Diary

Polls Deeply Flawed, Often Wrong

The presidential polls are tightening!

There are two kinds of polls: Polls run by the media to make Democrats look good in order to influence the election (Washington Post, NBC, CBS, CNN etc.); and a small number of polls that actually want to gauge the public and to call the election accurately (Battleground, Zogby etc.).

Next year, nobody is going to remember how far off the mark the Washington Post or CBS poll has been. But many people will remember which poll most accurately predicted this election. So there is a competition to get it right, and serious polls are tightening in order to get a legitimate outcome and maintain their credibility.

Today there is much discussion and consideration of the Investor’s Business Daily poll which was closest in 2004.

Here are some examples of bad and biased polling:

*One final poll said that John Kerry in 2004 was going to win Ohio by 4.5 points, yet he lost it by 2.5 points.

*A final poll in the 2008 New Hampshire primary said that Obama was going to win by 10 points and he lost by 2.

*The last Gallup poll in the 1980 presidential race showed the race getting close, yet Reagan defeated Carter by 10 points in the popular vote.

*Polls in 2002 and 2004 said that Republicans Elizabeth Dole in North Carolina and Lisa Murkowski in Alaska clearly were going to lose their US Senate races, but both won.

*Exit polls in 2004 taken as voters departed their voting places were touted heavily by the media on election evening showing John Kerry with a big victory. They were very wrong.

There are countless more examples, as well as many examples of fraudulent media practices like the media calling Florida for Gore on election night in 2000 before voting even had finished in all parts of Florida and in much of the nation. Gore ultimately lost Florida and the presidency.

To understand polling, there is a good book out by David Moore called The Opinion Makers. Moore should know about polls. He once was a senior editor at the Gallup Poll and now is a senior fellow at the University of New Hampshire Survey Center.

In a September C-Span appearance, Moore said that polls were established in the 1930s based on George Gallup’s belief that “the public ought to be part of the discussion on any given issue”. Since voters elect officials to do their legislative work, the public participates primarily through its vote and often there is no further involvement. So Gallup started polling.

Polls have evolved to the point today where there are more and more of them (go to pollster.com and see many) and they are fluctuating more wildly than ever this year. That is because first, more organizations want a piece of the action but are not necessarily dedicated to accuracy or equipped to deliver it; and second, the media this year are more biased than ever before and are going further into the tank for Obama.

So when a national poll says that Obama is ahead by 2 or 6 or 10 or 13 points, it is important to remember that these same pollsters assured us in October 2007 that the 2008 presidential race was going to pit Hillary Clinton against Rudy Giuliani.

Moore said that this mistake arose because the pollsters never really conducted any kind of rational survey to arrive at their conclusions. He indicated one major 2007 poll that didn’t call ‘decided’ and ‘likely’ voters in early-primary states like Iowa or New Hampshire whose outcomes often swing entire primaries. Instead it surveyed generic Republicans and Democrats across the country.

And that poll found that 75% of Republicans had not made up their minds, but of the remaining 25% that had, Giuliani led with 5% of the vote and the other 20% was broken down among the rest of the field. So Giuliani was declared the front-runner with just 5% of the vote from one nationwide sample.

On the Democrat side, it was 69% undecided, 17% for Hillary and 9% for Obama.

Moore said that pollsters “did a disservice to the democratic process” and “affected the democratic process” by their techniques, a pretty damning condemnation from an expert on polls.

Pollsters also get in trouble by asking “if the election were held today” questions, which put people on the spot. Because many voters are undecided up until election day and need to hear more before casting their ballot. And so their answer is hardly an accurate reflection.

Pollsters on public-policy issues also cause problems by asking what Moore calls ‘forced opinion’ questions, where people are asked if they believe this or that, even though they may know nothing about the subject. So often “pollsters give information to people, and then poll them” on that information, said Moore.

How?

Pollsters may inject lead-up statements into the poll like saying “As you know, the US maintains a terrorist detention center at Guantanamo Bay” for people who may not know about Guantanamo. And then the pollster may ask a forced-opinion question like whether the center should be closed. Even completely uninformed people may then offer a response, which is recorded as part of the poll.

Moore also said that good polls should include a crucial follow-up question like “Would you be upset if the opposite outcome came about?” because that would give pollsters more insight into whether the person really knows much about the issue, and into the respondent’s level of commitment to his/her answer.

Moore called the media “unimaginative” in their polling, and in their habit of using issues fed to them by the Bush administration.

On October 26, Tom Brokaw on *Meet the Press *virtually taunted John McCain with polls showing Obama way ahead of McCain, even by a whopping 39 points on the economy.

Yet NBC and other Big Media polls are known to excessively weight Democrats in their polling and to do other things to produce a pre-ordained outcome. Even the time of the day and the time of the week will affect a poll because conservative Republicans and liberal Democrats often lead very different types of lives. For instance, unionized public school teachers or welfare recipients who lean very liberal Democrat would more likely be at home by 4 o’clock in the afternoon than conservative corporate executives or rural factory workers.

Conservative Republicans by nature also are more likely to be reluctant to respond to a poll in the first place because they are less likely to be tolerant of someone snooping into their private opinions, especially people from the Big Media. When conservatives hear the words “Washington Post” or “CNN” on the other end of the phone, they are more likely than others to hang up. Therefore “Republicans” who are included in polls are more likely to be liberal.

Moore added one interesting fact, that one of the many reasons that polls don’t call cell-phone users is because the respondent has to pay by the minute to talk to a pollster and often won’t.

Now imagine that Brokaw and NBC reached their conclusion about Obama’s 39-point lead on the economy by asking one of the following lead-in questions, as Moore suggested they often do:

“As you know, Barack Obama is closely associated with Franklin Raines and James Johnson, two figures involved in the collapse of Fannie Mae. So do you think Obama is better equipped to deal with the economy than John McCain?”

Or how about this:“As you know, John McCain warned repeatedly in 2005 about the coming collapse of Fannie Mae. So do you think McCain or Obama is better qualified to fix our economy?”

That would produce a giant shift right there.

Meanwhile Brokaw unintentionally spilled the beans on the bias at NBC by essentially telling McCain that Obama cannot possibly be an economic redistributionist because he is associated with investor Warren Buffett and Reagan-era Federal Reserve chairman Paul Volcker. Because the other side of that story is that Brokaw and his media friends have done everything in their power to minimize other associations in Obama’s life like radicals Jeremiah Wright and William Ayers.

So why wouldn’t NBC & Co. do the same thing with poll questions?

Polling expert Moore said bluntly that “polls don’t tell the truth about elections”.

He also said that the so-called ‘Wilder effect’ – named after the first black governor of Virginia Douglas Wilder – in which white voters were said to be unwilling to tell pollsters honestly if they would vote for a black candidate, “doesn’t pertain anymore” because black candidates are much more common today than they were when Wilder ran in the 1990s.

Moore said that if he were polling just the state of New Hampshire today for the November 4 presidential election, that he would ask the following questions in the following way to get an accurate reading:

Question #1: Do you favor the Democrat ticket of Obama/Biden or the Republican ticket of McCain/Palin?Moore added that he would reverse the question, putting Republican first in every other call in order not to influence the poll with the order of the question.The answer to Question #1 would decide the poll, according to Moore. But he would ask follow-ups including:

Question #2: Are you firm in your decision or not?

Question #3 (if #1 and #2 do not give an adequate answer): Are you undecided or are you leaning to one candidate or the other?

This sounds like a fairly good method, and questions #2 and #3 would provide information for the ‘internals’ of the polls which detail how the poll was conducted. Moore warned, however, that ‘internals’ can be useful to poll analysts but also can be manipulated for effect.

Moore finally dropped a bomb about polling, saying that 80% of the randomly-generated land-line phone numbers called by pollsters belong to people who cannot be reached, who refuse to participate or who do not respond. And perhaps that is the most damaging statistic of all. Because increasing numbers of American do not believe polls and do not want to participate in them.

And it probably is no stretch to say that most well-informed Americans today are probably somewhat suspicious or highly suspicious of polls, leaving one wondering who exactly is being polled.

Those facts alone should tell us that the polls are increasingly untrustworthy.

And that the only real poll is on election day.

Please visit my website at www.nikitas3.com for more.