I should add a footnote to this morning's post on recent Presidential news with some local twists: it's a good idea to remain wary of individual poll numbers.
The national polls are interesting, not to mention prevalent. Yet, I'm still wary of reading too much into them given the effect early voting states tend to have on final results. Taken in combination, however, such as in the Real Clear Politics average, they can provide a worthwhile peak at the state of the race. More importantly perhaps, aggregate compilations of even more polling data, like that at pollster.com, can show additional insightful into the long-term developments in the race.
That latter link is especially noteworthy for the trends in the Republican contest. Giuliani has been slipping since his announcement bump several months back. Meanwhile, McCain has been steadily dropping since the beginning of the year. Thompson has soared in recent weeks, Romney keeps steadily rising, and Gingrich has begun to drop (no doubt thanks to Thompson's entry).
Such charts via Real Clear Politics or Pollster are infinitely better than individual polls, whose sample sizes are sometimes atrocious. Back in 2004, pollsters were generally sampling close to 1000 "likely voters" or more for surveys in the closing weeks of the election. Recent Presidential polls are dreadfully weak in methodology by comparison.
Take just two recent national polls. Today's Gallup poll tested the GOP race on only 393 Republican and Republican-leaning "adults" nationwide. A soon-to-be-released poll by RT Strategies for respected pundit Charlie Cook sampled a mere 330 "registered voters" who are either Republican or "Republican leaners."
In Presidential general election polls, pollsters work very hard to get their "likely voter screen" correct, to ensure the poll surveys with the most possible accuracy those who are actually going to vote. What then is the true value of primary polls that don't dig deeper than "adults" or "registered voters," when as little as 10% of adults actually participate in Presidential nominating contests?
State polls likewise are a tricky business, with many different pollsters at work - some of higher repute than others. Looking for trends in rolling averages of polls (and especially trends in regular polling by the same pollster) is one of the best means by which to gauge how competition in various states is unfolding. See the "state polls" section on the left side of this page for an excellent resource accordingly.
Again though, beware of the outliers. Take a recent Datamar poll in California, showing a near reversal of Rudy Giuliani and Mitt Romney since February. The same company has had numbers generally in line with other results in Florida, but this poll is just too disparate to be believed.
Likewise, in all fairness, the Mason-Dixon poll in South Carolina noted this morning also might be viewed as an outlier for how low McCain's numbers were. The problem for his campaign is the results fit the current story line of their guy's troubled candidacy. Tough to escape that in this case.
For good or for ill, poll numbers and fundraising results will be among the limited means by which we can assess the state of the Presidential race through much of the summer. Just remember to be skeptical first and swoon later.
Posted by Eric Earling at June 18, 2007 08:55 PM | Email This1. With the advent of cell phones a large portion of the population has abandoned land lines and are therefore left out of the 'scientific' polls.
2.They also have historically favor people who are more likely to be at home in the evening.
3. Lastly they favor people who have kept the same phone number.
All three of these bias hurt young peoples representation in the polls. The last two are somewhat made up using statistics, but the first one is a new one that has not been adjsuted for by any major polling place.
Online polls while certainly not scientific and not contained geographically, over represent young people and also politically active and motivated people. Both of these groups describe Ron Paul supporters to the T. Because of this I think people are going to be surprised by Ron Pauls campaign coffers come July as well as his primary votes come January (or is it december now?) :)
Posted by: Travis Pahl on June 18, 2007 10:06 PMAlso, FWIW, it is not really possible to adjust for cell phones. Some are trying to do it, but it's got the same fundamental flaws as land lines, but worse: they are even more heavily biased toward specific types of voters. Plus, it's illegal to make unsolicited calls to them, so you have to opt in, which makes the sample self-selected, which completely invalidates the results.
Online polls? Yes, they are unscientific, but it is far worse than you seem to think: you appear to believe that these polls represent what young people are actually thinking. They clearly do not.
It's not that the polls overrepresent young people. That is not the problem. The problem is you cannot even trust them as reasonably representative of ANY group, INCLUDING the group of people who actually vote! Not only is the sample self-selected (again, which completely invalidates it), but they have no significant protection from people who have voted multiple times.
Even if the problem were only self-selected samples, you would need to accept the fact that Paul supporters might simply be far more inclined to vote in online polls than supporters of other candidates. We saw the same thing in the exit polls, to a lesser degree, in 2004: Kerry supporters were far more inclined to participate in the polls, so the polls were skewed toward Kerry.
But while the problem of self-selection completely obliterates the validity of the poll for any use whatsoever in determining who favors which candidates, it gets even worse because of the fact that there is not a reliable 1:1 ratio of voter to vote.
I know a little something about this. Maybe you've heard of my exploits before (and that wasn't the first time!). I've also written online polling software. And while some (like MLB's) has been significantly improved over the years, most software still stinks, and are extremely easy to exploit.
So all you know from the online polls is that enough people cared about the polls to bother voting, and voting for Paul more than other candidates, and that alone means the poll is worthless, but it is compounded by the fact that you have absolutely no idea how many people voted how many times. As far as you know, one person could have voted 40,000 times.
We saw the exact same thing in 1996, and even in 2000 to a lesser extent: Alan Keyes cleaned up in the online polls. He didn't do so well when the actual vote came along, though.
Posted by: pudge on June 19, 2007 08:29 AMNow back to reality. Outside of polls, there are tons and tons of signs that Ron Pauls campaign is strong and growing.
He is filling auditoriums with 500+ people. He has more supporters outside the debates than the other candidates combined.
He has had more meetups than any other candidate (as well as members). His meetups are actually attended locally where as the Obama group can not even get more than 5-6 people at them.
When I am driving with my Ron Paul 2008 bumpersticker I get honks followed by thumbs up from strangers.
He has raised roughly 5 million in the past three months.
All these things are happening outside of polls. But you are probably right. It is probably the work of some multi millionaires in their dorm room. They have donated 5 million secretly by using hundreds of aliases. They follow me around and honk and give me thumbs up. They hire illegals to go to meetups and organize Ron Paul Rallys. They use elaborate systems of smoke and mirrors to make it look like Ron Paul filled the auditorium. It has to be the case that these internet geeks did all this because the alternative (that Ron Paul actually has alot of support) is just crazy talk!
Posted by: Travis Pahl on June 20, 2007 08:18 PMMaybe you should read a book on social science methodology?
You're treating these worthless polls like the old SNL "Change Bank" sketch, where they only make change, and charge no service fees of any kind. "People ask us how we make a profit. One word: volume."
As to the rest ... no, he does not have more supporters outside the debates than the other candidates combined. Who invented that number? Most voters and politically active people do not CARE about meetups. I've been heavily involved in the Internet for many years, and I've never even been to a meetup.
$5 million in three months? Well, be still my beating heart. He has 1/5th of what Romney pulls in for ONE month.
He has not the slightest chance of winning, he has nowhere near enough resources or support to win, and even if he wins the nomination (which is nearly impossible due to the lack of support) he has no chance to win the general election because the majority of Americans hate the majority of his views.
But you are probably right. It is probably the work of some multi millionaires in their dorm room.
Again you misrepresent me (you do that a lot). I never implied any such thing.
Posted by: pudge on June 20, 2007 11:41 PMYou are suggesting this is not true because you keep suggesting that the polls are being rigged by just a few people. There are plenty of real world examples that show that it is more than just a few people supporting Paul.
The words you think I am putting in your mouth was just a example I made up to put your theory to its logical conclision since you have not wanted to draw it to its conclusion on your own. If you are up to that task, feel free.
Posted by: Travis on June 21, 2007 12:07 PMRight, that's the problem. You have yet to deny this fact.
I have in fact been argueing instead that because they are self selected it is showing something different than just number of supporters but rather a combination of number of supporters and their relative dedication to their candidate versus opponents supporters dedication.
Except that no, it doesn't show that at all. It is not capable of showing that. Even if the polls showed a lot of people supporting him (which they do not), there are many other conclusions that could be drawn other than that they are more dedicated. Maybe the supporters of other candidates, for instance, simply are putting their efforts into other things to get their candidates elected. Things that are actually, you know, useful.
If anything it could possibly show that Ron Paul supporters are dumber, because they waste their time participating in meaningless polls.
You are suggesting this is not true because you keep suggesting that the polls are being rigged by just a few people.
No. That is false on several counts, not the least of which is that I did not suggest it was happening, only that you have no idea whether it is happening (which is, of course, absolutely true); and I did not "keep" suggesting it, as I only mentioned it once.
There are plenty of real world examples that show that it is more than just a few people supporting Paul.
I never said there wasn't. I said these polls do not show that more than just a few people support him, which is absolutely true.
The words you think I am putting in your mouth was just a example I made up to put your theory to its logical conclision since you have not wanted to draw it to its conclusion on your own.
False. You did not take anything I said to any logical conclusion. You falsely implied that I said Paul has the support of only a few people; I never said that. You falsely implied I said there were not a lot of people voting for Paul in online polls; I never said that. You either have a serious reading comprehension problem, or you simply don't care what people actually say and twist it however you feel like it, regardless of the truth.