Are Today's Polls Clueless?

Frisky070802 writes "As noted on electoral-vote, Jimmy Breslin has an interesting article in Newsday on why polls are broken. This is because they poll only landline phones, and a substantial fraction of younger people have only cell phones -- so they hit a biased demographic. If a majority of younger voters tend Democratic, the polls could be giving Kerry a raw deal. Hmm, could this be why two polls released this week vary so widely?"

but what percentage don't have landlines?

[rritterson]

The devil is in the details here. First, of the 168 million cell phones, how many of those are owned by people who have no landline? And of those, how many are likely to vote?

Using my unscientific survey (i.e. my life as a college student) about 40% of 18-22 year olds don't have a cell phone. I would estimate that segment of the population to own maybe ~35% of the cell phones. In the last election we voted at about 36%. Thus, .4*.35*.36*168 million is about 8 million votes that aren't included in the poll. Of those (at the very most). I bet it's 60/40 Kerry/Bush. I don't think it's really large enough to cause a dramatic turnaround in the election, but it is big enough to increase the margin of error in the polls.

On a side note: does anyone know if they survey all of the likely voters in a household, or just the person who answers? (I've never been polled)

As someone who has been in thepolling biz for 3 yr

[LennyDotCom]

I have seen a lot of sloppy polling. You have the big problem of the callers cheating, faking data and all kinds of crap you wouldn't belive. when they say + or - whatever % don't belive it for a minute

The problem is not young people with cellphones

[waynegoode]

...a substantial fraction of younger people have only cell phones--so they hit a biased demographic.

I don't think this is the problem. Demographics like gender, race and political party, preference, etc., are usually corrected for, although I don't know about these polls specifically. They will either adjust the group they poll so that they are half men and half women, for instance, or adjust the weighting of the answers so they are effectively half men and half women. Unless people with cell phones hold different opinions that those with land lines--that is not accounted for by gender, race or political party, etc.--this will not be a problem.

I think the difference is just the inherent inaccuracy in conducting a political poll.

Zogby, anyone?

[Asprin]

HA-HA!

What comes around goes around, I guess.

About 10 or 15 years ago, some dude named John Zogby surmised that the standard political telephone polls we skewed toward the left because their methodologies involved making the calls during the day, when older Americans -- who tended to be more conservative -- were more likely to be preoccupied with activities like working, shopping and running errands. He started company [zogby.com] to prove he was right. Here's his bio. [zogby.com]

My pollers just talked to the first voter

[waynegoode]

I've been polled twice. They wanted to talk to a regsitered voter. I suspect they just talk to the first one, or in one case play recorded messages and record touchtones of the first one.

What a horrible article

[Skeezix]

Any credible argument or salient points are pretty much wiped out by statements such as:

The people who say they want to vote for Bush are generally in the older age brackets, and they don't have as much trouble with the lies told by Bush and his people.

Yeah, because we all know that older people don't mind when a president and "his people" lie to the nation. And clearly everyone knows the president has lied to all of us. It's just that older people don't mind. Huh?

The young people on cell phones appear not to be listening and they hear every syllable. They punch out a number without looking. They are quicker, and probably smarter at this time, and almost doubtlessly more in favor of Kerry than Bush.

Yeah, and we all know that the younger people who are also smarter will doubtless vote for Kerry (probably a direct consequence of their increased intelligence). Only the old, stupid, slow people would not mind Bush's lies and vote for him and "his people."

Older people complain about Kerry's performance as a candidate. Younger people don't want to get shot at in a war that most believe, and firmly, never should have started because it was started with a president lying.

And obviously the older generation will be more concerned with trivial details such as the candidate's "political record" and "performance" while the younger, smarter people don't want to die and therefore don't want to vote for a liar who sends people to their death for a pointless cause.

Gallop poll is dubious

[green.vervet]

The reason for Gallop's very high poll numbers for Bush was based on its bizarre assumptions on turnout. This is well documented in Zogby's critique of Gallop:

http://www.zogby.com/news/ReadNews.dbm?ID=859 [zogby.com]

Gallop assumes for that poll assumes that the turnout on election day will break down as follows:

Total Sample: 767
GOP: 305 (40%)
Dem: 253 (33%)
Ind: 208 (28%)

However, as zogby noted:

If we look at the three last Presidential elections, the spread was 34% Democrats, 34% Republicans and 33% Independents (in 1992 with Ross Perot in the race); 39% Democrats, 34% Republicans, and 27% Independents in 1996; and 39% Democrats, 35% Republicans and 26% Independents in 2000

So Republicans are badly over-sampled and Democrats badly under-sampled, giving systematically biased results. Awful polling, but used to keep Republicans motivated and Democrats depressed.

Younger voters leaning towards democrats

[PIPBoy3000]

This article [washingtonpost.com] may help support my comment. To be fair, that age group tends to be pretty volatile. Earlier this year, I think they were fairly evenly tied. In recent months, the war on Iraq is making a greater difference in that age group, probably due to worries about a potential draft.

Re:but what percentage don't have landlines?

[nelsonal]

The actual data is available from the phone company web pages. While there has been a big stir about lines being swapped from land line to cellular, it hasn't been that big a factor. The bells have about 130 million lines (VZ doesn't disclose second lines from primaries, but that includes business and second lines. There are also almost 20 million lines under things like the MCI neighborhood plan and other CLECs (I don't have data on if those are primary or secondary lines but I'd suspect they are almost all primary lines as CLECs are in dense areas and offer DSL). If you back business losses, UNE-P (CLEC) losses, and the few VoIP losses to date, all of which can be polled or do not represent actual households you are left a fairly constant number of lines over the past two years. Household growth in the US is generally in the 1-2% range so it is unlikely that there are more than 3% of households with wireless and no land line. Also the swing states are generally not the areas that would be as likely to have wireless only users (who would be more likely to be in dense urban areas such as NY or CA) Even if that group split 60-40 for Kerry, I doubt it would throw the election off enough to account for any electoral votes.
Btw, there are about 150 million subscribers or roughly half the population with cellular phones now. In some European countries the penetration rate is north of 80% which is pretty impressive. In several, Mediteranian countries it is north of 100% which is bizarre.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Dewey Defeats Truman...

[Big Sean O]

Lots of people remember the Chicago Daily News headline, but this story harkens back to the 1948 race.

Back in 1948, Thomas Dewey (he-of-the-new-york-state-thruway-fame) was polling ahead of President Truman. No one expected that Truman would win. However, after the votes were counted, Truman won.

Afterward it was discovered that extra Truman support came from urban and rural poor, the people who didn't have phones, and therefore they weren't polled.

There was even a third-party candidate back then: Strom Thurmond, the "Dixiecrat [rotten.com]" who bailed on the Democratic party because Truman had the gall to support civil rights reforms (like integrating the military). "Ol' Lizard King", as I like to call Thurmond, apparently felt it was okay to secretly father children with "Negroes" (although he preferred a different N-word [stromwatch.com]), but southern states shouldn't have to give up segregation.

Of course, back in 1948 you had two decent, qualified people running for president, today we're lucky if we get one.

Re:Biased.

[QSO_Wizard]

The other funny thing is that the last several elections have offered polls which tended CONSISTENTLY several points to the left of reality. I'm talking off by 5-10 points just before the election - and always to the left.

Funny, when I looked at the CNN/Time poll taken a few days before the 2000 election, I see that they predict Bush with a comfortable lead (49% to 43%).

http://www.cnn.com/2000/ALLPOLITICS/stories/10/27/ cnntime.poll/index.html [cnn.com]

This article claims to be in agreement with a CNN/USA Today/Gallup poll that showed an even larger lead for Bush (52% to 39%). A few days later Gore received the majority of the popular vote, so both of these polls were garbage. They leaned way to far to the right, not left as you claim.

I wonder what makes them automatically overlook Kerry's lies?

I don't know about Kerry's lies, but yours were easy to disprove...