were bogus.

There is selection bias and you miss those people who don't have a home computer.

[ccpup]

8. or people who check it as "Junk"

and then delete it without responding.

[Mutineer]

9. And if they showed HRC in the lead

some of you would be celebrating that it was totally reliable.

[durrrty libby]

13. That is silly

[Godhumor]

11. Statistics 101

The only true sample is a 100% random sample. Anything not taken as random is not a true reflection of the population.

Zogby's online poll is reliable for one population--the people who are registered in his online polling. Using that sample, it is fair to say that out of the people who participate in interactive Zogby polls Clinton is not doing well (assuming he used a random sample from those who registered), but there is no credible way to move that conclusion on to the general population of the US. The simple act of registering to participate removes the randomizing element.

This isn't building a poll up or disliking results--it is simple stats.

[Captain Hilts]

14. Random, yes and no. When you call, you first ask to speak to the youngest person...

in the household registered to vote. Some firms ask to speak to the youngest male registered to vote because they're the toughest to get on the phone. You don't necessarily want to talk to the person who answers the phone, or it would be a survey of older women. Women answer the phone and men answer the door.

You get your toughest demographics first - young men, then young women, then work on others.

But, you have to balance the sample to reflect the VOTING population. The voting population is very different from the Internet population. It would take a great deal of adjustment.

[Hawkowl]

18. Not true.

Look, in statistics, the size of a sample relative to the total population, in this case over 9,000 vs the total electorate of about 90,000,000, mitigates to a large degree the need for randomness.

This is true from a strictly mathematical standpoint. Granted you could gain a greater validity, i.e; smaller percentage of error, with a random sample that is much smaller, say between 40 and 1000 potential voters. Statistics are not well understood by the layman and often used to obfuscate rather than illuminate.

[Godhumor]

21. I might be misreading your intention

But, generally speaking, the larger the sample size the less sample error of the means you're going to end up with, which is why Z0gby claims an error of only 1% due to his sample size. However, there is a few substantial problems with this:

First, all polls have a certain amount of bias in them, whether it is that people act differently on the phone compare to in person or to even catching the people in wrong mood. The trick is to adjust for these biases, assuming they are small enough to be adjusted for.

The online survey uses the large sample to reduce the sample error well beyond what a normal smaller poll would be able to do (I'm not sure what confidence level Zogby assigned to his poll, but I assume he used a standard 95%), but that is with the assumption that the biases in this form are a) small and b) can be accounted for and corrected in the final outcome. The argument, and it is a massive one, is that the biases inherent in online polling is beyond the ability to correct and to negate with larger sample sizes. At the heart of it is a rather simple argument:

Online polling can not and does not account accurately for those in lower incomes, as the computer activity and generally educated state of those who respond to the surveys make the sample irrevocably skewed toward the middle and upper class (It also tends to ignore the older voter.).


The fact is that we need to see a regression from Zogby when results are concrete and see just how close his model fits the regression equation.

Finally, the large sample never negates the need for correctable bias, as it will only form a normal distribution if the problems can be accounted for adequately. If that part does not exist, it does not matter how large his sample is--the result will not be a true reflection of reality.

[DemocratSinceBirth]

28. It's Not The Quantity But The Quality Of The Sample

Once people register to be polled it becomes a non probability sample...Only those that registered have any probability of being sampled...The poll ignores everybody that didn't register...In a true probability sample everybody would have the same probability of being sampled...

[PurityOfEssence]

15. Okay. How about Rasmussen's 52% unfavorable for Senator Clinton?

Care to wax on that waning?

[Clintonista2]

16. Legitimate. n/t.

[PurityOfEssence]

22. That made me get up, walk out, get a hat, and return so I could doff it to you

So if this is true, is she electable? More importantly: is it worth the risk?

(Hey, I'm not even getting into the personal revulsion part; this is just sheer practicality.)

[Clintonista2]

24. Answer:

It's currently about a year away from the GE. Unless Hillary is not going to bother campaigning in the GE, I don't see how polls like that can influence anybody. Polls provide a snapshot, and they shouldn't be used to "prove" that a candidate is unelectable, before campaigning has even begun.

[BenDavid]

17. This is a bogus poll. WHY? I emailed

over 50 of my friends across this nation that belong to the Zogby International in which this poll came from. To our amazement of the 50 I sent and received emails from only 5 reported they had received the zogby polling in their email. I find that rather strange, but I do have a conclusion. In prior polling they ask are you a Democrat? Then they ask are you a strong Democrat? I am under the conclusion they ran this poll with it front loaded with republicans and right leaning independents, and thus you have a bogus poll.

Now I do know there was a Gallup poll out that showed HRC winning over all the republicans and that Obama was tied with Rudy.. I understand how the game is played and I would not be to hyper on this Zogby poll....

But the media will use this poll just like they did the abc poll and repeat it over and over again and disregard all other polling showing hRC in the lead from 2 to 10 points.

I like the game of politics but I do not like to play when the deck is already stacked against your candidate.

Ben David

[Hawkowl]

19. Completely disingenuous

That is blatantly not true. Zogby polls have a much better than average record of predicting election outcomes.

[Clintonista2]

20. Proof? Link?

Thanks in advance!

[durrrty libby]

23. lol

[DemocratSinceBirth]

27. They Are Fundamentally Different

They are fundamentally different than your run of the mill internet poll but they still suffer from the same problem...They are non-probability samples...As long as respondents to your poll are self selected; they request to be polled, then your sample is flawed...