TJ | 5th Nov, 2012

Predict An Election Yourself

Print Print

Eleanor Roosevelt Voting

This piece from 2012 had so many hits I thought it should be revived. There is no mystery behind predicting election winners if you know how. You need to research past patterns for the attached demographic groups then closely follow the turnout data. Which groups are running ahead or behind their recent patterns. Lest you scoff at this, using this basic methodology, this blog was the only source to correctly predict the 2012 Iowa results within two percentage points and was among the earliest to call the Presidential race that year.


Predicting the Presidential election winner will be the name of the game tomorrow night and every network, blog, and tweat will be competing to be the first to name the winner.  The networks and the pundits make it sound like it takes a supercomputer with some magic formula to do the job.

While there is no denying the ability of such tools to predict down to the precinct level and the strength of their prediction models, it is possible for viewers to make their own predictions while watching the election returns and come pretty close to matching those multimillion dollar machines. If you are having a party that night you can even match predictions with friends.

If you have ever sat in on a campaign headquarters on election night, this is exactly what they do. Most candidates know if they have won or lost before the networks “call” the election because they know which precincts they have to win and by what margin and what the margin needs to be in the precincts they will probably lose. In more sophisticated campaigns they do their own polling right along with the networks.

The Real Focus: Turnout

This site contains no big bucks from advertisers and no black box formulas, just information so you can make your own prediction.  If nothing else the analysis sheets should help liven up the evening for they provide a perspective you will not hear from the networks. Hopefully they also will spark some good conversation. In fact the sheets should help you make a better pundit than a fair number who will be taking up air time Tuesday night.

For all the smoke and mirrors consultants and pundits are throwing up around election results, the key to predicting elections is quite simple and has been since someone put them down on papyrus: the candidate who turns out more of his or her people wins. Based on results from the primaries, polling and other data we know which groups tend to lean towards Obama and which lean towards Romney. So predicting each state is a matter of seeing which voters from each of the candidate’s main support groups turn out in higher numbers. For six years this methodology has scooped the networks on every election except one. It produced THE most accurate prediction of the Iowa primary.

The Format

The charts for you to use are all in an attached Word or PDF file you can download and print them out. That way if you don’t want to read all of this essay, you can just print the sheets. You can jot your own comments on the back. Hopefully they will liven up your evening.

The spreadsheets are in one file with separate pages for each swing state.  That way you do not have to download and print separate files for each state.

The States

By now all of you know the election has come down to several swing states. There is some disagreement about some of these, so my choices are based on today’s polling data. These polling data come from RealClearPolitics which maintains a composite of all the major polls. Right now RCP has the following states as swing states: Ohio, Virginia, Colorado, North Carolina, Nevada, Florida, New Hampshire, Michigan, Wisconsin, Pennsylvania and Iowa.

According to RCP, Obama currently has 201 electoral votes from states that are solidly in his column while Romney has 191. The final 146 are either leaning towards one of the candidates or are one of the above swing states.  Add in the leaning votes (states where either candidate has at least a 5% advantage) and Obama has 219 electoral votes and Romney 202.

To make things easier, my definition of a swing state is any state in which the polling margin is basically zero–that is either candidate has a three percent lead or less since three percent is the usual statistical margin of error in political polls. That eliminates Wisconsin (4.2% for Obama), Pennsylvania (4.1% for Obama), North Carolina (3.8% for Romney) and Michigan (3.8% for Obama).  Wisconsin, Pennsylvania and Michigan have gone Democratic in every election since 2000.  The one wild card is VP Candidate Ryan who is from Wisconsin. Obama squeaked out a slim victory in North Carolina, which previously had gone for Bush, largely on the strength of the African American vote.

Admittedly this is cutting things pretty fine, but it leaves us with Ohio, Virginia, Colorado, Nevada, Florida, New Hampshire, and Iowa. We might call these the Big Four and the Little Three in that the combined electoral votes of  Nevada, New Hampshire, and Iowa are 16, less than Ohio and Florida and three more than Virginia.

Given the closeness of the race the winner cannot lose both Florida and Ohio with their 47 electoral votes.  If Obama wins them both he is within four electoral votes of the needed 270.  Romney would be within 20.  A combination of three out of the Big Four would put either candidate in the White House.

With that in mind, this Predict the Winner analysis will focus on those four states.

The Sheets

On each sheet you will find a series of columns across the top for various demographic groups. The first set of numbers below these groups are the most current data. For example, swing state “A” contains 12% African Americans. A few explanations are needed for some of the symbols and abbreviations used in each column to make the charts more readable and able to fit on one page. AA is for African American. Young is voters under 34. Old is voters over 62. ED is voters without a college degree. POV is the percentage with family incomes below the poverty line. UNION is the percentage of union members.

Census data are from the Census Bureau. Union membership comes from the AFL-CIO.

The open spaces below are for you to fill in the actual turnout data as you find it either from the networks or more likely on the Internet. CNN still maintains the most complete database, but to their credit they tend to release it late. So you will need to do a bit of detective work to find the early data.

A note here about the network data and that of the Associated Press: their exit polling is conducted by one organization, Edison Research. Edison’s methods leave me a bit uneasy since their pollsters often are hired just for this event and receive questionable training (some is via videotape). Several elections ago the networks and AP all pooled their resources and hired Edison, so when you hear Fox or CBS talk about their polling; they really are talking about the same poll and the same data–Edison’s.

This practice of using the same data to call election winners bothers me. Essentially the election is in Edison’s hands since their exit polling data declares the winner long before all the ballots are counted.  Most states have their own independent pollsters, who often are more accurate than Edison. So a good place to start is the major paper in your state.

The Colors

The red-colored numbers indicate increases from the 2008 Presidential election.  The most notable–and most commented on–is the increase in Latino voters.  In the four states it totals about 7%.  Given polls show those voters are leaning towards Obama, that gives him that many more votes over what he had in 2008.

The other increase is even more interesting.  Reports about the death of the union movement are exaggerated for in three of the four states union membership INCREASED since 2008. With the GOP pushing right-to-work measures this should bring out union voters who have increased 3.7% in the four states. This alone would be enough to swing those states to Obama.

The blue colored numbers represent the 2008 percentage turnout for that particular group. It is important to remember that this DOES NOT represent the percentages in that group voting for Obama. For example Obama won 95% of all African Americans nationally and 56% of women.  If his support among African Americans falls to the low nineties or below ninety he is going to have a tough time winning.  These turnout percentages are useful to compare to this year’s results. Should Obama’s core voters turn out in smaller percentages than 2008 he will be in trouble.

The States

If you would rather not read the BS below, click below for the Word or PDF file.

2012 prediction sheets (Word)


The left hand column headed “Time” identifies the time of the sample. “Pre” means pre-election. “2008” are data taken from the CNN exit polls for the last Presidential election. “2nd data” and “3rd data” are for you to fill in data as you find it or to just make a prediction.  For example, you could use 2nd Data to predict and 3rd data for the first real data. “Final” is for the actual results. It can be compared with 2008 and the Census data to create your own analysis of what happened and maybe spark a few early morning conversations or arguments at work the next day.

Have fun. I will be adding my own, only I will not start making predictions after the polls close.


Obama won Colorado handily in 2008, in part on the gender gap with 56% of women voting for him. He also won every age group except those over 65 along with 62% of the Latino vote.

The fact that Colorado is a tighter race this year may have to do with a falling off among these groups.  If the turnout by women is low along with that of Latinos, he will have a tough time winning this state.


In 2008 Obama had a lead of 4.2% in Florida on the eve of the election. This year Romney has a lead of 1.4% and is expected to win the state. At one point there was talk of Obama not even contesting Florida.

Florida is a strange state in that the combined percentage of Latino and African American voters is 39.4%.  That should be more than enough to swing the state Democratic in every election, but that has not happened. One explanation is that the Latino vote is a wild card since Florida’s large Cuban-American population has tended to vote Republican. But that has changed in recent years so it does not sufficiently explain Florida’s results.

If you remember back to the Bush-Gore Supreme Court case some explanation can be found in evidence and testimony suggesting African Americans were discouraged from voting or had problems voting.  Pay careful attention to the vote from Dade County, which was at the center of Bush v. Gore.  If the turnout is high there, Romney may be in trouble. Results from Dade County tend to come in late, so if Florida is close and we don’t yet have the results from Dade we could be in for a long night and perhaps even another court case.


In 2008 the polls showed Obama with a 4.2% lead.  Now it is 2.8%.  The economy is a huge issue here with a high foreclosure rate, fear over potential plant shutdowns and the financial crisis.  There also is union anger at the right-to-work initiative. Blue collar voters are a key swing vote in this state and if they go heavily for Obama, as they seem to be trending towards, then that lead could widen.

If African American support falls below 2008 that means that the enthusiastic support for the President that characterized the African American vote in 2008 has waned. If you wonder why you are seeing the Colin Powell ad so many times this is why.

Like Florida, Ohio has a history of election shenanigans almost as bad as Florida. It is an inside joke that both campaigns already have more lawyers than campaign staff. If this one goes early for Obama it could signal a big night for him.


So we come to the last toss-up state. You need only look at the demographics to see why. First, it is one of the few states where younger voters outnumber older. Second, among the swing states it has one of the highest percentages of people with a college degree. Third African American and Hispanic voters make up almost a third of the electorate.

The state has been trending Democratic and a solid vote for Obama could solidify that. This will be a state where the African American and Latino turnouts will decide the winner.


A key group will be low income voters. No Democrat has won without them.  The GOP has been running ads blaming Obama for an increase in the poverty rate.  It will be interesting to see which party these voters blame for their situation.

While it is doubtful these voters will swing Republican they have a recent history of staying away from the polls, with one of the lowest voting rates of any group. If they vote significantly below their demographic percentage it means Obama has lost the confidence of this group that he can handle the economy.  If that occurs expect for Obama to have a long night.

Although much of this analysis is predicated on polls showing older voters trending for Romney, I think they are the demographic to watch in this election.  Older voters used to be a Democratic-leaning group.  The question will be whether recent moves by Republican radicals to privatize Social Security and Medicare have scared these voters back into the Democratic column.  If these voters record high numbers for Obama, especially in the two Southern states, Florida and Virginia, it is all over for Romney.

Finally, the combined Latino and African American vote, which in my book has been the swing vote in every election for at least half a century, reached its maturity in 2008. The victory of an African  American whose margin of victory was due in large part to people of color signaled the beginning of a major transition period in American politics.  The big question is whether the economy has thrown a wrench into this.

If he loses, Barack Obama will be the first Democrat since Grover Cleveland to campaign for a second term with questions being raised about a major economic crisis. Jimmy Carter’s was minor compared to the Great Recession. Most people believed FDR’s New Deal was working. Cleveland faced the Panic of 1893 and lost the support of his own party for failing to take on Wall Street, intervening in the Pullman Strike against the unions, having to borrow money from J.P. Morgan to prevent the collapse of the Treasury and failing to aid starving farmers.

While not as ineffective as Cleveland, Obama’s policies echo Cleveland’s refusal to take on Wall Street. For example,  Obama has failed to prosecute three banks for violation of the Riegle-Neal Act. Most people have never heard of Riegle-Neal, but there is the feeling Obama has gone soft on Wall Street, in part because that is the stance of his chief campaigner Bill Clinton and Treasury Secretary Timothy Geithner. [More on that after the election.]

The Wild Cards

This will be the first Presidential election to take place in the era of social media. If you followed coverage of the debates, you heard a lot of pseudo-analysis coming from network pundits generalizing about Twitter and Facebook postings. Expect that to continue.

Both sides will make huge efforts to get control of the social media early in the day. This will turn into a deluge after the first poll closings. There will be some crude attempts to sway voters in states where the polls have not yet closed by claiming that their candidate is ahead and posting rumors of election irregularities.  Twitter and Facebook are under no rules to not broadcast exit polling.  It is going to be tough to keep the lid on this and to separate the noise and BS from what is real.

The other wild card is Sandy. It devastated New Jersey where Obama currently holds an eleven percent lead.  Reports stress low income people suffered disproportionally from the hurricane with many of them still without power and shelter and some scrambling just to find something to eat. There is no question turnout will be lower because of Sandy and that it will hit Obama voters more heavily.  If New Jersey becomes close Obama could have a tough hill to climb.

Expect some law suits to come from attempts to allow people to vote away from the polls. The GOP has no choice but to file these suits since its strategy has been to discourage alternative voting. If alternative voting works in the wake of Sandy it will give a big boost to those who are pushing to do away with voting practices that have not changed much since the nineteenth century.

See you on election eve.

Print Print