Monday, November 1, 2004

previous entry | main | next entry | TrackBack (2)

Tentative answers to some big voting questions

A quick follow-up to my last election post about possibilities not included in the polls:

1) Looking at the latest batch of polls, I notice that some of them include Nader, but I haven't seen any of them include Badnarik (if I'm wrong about this plase post a comment). Again, my hunch is that the Libertarian party candidate will be the equivalent of Nader for disaffected right-leaning voters.

2) Peter Wallsten wrote a story last week in the Los Angeles Times suggesting that the evangelical vote -- a vital Bush constituency -- might not turn out as much as the administration hopes:

An estimated 80% of the evangelical vote went to Bush in 2000. But Bush's senior political strategist, Karl Rove, said after the 2000 election that the president might have won the race against Democrat Al Gore by a comfortable margin had 4 million more evangelicals gone to the polls rather than sitting out the election.

This year, the Bush campaign and conservative groups have made enormous efforts to mobilize evangelicals, a group that includes more than 70 denominations, and which generally sees the Bible as the authoritative word of God, emphasizes "born again" religious conversion, and has committed to spreading its faith and values. Evangelicals are thought to make up about a quarter of the electorate.

In appeals to evangelicals, the president's supporters have pointed to Bush's stance against abortion, his appointment of conservative judges and his support for a constitutional ban on gay marriage. And yet a recent poll found a slight slippage in the president's support.

A poll published last week by the Pew Research Center for the People and the Press found that 70% of self-described evangelicals or born-again Christians planned to vote for the president, down from 74% in the same survey three weeks earlier. That was not only a slight decline, but lower than the 80% to 90% support that Bush campaign officials had been forecasting. (emphases added)

UPDATE: Chris Sullentrop speculates that there's another problem -- the Republican effort to get out the evangelical vote also triggered greater turnout among Democratic-leaning non-voters:

It's possible that Rove and the Bush campaign have turned up a huge trove of conservative nonvoters who were registered to vote four years ago and who therefore aren't showing up in the numbers of new registered voters. Unless that's true, however, the early indications are that Rove's repudiation of centrist politics will backfire. The secret of Bill Clinton's campaigns and of George W. Bush's election in 2000 was the much-maligned politics of small differences: Find the smallest possible majority (well, of electoral votes, for both men) that gets you to the White House. In political science, something called the "median voter theorem" dictates that in a two-party system, both parties will rush to the center looking for that lone voter—the median voter—who has 50.1 percent of the public to the right (or left) of him. Win that person's vote, and you've won the election.

Rove has tried to use the Bush campaign to disprove the politics of the median voter. It was as big a gamble as any of the big bets President Bush has placed over the past four years. It has the potential to pay off spectacularly. After all, everyone always talks about how there are as many people who don't vote in this country as people who do vote. Rove decided to try to get the president to excite those people. Whether Bush wins or loses, it looks like he succeeded.

3) The cell phone vote tilts towards Kerry -- maybe. Zogby has a poll:

Polling firm Zogby International and partner Rock the Vote found Massachusetts Senator John Kerry leading President Bush 55% to 40% among 18-29 year-old likely voters in their first joint Rock the Vote Mobile political poll, conducted exclusively on mobile phones October 27 through 30, 2004. Independent Ralph Nader received 1.6%, while 4% remain undecided in the survey of 6,039 likely voters. The poll is centered on subscribers to the Rock the Vote Mobile (RTVMO) platform, a civic engagement initiative launched last March by Rock the Vote and Motorola, Inc., responded to this poll between October 27 and October 30.

The problem with this poll is that while it went after cell phone users, it apparently did not identify those people who have no land line -- so there's no way to know the magnitude of any sample bias in more traditional polls. [Isn't another problem with this poll that they used Rock the Vote's database, which might be nonpartisan in theory but is undoubtedly Democrat-heavy in practice?--ed. Zogby says "The results of the survey are weighted for region, gender, and political party," so I'm assuming he's compensated for that kind of sample bias -- but this is open for debate.]

Again, remember the electoral projection motto of "I don't know who's going to win -- and you don't know either."

UPDATE: The three things mentioned in this post trend towards Kerry, so here's a thought that trends towards Bush. If I remember correctly, last time around Zogby's polling trended strongly towards Bush in the last week or two of the election, leading to one poll suggesting that California was a dead heat between Bush and Gore. Obviously, those polls underestimated Gore's growing strength over the final few days.

Now a lot of people are assuming that the polls will kick the same way this time, and that therefore a tie really means Kerry is up by a few percentage points. Click here for an example. However, what if the trend that the polls missed wasn't the late surge towards a Democrat, but the last surge towards the incumbent party? I know this flies in the face of the incumbent rule, but it's still worth keeping in mind.

LAST UPDATE: Will Saletan et al at Slate get the final word:

Here is the math that matters: If all the states in which the data lean discernibly to either candidate vote as the polls suggest, the election will come down to Florida and Ohio. If Bush takes both, he wins. If Kerry takes either, he wins.

posted by Dan on 11.01.04 at 01:18 AM


The Gallup prediction of the election on CNN is 49% Bush, 49% Kerry, 1% Nader, 1% other. So they take into account other. But who knows--its not a poll, just a prediction.

posted by: Damascus on 11.01.04 at 01:18 AM [permalink]

The weird thing is that it looks like both of them are going to win. That spells trouble.

posted by: oldman on 11.01.04 at 01:18 AM [permalink]

They started with a sample of young adults who signed up with Rock The Vote and Kerry only barely won. That doesn't sound good - for Kerry.

posted by: Chad on 11.01.04 at 01:18 AM [permalink]

No, I don't "know" who's going to win, but how about having the moxie to throw confidence intervals to the side and call the damned thing anyway? Here's the memo I sent to my address book last night...

Here's how I score the upcoming presidential election. The folks at have conducted a rudimentary meta-analysis of the most recent state-level polls. This approach has the advantage of removing many states from the "battleground" column into relatively safe states for one candidate or another. For example, this meta-analysis puts Pennsylvania solidly in the Kerry column and Florida solidly in the Bush column. It has the disadvantage of possibly missing very late trends in the polling data. This approach puts all states but five solidly (85% chance or more of winning) in one camp or another. Of those five states, I don't believe that three (Arizona, Hawaii, and Michigan) are really in play. Arizona will go for Bush, Hawaii and Michigan for Kerry. That leaves only two states really in play: Iowa, with seven electoral votes, and Ohio, with twenty.

An appendix with how I call the states is listed at the bottom. But the real bottom line is that Bush has 259 safe electoral votes, and Kerry 252. That makes Iowa irrelevant, and Ohio the state that will make a big difference. (Unless, of course, Hawaii really is in play, in which case Bush could lose Ohio, but win, with 270 electoral votes, by carrying Iowa and Hawaii).

If you're looking for an early indicator, check New Hampshire. Kerry should take this one; if he doesn't it's going to be a long night for Democrats. There probably isn't a good early indicator for Bush. Florida would be, of course, but after what happened in 2000 the networks are going to wait forever to call that one.

So, the election really is too close to call, but saying that is the coward's way out. Who's really going to win? The Democrats could win if they dramatically increase voter turnout, but I don't believe they really have the troops to do that. Blame this on Tony Coelho and his inheritors. Failing sharply increased turnout, the election will turn on late trends. So how will the late deciders break? They're going to break for Bush. The one issue on which he has consistently led Kerry throughout the campaign is ability to fight the war on terror. (I'm not saying he is better, just that he's consistently perceived to be better, and by a large margin). With Friday's release of the Osama bin Laden tape, that issue is going to have more salience in voters' minds. Hence, Bush wins.

As for the Senate and House, don't sweat it. They're both staying Republican.

Happy election! Vote early and often!

BUSH STATES (electoral votes in parentheses)
Alaska (3)
Alabama (7)
Arkansas (6)
Arizona (10)
Colorado (9)
Florida (27)
Georgia (15)
Idaho (4)
Indiana (11)
Kansas (6)
Kentucky (8)
Louisiana (9)
Missouri (11)
Mississippi (6)
Montana (3)
North Carolina (15)
North Dakota (3)
Nebraska (5)
New Mexico (5)
Nevada (5)
Oklahoma (7)
South Carolina (8)
South Dakota (3)
Tennessee (11)
Texas (34)
Utah (5)
Virginia (13)
West Virginia (5)
Wyoming (3)

California (55)
Connecticut (7)
D.C. (3)
Delaware (3)
Hawaii (4)
Illinois (21)
Massachusetts (12)
Maryland (10)
Maine (4)
Michigan (17)
Minnesota (10)
New Hampshire (4)
New Jersey (15)
New York (31)
Oregon (7)
Pennsylvania (21)
Rhode Island (4)
Vermont (3)
Washington (11)
Wisconsin (10)

Iowa (7)
Ohio (20)

posted by: Michael on 11.01.04 at 01:18 AM [permalink]

My first-best outcome for this election is for Bush to win the popular vote, Kerry to win the electoral vote, and the Libertarians to play a 'spoiler' role as Nader was alleged to play in 2000. I have this idealistic belief that such an outcome would take the hot air out of a lot of sails.

posted by: dubious on 11.01.04 at 01:18 AM [permalink]

I have no clue. I have no prediction. I have one fear -- the lawyers will grab hold of this somehow, and attempt to delegitimize the result. My guess is that the state that will be victimized this time will be Ohio.

Personally, I'd like a gentleman's agreement by the candidates that there be a hand recount of all ballots in a disputed state (to the extent possible), and a threat by Congress to reject the electors of any state that permits either campaign to cherrypick votes from selected counties.

posted by: Appalled Moderate on 11.01.04 at 01:18 AM [permalink]

Can someone explain something to me?

I can understand that polls are being used by the campaigns to (re-)focus their resources. Identify where the race is tight and where more campaigning might have the most impact. Fine.

But I don't quite understand why polls are being spun - by both sides - to paint a particularly bright picture of one's candidates prospects.

If a candidate leads in a poll or looks like the winner, will this fact make more people vote for him? Couldn't it suggest to some of this candidate's supporters that there is no urgent need anymore to go out and vote (especially if it rains, as it will across quite a few battleground states tomorrow)? Whereas the opponent's supporters might be more energized trying to turn this around?

One should also consider voters who are unhappy about both choices and reluctantly willing to vote for one candidate because they don't want the other one to win. These voters might just be looking for an excuse not to vote.

As to my own prediction: well, we are seeing a similar last-minute change in polls as in 2000. Bush was ahead by up to 5 points in many polls just a few days ago, now the polls are (almost) even, with some even showing a Kerry lead (Fox News is showing a 2-point Kerry lead!). The NY Times reported that more people are making the switch from Bush to Kerry than from Gore to Bush. New voters are supposedly voting for Kerry by a wide margin. All this suggests a Kerry victory.

posted by: gw on 11.01.04 at 01:18 AM [permalink]

They started with a sample of young adults who signed up with Rock The Vote and Kerry only barely won. That doesn't sound good - for Kerry.

55% to 40% doesn't sound liked barely won to me.

Perhaps your mental model of "young adults who signed up with Rock The Vote" is different than mine? I image them to be the a reasonable cross section of middle American youth, more suburban and rural, than urban, more white than non-white, and perhaps excluding both the religious fundamentalists/evangelicals and the self-consiously intellectual.

I predict Kerry wins the popular vote by more than 5%.

posted by: Ray on 11.01.04 at 01:18 AM [permalink]

Here at UW, I haven't noticed much partisan stamp on Rock the Vote one way or the other. Some perceive them as Democratic because they played the draft card, but through paranoia if nothing else college-aged males do worry about it, and did long before it became a campaign issue. Some who don't follow news closely wondered about it during the Kosovo War. That's just the way it is.

posted by: Brian Ulrich on 11.01.04 at 01:18 AM [permalink]

The humor lies in the fact that the Election will hinge on factors, or political concerns, where both Candidates have stances which are wrong. List of Issue which will decide the Elections:
1)Who will get Us out of Iraq fastest?
2) How angry are Seniors about the Co-payment and Timeframe of Implementation schedules of the Proscription Law? Who is best likely to fix this boondoggle for Seniors.
3) Who is going to signal the return of good-paying Jobs and Pension benefits?
4) Who is most likely to get a Cap on health care expense increases?
5) Will Bush keep the military vote (including Dependents and families) with high Casualties, second-term deployments, and Wounded being mishandled?
6) Do the bulk of Republicans--Small Town and Suburb Storeowners and Workers--think they have been served by the Bush Tax Cuts? Remember they received only about one-third the benefit from the Tax Cuts, as did their Suppliers who raised Prices throughout.
7) How many Voters have a personal family member or friend out of Work, or employed at a lower paying Job than in 2000?

The above are the major worries and considerations for both Candidates. Both do not stand well on these Issues. lgl

posted by: lgl on 11.01.04 at 01:18 AM [permalink]

Bush wins all his leaners, steals Hawaii, wins the battlegrounds of Florida,Iowa,and Minnesota and earns a split in Maine. George W. Bush 281-Neville Chamberlain-257.

posted by: Rocketman on 11.01.04 at 01:18 AM [permalink]

There doesn't have to be a late surge for Kerry -- or Bush -- to win. There only needs to be a late surge to either candidate in a small number of states.

I can put together scenarios that have Kerry winning both Florida and Ohio and still losing the election. They're not impossible, but they are improbable -- they require Bush winning all the Upper Midwest battleground states and picking up a few extra votes in states like New Hampshire. So Saletan's probably right. I've predicted a Kerry win before, but I'm not claiming clairvoyance. At this point I think I'd settle for knowing the name of the winner 48 hours from now.

posted by: Zathras on 11.01.04 at 01:18 AM [permalink]

gw: But I don't quite understand why polls are being spun - by both sides - to paint a particularly bright picture of one's candidates prospects.

I think it has to do with perception of momentum and the bandwagon effect. I think the effect is a lot more obvious in the primaries (and this year, after the debates), but it's still there. People would probably rather vote for the clear winner than for a lost cause.

posted by: fling93 on 11.01.04 at 01:18 AM [permalink]

Seems like the Christian & Evangelical voter blocks are of 2 minds - those deeply committed to Bush as a man of God (regardless of all other issues), and those who question the current healthiness of the tight relationship between the GOP and the Church.

From what I read, the Confederacy of the Catechism has promoted a very inconsistent, and very political, pro-life stance. It's become single issue: abortion. Yet the death penalty, euthanasia, fed $$$ for chronically ill poor, suicide prevention, etc.. have been nearly absent in their national message. This phenomenon happens to align far too closely to the GOP message this campaign, and over the last 4 years.

posted by: wishIwuz2 on 11.01.04 at 01:18 AM [permalink]

Are there really other indicators for a dramatically increased number of voters this election, other than the increased number of registrations? I seem to recall back in the spring when the pundits were atwitter that there would be a flood of these cell-phone-using, internet dwelling young enthusiasts who would carry Dean on to the nomination. I think we all know how that turned out...

The thing I think one should examine is how many of the new registrations (of those that are actual people, and only counting each person one time...) actually went out and said, "I need to register to vote!" rather than those who were harangued and a shopping mall or by someone who came out to their house. Given the number of new registrations both sides are touting, I'm guessing these were not the most motivated of folks (or they would have voted last time around, for starters).

If there is not a huge increase in participation, it affects the polls -- at least those polls that adjusted their metrics to account for this increased turnout. Should that turnout prove a chimera, it would seem that, given everyone assumes the increased turnout would favor Kerry, Bush would be looking like a very good bet.

BTW, did that Rock the Vote poll only call people it signed up? That would seem to me to be a huge selection bias with the sample -- much like an internet poll, but without the possibility of multiple votes.

posted by: DC Corporate Lawyer on 11.01.04 at 01:18 AM [permalink]

Zathras is right: Saletan's analysis is almost perfect. But they could mirror 2000, with Bush taking FL and OH, while Kerry picks up NH and CO and Kerry could win. But that's the only exception I could project with the current numbers to Saletan's take.

Most polling indicates 10%-11% are first time voters this time, but that's only meaningful by comparing to previous presidential years. However, state registrars are saying the numbers of new registrants, absentee ballots returned, etc, indicates turnout will parallel 1992, and the only higher one in several decades was 1984.

posted by: Kevin Hayden on 11.01.04 at 01:18 AM [permalink]

Post a Comment:


Email Address:



Remember your info?