#FF00FF 2012

An FAQ for the 2012 US Presidential Election

This is an FAQ (Frequently Asked Questions list) for the 2012 United States Presidential Election. I need to disclose up front that I support President Obama. However, with the exception of the very last question, this FAQ is designed as a collection of factual information (such as the latest poll results) and of analysis that is as objective as possible. Why did I do this? To educate interested readers. My ambitions are not as grandiose as Sam Wang of the Princeton Election Consortium, who wrote " When I started doing the Meta-Analysis of State Polls in 2004, I thought it would be a useful tool to get rid of media noise about individual polls. ... Space would be opened up for discussion of what really mattered in the campaign - or even discussion of policies. To my disappointment, this has not happened. Maybe it just takes time. Or perhaps polling nerds need to get a few more races right. Let's see if we move the ball forward for Team Geek on Tuesday."

Table of Contents (links stay on this page)

What does "#FF00FF" mean? 
Who's winning?
What's the score?
Wanna bet?
Can Romney win? How?
How accurate are polls?
Where do I vote?
What about voter fraud?

What do the candidates say?
What about the Economy?
What about Taxes?
What about Spending?
What about State vs. Federal?
Who endorses each candidate?
Are the candidates always honest?
How honest or dishonest are they? 

What happens in case of an electoral vote tie?
Who else is running?
Is it rational to vote?
Who do you think will win?
Why do you support Obama?

What does "#FF00FF" mean? Why the garish color?

In the naming convention for web colors, maximum red is #FF0000, maximum blue is #0000FF, and when you mash them together you get magenta: #FF00FF. It is garish, but that's what you get when you combine two extremes.

Who's winning?

Check the following sites for up-to-the-day aggregation of polls. We list their predictions for the probability of Obama winning (in percent) for those sites that give a probability, and for electoral votes (EV) for each candidate (as of Nov. 3):

Obama PctObama EVRomney EVTossup EVSite
98%233191114 Princeton Election Consortium (Sam Wang, Princeton)
?29422024 electoral-vote.com (Andrew Tanenbaum, Vrije Univ)
?201191146 Real Clear Politics (John McIntyre)
?3322060 Votamatic (Drew Linzer, Emory Univ.)
86%3072310 538 (Nate Silver)
?27719170 Pollster (Mark Blumenthal)
?29019157 Polltracker.com (Charles Franklin)
83%201181156 NerdWallet (Joanna Pratt)

Different analysts combine these polls in different ways, and end up with different predictions. For Romney to win, we would need a swing of about 3% of voters in his direction. This has happened before, so it is possible, but time is running out for Romney. Alternately, the polls would have to be consistently wrong by 3%, which is very unlikely. Romney's best hope is that the polls underestimate his support by about 2% (the largest amount that is remotely likely) and that there is some event in the last week that causes another 1% to change their mind towards him.

Interestingly, these poll aggregators are all "hobbyist/academic" statisticians, not big news organizations (although 538 and Pollster have acquired sponsors since the last election).

To track who is winning in real time on election day, watch first for the betting markets (such as Intrade, and later in the day look for election result news. Also check out Sam Wang's Geek's Guide to Election 2012.

What's the score?

No votes will be counted until Nov. 6, but millions of votes have been cast. Some states release the party affiliation of early voters, which does not necessarily indicate who the votes were cast for, but it is a hint. Here are some early voting registration totals as of Nov. 2:

N. Carolina2.1M48%32%

A Reuters/Ipsos poll on Oct. 14 of early voters in all states had Obama ahead by 59% to 31%, and a Reuters poll on Oct. 27 had Obama ahead 54% to 39%, but a Gallup poll Oct. 22-28 had Romney ahead by 52% to 46%.

If you want to dig into the complete stats, the George Mason University elections project has comprehensive early voting statistics.

Wanna bet?

Recently there has been much interest in prediction markets. The idea is that if people bet real money on an outcome (such as the presidential race), they may have more honest and more considered opinions than typical poll results. There are several markets covering the election; here are the current numbers from the Democratic Party point of view, expressed as a probability of winning:

For interesting commentary on political futures markets, see Worlds of Spike. Justin Wolfers has a WSJ article that covers the "favorite-longshot bias" -- the idea that people bet too much on longshots. You can see this at horse races, and it is clear why it happens: if a horse is a 20-1 longshot, bettors say to themselves "if this one pays off, I'll be rich!," potentially changing their lifestyle, whereas if they go with the favorite, their payoff will be much more modest.

Can Romney win? How?

Yes. Although the odds are against him. This chart from Nate Silver summarizes it best. These are the 20 most competitive states, with their number of electoral votes, the current forecast based on polls, and the odds of the leader holding on to win.

Here's how: to get to a Romney victory, start at the bottom and move up. You see the cumulative electoral votes for Romney in the "R" column. If Romney holds all the states he currently leads in, he would have 235 electoral votes (that's the "R" column in the "VA" row -- this indicates that Romney has won all the states below VA, but not VA itself). So Romney needs 34 more electoral votes (we'll assume that Romney wins in case of a tie). Of the 5 next states, Obama leads in all 5, and even if Romney took 4 of the 5 without Ohio, that would only give him 32. So this suggests that either Romney needs Ohio, or he needs a major upset in a state where Obama currently has an 85% or greater chance of winning. Romney's easiest path to victory is VA + CO + OH. If the polls are accurate and each state vote is independent, then the odds for Romney would be (100-80)×(100-63)×(100-61)%, which is about 3%. But we don't know how accurate the polls are, and voting in each state is certainly not independent, so different models of the election arrive at different percentages for a Romney victory -- all higher than 3%, but ranging from 4% to 35%.

(Note: If Romney loses OH (but wins CO and VA), the easiest way for him to reach victory would be to win NV and IA (where he currently has 15% and 22% chances), thereby giving him 269 electoral votes, a tie, which the House would presumably resolve in his favor.)

The New York Times displays a tree of all 29 = 512 possible outcomes of the nine swing states, with winners for each. Summary: If Obama wins FL, Romney has to win all the other states. Otherwise, if Obama wins OH, he only has to win one more state (other than NH, which is only 4 EVs). If Romney wins both FL and OH, then each candidate has multiple paths to victory.

How accurate are polls?

Historically, individual presidential polls, such as Gallup, have missed by an average of about 3% off the final vote tally. But if you average together multiple polls, the historical accuracy is about 1% off the final total. Scott Keeter of the Pew Research Center sums it up: "In 2004, nearly every national pollster correctly forecast that Bush would win in a close election, and the average of the polls predicted a Bush total within a few tenths of a percent of what he achieved. Among statewide polls in races for governor and U.S. Senate, 90% correctly forecast the winner, and many that did not were still within the margin of sampling error. The record in 2000 was similar, though that was an even closer [presidential] election." In 2008, the election was not as close: Obama won by 7.2 percentage points, which is 0.3 points off of the average of the final pre-election polls.

For a Presidential election, we are more interested in who wins, which is determined by the electoral college, than we are in the total vote percentages. Poll aggregators do even better at predicting the electoral vote; for example, the Princeton Election Consortium missed by 1 electoral vote in 2008, and by zero in 2004.

Polls can be wrong for a variety of reasons. A poll samples a small collection of possible voters, and from their responses makes inferences about all voters. Any sampling process can be off for two reasons: random variation and systematic error.

Random variation: The random variation is easiest to explain, and for some reason everyone explains it by referring to a jar full of, say, blue and red jelly-beans. You shake the jar to make sure the jelly-beans are mixed up, and then reach in and grab a sample. Let's say you sample 6 red and 4 blue. Does that mean that exactly 60% of all the jelly-beans are red? Of course not; if you repeated the experiment, you might get 5 or 7 or 4 reds out of 10 the next time. By sampling only 10 individuals, you can't get a very reliable picture. But, if you sampled 1,000 jelly-beans and got, say, 54% blue, then you could be more certain that this was a good estimate. In fact, some simple math will tell you that there is a 95% chance that the actual portion of blue jelly-beans is within 3% of the sampled 54%, assuming the sample was done fairly. This 3% is the "margin of error" that is often published along with polls. The margin of error is proportional to the square root of the number of samples, so in order to cut the margin in half, we would need 4 times more samples. To cut the error by a factor of 10, we would need 100 times more samples.

Note that the term "margin of error" is misleading because some people think of it as an absolute margin, when it is actually the 95% confidence interval. When there is a 95% chance that the actual result is within 3% of the sampled 54% result, there is also a 68% chance of being within 1.5%, and a 99.993% chance of being within 6%. Sometimes you will hear innumerate reporters say "candidate X leads candidate Y by 3%, and because the margin of error is 3%, that means the race is a statistical tie." Not true. That result would mean a 95% chance that candidate X is winning. And if X was ahead by 1.5% in the poll, there would be a 68% chance that s/he was actually leading.

Why don't pollsters cut the margin of error by including more people in their sample? Mostly because it is expensive. They feel embarrassed to publish results with much fewer than 1000 samples (because the margin of error would be higher than their peers) but there is no real incentive to add more sample points. The pollster gets more publicity (and more money) from releasing 4 polls with 1000 samples each (and thus a 3% margin of error) than by releasing one poll with 4000 samples (and thus a 1.5% margin of error). But a bigger reason for not including more samples is that it would be pointless, because the systematic error would dwarf the random variation. The dirty not-so-little not-so-secret of polling is that systematic error is much harder to control, and because there is no nice clean Bernoulli formula for systematic error, the margin of error figure covers only the variation.

Systematic error: Getting back to the jelly-beans, systematic error could occur if the red jelly-beans were a slightly different size or shape than the blues, making the reds settle to the bottom of the jar. Then the sample we draw would reflect the statistics of the top of the jar, but not of the whole jar, and we would get a bad result. The "margin of error" might still be 3% (meaning that if we repeated the same flawed experiment we would get the same flawed answer (within 3%) most of the time), but the actual result could be off by 5%, 10%, or more!

Some statistical analysis can quantify the chance of systematic error; Drew Linzer of votamatic.org has a nice analysis. For a thoughtful analysis of how this year's polls might have systematic error, see Ted Frank's election prediction, which reduces Obama's 80% chance of winning at 538 down to 60%, because of several possible sources of systematic error.

Most election polls work by random digit dialing: choosing phone numbers at random (with area codes and exchanges that are known to be in the right locale). Pollsters know that this method does not give an exactly equal chance of reaching each voter: it is like a jar that is not evenly shaken. There are six main sources of systematic error that pollsters try to control for:

  1. Sample design error: Pollsters know they don't really have an equal chance of reaching every voter. Most pollsters dial randomly from a list of land-line exchanges. But that leaves out those who only have a cell phone; pollsters who also call cell-phone-only voters report about a 4% larger lead for Obama (see chart at right from Nate Silver where cell-phone pollsters are in yellow). Cell phones are difficult for pollsters to deal with. It is illegal to call a cell phone with an automate dialing system (because it would involve an expense for the callee), and it is expensive to poll with manual dialing, so pollsters either ignore cell phones, or sample a few and then try to extrapolate. Who has the right model for how to count these cell-phone-only voters? It is not clear.
  2. Weighting: Pollsters try to correct sample errors by weighting the results they get. For example, about 52% of voters in 2004 were women. A pollster whose sample gets 60% men and 40% women might want to weight the samples to get back to the 2004 percentages. Or they might want to weight to a different percentage that they predict will occur in 2008; maybe the presence of a female VP candidate will bring out more women voters. A pollster has to guess (er, I mean estimate) the percentage, and if you they wrong, the poll result will be wrong. Here are some factors that pollsters weight for:
  3. Availability: What happens if there is no answer? Pollsters move on, but feel uneasy: they'd like to have a model that weights unavailable voters. Some pollsters to persistently call back, get a feel for the way unavailable voters are going, and weight them accordingly.
  4. Order and wording of questions: You'd think this wouldn't matter much in a presidential poll, but remember that pollsters ask multiple questions. For example, an IBD-TIPP poll that reported a 1% Obama lead at a time when the average was 7% asked the question "Are we ready for socialism?" before asking the question on presidential preference. By asking questions that prime positive or negative associations in the mind of respondents, pollsters can move results several points in the direction they choose. Psychologist Elizabeth Loftus demonstrated this in an experiment where she showed subjects a video clip of a traffic accident, then asked half of the subjects "how fast were the cars going when they smashed?" and the other half "how fast were the cars going when they bumped?" The first group reported significantly faster speeds. Words matter.
  5. Refusal rates and truthfulness of responses: Some respondents refuse to answer pollster questions. If refusing to answer is correlated with voting for one candidate, then that candidate will be underrepresented in the polls. Some pollsters try to get at this problem by calling back repeatedly, but this in at best a partial solution. In some cases, voters may lie, although there does not seem to be a motivation to do so. The Bradley Effect is an instance of voters lying, but as reported above, seems to be a minor effect (if any) today.
  6. Passage of time: Obviously, the polls are taken before the election, and voters may change their minds between the poll and election day.

Now, after each individual poller does their best to collect and make sense of the data they collect, then the poll aggregators step in. Each has their own model of how to interpret the individual polls. For example, if one poller is consistently different than the others, should they be ignored or counted less? How much should polls in the past be discounted compared to more recent ones? Each aggregator makes their choice in defining a model of the election, then feed the individual poll data into the model.

One important part of the model is: how much change do you expect from the current polls? If we had thousands of past presidential elections, this would be easy: we would have a good idea of the range of differences from past experience. But the sad fact is that presidential elections occur only every 4 years, and this level of intensive polling has only been going on for about 30 years, so it is hard to estimate how much possiblke variation there might be. Different aggregators make different choices. At one end, Sam Wang at Princeton uses a minimal intervention approach, letting the data speak for itself as much as possible. So he ends up with (as of Oct. 31), a 95% probability for an Obama win. But Nate Silver builds in more of a possibility that things will change (in unpredictable ways in either direction), and arrives at a 77% chance for Obama. Most of the aggregators agree on who is winning in each state; they just disagree on how much chance there is that that will change in the remaining week. Simon Jackman of Stanford University explains how in his model, he starts with a 99% probability of an Obama electoral vote lead today (based on the amnount of random error in the polls), and arrives at a 71% chance of an Obama victory (because opinions might change, or the polls might have some systematic error). To make things even more explicit, he shows the latest 110 polls in battleground states (CO, FL, IA, NC, NH, NV, OH, VA, WI), plots them by percent of the two-party vote (that is, not counting undecided or third parties) and gets this graph, which shows Obama with a lead of 50.6%, which is small, but extremely probable given the polls (over 99%):

To reiterate, the problem for Romney is that, even if you give him FL and NC, he still needs 34 more electoral votes. The easiest way for him to get there is with VA + CO + OH, three states in which he trails by 2 or 3 points.

Nate Silver has a nice summary of where the aggregators stand, based on these results:

You can see that there is only one state, Florida (FL) where there is disagreement over who is winning, and then only one of the aggregators has Obama in the lead. (There are also four states (NH, IA, CO, VA) where one or more aggregator reports a tie, and the others report Obama in the lead.)

Now, I have to admit to a bias here. Not a Democratic/Republican bias, but rather a Data/Intuition bias. I'm on the side of data. But not everyone is: talk show host Joe Scarborough complained

Nate Silver says this is a 73.6 percent chance that the president's going to win. Nobody in that campaign thinks they have a 73.6 percent -- they think they have a 50.1 percent chance of winning. .... Anybody that thinks that this race is anything but a tossup right now is such an ideologue [that] they should be kept away from typewriters, computers, laptops, and microphones for the next ten days, because they're jokes.
Nate counters with
We can debate how much of a favorite Obama is; Romney, clearly, could still win. But this is not wizardry or rocket science," Silver told POLITICO. "All you have to do is take an average, and count to 270. It's a pretty simple set of facts. I'm sorry that Joe is math-challenged.
Joe is saying that he has an impression that this is a close race, and his impression should count more than data that contradicts his impression. Peggy Noonan says "all the vibrations are right" for Romney, and that vibrations should count more than polling data. You can see similar arguments, ignoring the data and going with the gut feeling, from Alex Castellanos, Jay Cost, Michael Barrone, and others. Perhaps he is also motivated to criticize the data-driven predictions because (1) they currently predict a loss for Joe's preferred candidate, and (2) if those predictions actually work, then there is less need for people like Joe, who make their living by talking or writing about their intuitions. Nate is saying that yes, things could change, but we have a pretty good idea of where they stand today (thanks to the plethora of polls), and we have some historical evidence of how volatile voters opinions are; how much they can possibly change per week. The Atlantic covers this battle between the nerds and the pundits.

Where do I vote?

Google has helpfully provided a Voter Information Tool. You can also find your polling place info from the National Association of Secretaries of State here.

What about voter fraud?

How big a problem is voter fraud? In 2002, the Bush administration thought it was serious indeed, and launched a crackdown. From 2002 through April 2007, this program yielded 86 convictions: 40 for vote buying/intimidation or ballot forgery, 18 for ineligible voting (e.g. by an non-citizen immigrant or a criminal on probation), 5 for voting twice, 5 for registration fraud (such as a non-citizen registering but not voting), and 2 for civil rights violations. The vote-buying schemes were all for local races; there were zero convictions related to state or national races. Ronald Michaelson of the McCain-Palin Honest and Open Election Committee says he does not know of a documented case of voter fraud that resulted from a phony registration form. Another study of 197 million votes in the 2002 and 2004 elections, there were 26 convictions for illegal registration or casting illegal ballots, but no convictions (or evidence) of someone impersonating another voter.

Of course, where there are 13 or 14 convictions per year, there may be more uncaught crimes. For the sake of argument, let's say that for each conviction there are 100 cases of fraud that go undetected. If that were the case, fraud would amount to about 0.003% of the vote, which would not be enough to affect even the close 2000 presidential election. Fox News looked into the problem and concluded "One doesn't have to look far to find instances of fraudulent ballots cast in actual elections." But readers of that article have commented that Fox did have to look pretty far: all the way back to 1984, when a study found "thousands" of fraudulent votes between 1968 and 1982 in state races. Again, this comes out to somewhere in the range of 0.001% of votes. While every fraudulent vote is a felony that should be taken seriously, there is no way this level of activity could affect a national race. As Steve Schmidt (strategist for the 2008 McCain campaign) said, "It's part of the mythology now in the Republican Party that there's widespread voter fraud across the country. In fact there's not."

On the other side of the aisle, Democrats complain about voter suppression. The Help America Vote Act of 2002 has a provision for removing erroneous registration records from the system. That seems like a good idea, but the problem is that most errors mean that the problem should be corrected, not that the voter should be purged. For example, "Joe the Plumber", or Samuel Joseph Wurzelbacher, was actually registered under "Worzelbacher." It seems clear he should be allowed to vote. The Republican strategy is to challenge registrations that have a typo like Joe's, forcing them to cast a provisional ballot, a process that is more time consuming and might cause frustrated voters to give up and go home rather than wait through a lengthy process. Why would Republicans use this strategy? Historically, because those who change addresses more often (and thus are more likely to have registration problems) are more likely to be low-income, and that means more likely to be Democrats.

Aware of this issue, the Obama campaign made a concerted effort to encourage early voting. On the Jay Leno show, Obama commented on this tactic. In response, Republicans tried to restrict early voting, but the courts have mostly restored or expanded voting opportunities. Overall, this issue seems unlikely to be decisive.

The third problem to worry about is voting machines (of all types). In 2000, over 2 million punch-card ballots could not be read, because of either multiple or inconclusive punches. Security of voting machines has ranged from poor to incompetent. I believe that many voting machines are insecure, and should be redesigned. However, I think there are sufficient human, non-software safeguards that software fraud is not a major threat: the voting system is mostly secure, even if some individual machines are not.

Despite all the claims of fraud in recent elections, there is no evidence of any fraud that makes a difference. In the 2004 election, the final total of 51% Bush to 48% Kerry closely matched the polls leading up to the election (various, but average within 1% of the final total). Sam Wong's state-by-state meta-analysis of polls predicted the final electoral vote result exactly. There was a larger discrepancy in the exit polls--everyone agrees that the exit poll totals varied significantly from the final results--but careful analysis shows that the exit polls do not conclusively show a problem (nor do they rule out the possibility of a problem). There just doesn't seem to be any clear evidence that the final vote was anything different than what the polls are telling us. Similar results hold for 2000, 2002 and 2006.

My conclusion is that voter fraud contributes much less than 0.001% of error, voter suppression contributes about 0.1% (about 100,000 votes) and voting machine errors (including honest mistakes by voters) are harder to estimate but probably less than 0.1%. Both parties have armies of lawyers standing by, so if there are any significant irregularities, you will certainly hear about it.

(On the other hand, you could say that the 2000 election was decided by voting error (due to the butterfly ballots), but you could equally say it was decided by Ralph Nader's ego, by Al Gore's sighs, by Jeb Bush and Katherine Harris's failure to recuse themselves, by voter suppression (by some counts nearly 3% of African Americans in Florida were removed from rolls), by the suppression of vote-pairing web sites, or by the whim of the SCOTUS.)

What do the candidates say?

You can catch soundbites from the candidates on daily newspapers, TV, or radio (I like XM radio POTUS, channel 124). For written summaries of the candidates stands on the issues, see CNN or Pro/Con.

You can always go direct to the position papers from Obama or Romney, or to wikipedia for summaries of the positions for Obama and Romney.

What about the economy?

President Clinton said it first: It's the economy, stupid.

Romney's main argument seems to be: Obama had a chance, but the economy is still in a mess; look, it didn't work; time to try a different way.

Obama's argument is that he inherited the worst recession since 1930, and we're recovering about as well as could be expected; a Romney presidency would take us back to the failed policies of the Bush years.

So we should ask: how has the economy done? One answer is: better than other countries. Below left we see that the US GDP is recovering (from the crisis in 2008-2009) faster than the UK, and faster than the combined European countries. Below right we see that US unemployment is recovering (going back down) while EU unemployment is getting worse:

My conclusion is that Obama is not getting an A+, and you can argue whether it is an A-, a B-, or somewhere in between. But it is clear that he is doing better than all his classmates. (Some of the performance is due to Obama's choices; other factors are out of his control.) On the other hand, when The Economist asks experts if the Romney-Ryan plan could sustain America's recovery, the answer is No, by a margin of 4-1. Their poll of experts mostly is negative towards both candidates but it is worse for Romney than Obama, with the caveat that many experts refuse to answer on the grounds that they don't know what Romney's plan is.

What about taxes?

Adam Smith, the founder of capitalism (and of economics) said "It is not very unreasonable that the rich should contribute to the public expense, not only in proportion to their revenue, but something more than in that proportion." Billionaire Warren Buffet complains that his taxes are too low: he paid 18% in taxes, and his secretary and other office staff members averaged 33%. He thinks he should pay more. Obama agrees; Romney thinks the rich should pay less, that his 14% rate is about right, and to tax the rich any more would kill jobs.

So: do tax cuts for the rich spur economic growth? Most evidence says: no. Conceptualmath.org is nonpartisan (except that they are partisan towards math), and concludes that "The historical evidence suggests that an economic decline will follow a tax cut to the rich, and economic growth may follow a tax increase to the rich. The evidence suggests that the optimum tax marginal tax rate on the rich is higher than 60%."

The nonpartisan Congressional Research Service concludes that "Tax rates affecting taxpayers at the top of the income distribution are currently at their lowest levels since the end of the second World War. The results of the analysis suggest that changes over the past 65 years in the top marginal tax rate and the top capital gains tax rate do not appear correlated with economic growth. The reduction in the top tax rates appears to be uncorrelated with saving, investment, and productivity growth. The top tax rates appear to have little or no relation to the size of the economic pie." Here are their charts showing real per capita GDP growth (positive or negative, as a percent) plotted against top tax rates (top marginal income tax rate, and top capital gains rate). It is clear there is very little correlation.

The US currently has a top marginal rate of 35%. So taxes on the rich are historically low, and the economy would (perhaps) improve if they were raised, or would perhaps be unaffected (we can't say for sure, since circumstances change over time, and since the past data is correlational, not necessarily causal).

What about overall tax rates compared to other countries? The US is in the middle in tax revenue as a share of GDP among all countries, and near the bottom among developed countries, with 27%, compared to 32% for Canada, 39% for the UK, 40% for Germany, and 49% for Denmark.

Even if tax cuts are not correlated with growth, wouldn't it be nice to pay less taxes? Obama wants 98% of Americans to pay less, exempting the richest 2% from cuts. Romney wants everyone to pay 20% less, but to eliminate some tax loopholes/deductions to recover the missing income. The nonpartisan Tax Policy Center estimates that Romney's plan would leave a $900 billion shortfall in 2015. The congressional Joint Center on Taxation reported that to remain revenue neutral, Romney could only cut rates by 4%, not 20%. Romney (and Ryan) have refused to answer questions about the specifics of their plan, so we can't say whether they would end up raising or lowering taxes on most people. Romney cites six blog posts and policy papers that say his plan would work. You can see a thorough discussion at debategraph.org.

Where do tax revenues come from? This chart shows that corporate taxes are decreasing as a percentage of GDP, so the scary headlines like US corporate tax rate: No. 1 in the world are silly; it doesn't matter what the nominal rate is, what matters is what companies end up paying after all the deductions. Overall the chart suggests that companies are getting a great tax break (as Warren Buffett points out), individuals are paying near historic lows in income and misc. taxes, but social insurance is going up.

What about spending?

Depending on what news sources you frequent, you can find facts, figures, and charts that show that the Obama administration has been the most wasteful spenders in all of US history, or the most prudent budgeters in the past 40 years. That is because there are many figures to consider, and each tells a different story. I'm going to focus on two figures: total spending in constant dollars (that is, adjusted for inflation, in this case to a baseline in 2005), and the percent increase in spending, compared to a baseline measured over the previous 16 years (again, adjusted for inflation). Here's what we get, first for spending:

Here's what it means: government spending goes up over time, at a fairly constant rate. This is not surprising; the country is growing and has more people to serve; also, as the country (and world) gets richer, people expect that government can take on some things now that we couldn't afford to take on in 1930. So, yes, you can say that Obama has spent more than any president in history. You could also say that about every other president before him, going back to Herbert Hoover on this chart. You can also see that the rise is starting to flatten out under Obama. How much? It looks like more flattening than any other president, but that's what the next chart is for:

Here we see that the biggest increases were around the big social programs of the 20th century: the new deal under FDR, and the great society under LBJ. Obama's percent increase is at a historic low (mostly because George W. Bush's spending was high). Overall, I look at this and say that Obama's spending has been surprisingly low--he kept us out of a depression with a small increase; not only much smaller than FDR's response to the great depression, but smaller than all recent presidents.

Now let's bring the last two questions together. When we combine spending and revenue (which is mostly taxes), we get at the deficit. Here is what the numbers look like:

Here we see a fairly large gap between spending and revenue in the 1930s; a stimulus was needed to get out of the great depression. We returned briefly to a balanced budget, but World War II required a huge increase in spending. That investment was worth it, and from 1950 until the mid-1970s we had roughly balanced budgets. Then the recession in the Carter years and the spending under Reagan opened up a deficit. Clinton was able to close the deficit, but Bush opened it up again (by cutting taxes more than by increasing spending). The 2009 recession required another stimulus, so spending was up by a huge amount, $535 billion, in 2009 (although this is less, in relative terms, than in the great depression and much less than in WWII). About 2/3 to 3/4 of this spending was due to Bush's actions; the rest by Obama. Spending has stabilized over the last 3 years, but tax revenues remain low, so we see a deficit gap. Overall, I think Obama's actions have been an appropriate response to the recession. We saw that Bush was committed to a large part of this spending, and it is not clear that Romney, had he been president, would have acted significantly differently. Now that the financial crisis has subsided, the question is how to move forward. Obama has spending level for the last three years, and has a plan to increase revenue, mostly by letting the Bush tax cuts on the rich expire. I can't rate Romney's plan, because I don't know what it is.

What about State vs. Federal?

On Oct. 28, as Hurricane Sandy was bearing down on the East Coast, a video clip from September emerged of Romney implying that he would shut down FEMA, the Federal Emergency Management Agency. This makes him look bad in many people's eyes, and he had to backtrack and say that he would keep FEMA after all.

But let's forget about FEMA for the moment and concentrate on something Romney said: "Every time you have an occasion to take something from the federal government and send it back to the states, that's the right direction." Now, Romney can't possibly mean that, for example, there should be no United States military, just 50 state militias. So let's assume that what Romney really meant is that most of the time functions should go to the state level. Is that reasonable?

The constitution lays out what powers go to the federal government, the states, or are shared between them. The most important conflict in state's rights was the Civil War, so there is still an association of state's rights with the right to subjugate part of the population. That said, within the bounds of the constitution there will always be a tradeoff. Doing things locally makes sense if local judgment will be better. It probably would be better for farmers and fisherman to make most decisions based on the specifics of their local lands and waters, rather than having a Federal official trying to understand all the different local conditions. On the other hand, it doesn't make sense to duplicate work 50 times over. If there is, say, a chemical in common use that may or may not be dangerous, it makes sense for a Federal agency to go to the expense of determining the truth, rather than have each state contract to do expensive testing. I am concerned that if Romney literally believes that his main goal is to move powers to the states, he will make inefficient choices, leading to more spending, inefficient use of shared resources, and slower adoption of best practices and knowledge. He should strive to have decisions and programs at the right level--Federal, State, County, or City--not to automatically move everything away from the Federal level.

Obama has taken a policy approach, meaning that he considers problems as large interacting systems, and looks for solutions to the big picture, rather than just considering one bill at a time. I think this is the correct way to look at issues. Ironically, this is exactly how succesful companies approach their business, but in the case of Federal/State decisions it is Obama, not Romney, who advocates this problem-solving approach.

Who endorses each candidate?

You can see endorsements from newspapers (an alternate source); currently Obama leads with endorsements from 55% of the newspapers (accounting for 64% of the circulation, and it would be 67% if The New Yorker counted as a newspaper). Also other endorsements for Obama and Romney. My friend Paul Graham, polled some of his peers (not me), and found that Silicon Valley leaders favored Obama by a more then 2-1 margin. Even though these were overwhelmingly rich business-people who would pay less taxes under Romney, apparently they were attracted to Obama's business and/or social policies. (My guess is that this is due to an alignment of personalities, more than a careful analysis of policies. Silicon Valley leaders have, for the most part, succeeded by taking risks, being open to new experiences, and embracing scientific ideas even if they lead to unexpected results. This is consistent with a liberal/progressive outlook, and less consistent with a conservative outlook.)

In general, Republicans endorse Romney and Democrats endorse Obama, to nobody's surprise. A few recent semi-surprises: Obama was endorsed by centrist NYC mayor Michael Bloomberg, in part because of Obama's treatment of Hurricane Sandy, but also because of climate change in general, and for women's rights. Obama was also endorsed by the centrist magazine The Economist, in a lukewarm way. The Economist says America could be better than Obama, but sadly, Romney is worse: "For all his businesslike intentions, Mr Romney has an economic plan that works only if you don't believe most of what he says."

Are the candidates always honest?


How honest or dishonest are they?

Probably the best referee in this game is Factcheck.org, from UPenn's Annenberg Political Policy Center, a nonpartisan nonprofit consumer advocate for voters. You can read a recent articles from Factcheck.org, and also from CNN's fact check.

The Washington Post has some fun with their ratings, giving them in terms of number of Pinocchios. The St. Petersburg (FL) Times, an independent non-profit paper, runs PolitiFact and has a similarly fanciful scale that runs from "True" to "Liar, Liar, Pants on Fire." Romney's statements are rated as 16% false and 9% pants-on-fire, while Obama has 14% false and 2% pants-on-fire. So, according to PolitiFact, Romney is 4.5 times more likely to tell a whopper of a lie, and tells more overall lies at a 5/3 margin over Obama.

You can see Obama's promises kept, Obama's file, and Romney's file.

Recently, Romney has come under unusual criticism for claiming that GM and Chrysler plan to ship jobs to China. Usually big companys like to stay out of the fray (because they know they have customers on all sides). But this time a GM spokesperson stated that "This ad was so misleading, that we had no choice but to offer a vigorous defense of our progress since emerging from bankruptcy. ... The ad is cynical campaign politics at its worst. ... We think creating jobs in the U.S. and repatriating profits back in this country should be a source of bipartisan pride." Similarly, Chrysler responded "I feel obliged to unambiguously restate our position: Jeep production will not be moved from the United States to China. ... It is inaccurate to suggest anything different."

Of course, dishonesty in presidential elections is nothing new. In the 1800 election, according to Wikipedia, "Federalists spread rumors that the Democratic-Republicans were radicals who would murder their opponents, burn churches, and destroy the country."

What happens in case of an electoral vote tie?

It could happen. (There was a tie in 1800.) Nate Silver's simulations at 538 give about a 0.2% chance of a tie this time. This is a lower chance of a tie than in 2008; even though this is a closer election than 2008, the way the states break down, it is unlikely that they will fall exactly 269-269.

But if there is a tie, the 12th amendment dictates that the incoming congress then gets to choose, but the exact rules are complex. The Senate chooses the vice-president; one senator one vote. Currently the prediction is for a Democratic majority, so this scenario welcomes Joe Biden as VP. Meanwhile, the House of Representatives chooses the president. But here it is one state one vote. The 53 Californian representatives get together to make their choice, and the 1 North Dakotan representative has an equal vote. Most likely, the Republicans will hold a majority of states and will pick Mitt Romney.

Who else is running?

Candidates include Libertarian Gary Johnson, Virgil Goode of the Constitution Party, and Jill Stein of the Green Party. You can see the complete list of third-party candidates. Most of them are polling well below 1% in most states, but Johnson and Goode in particular get in the 1% to 2% range in several states (even 4% in some polls) and could potentially affect the outcome (as Nader did in Florida in 2000). Historically, third-party candidates are better at getting polling numbers than actual votes; a candidate with 3% in pre-election polls might end up with 1%, as supporters shift their vote at the last minute to someone with an actual chance of winning.

I have been criticized for not paying more attention to third party candidates. In leaving them out, I am not making any comment on their suitability to be president. Rather, I am acknowledging the fact that their probability of becoming president is near-zero.

A vote for president does two things: first, it has a chance of actually making the difference that gives one candidate a win. Second, it records a tally, regardless of the winner. A vote for a third party candidate has no effect on the first issue (which is what this FAQ is about) but does have an effect on the second. In fact, I have exercised this option in the past, voting for a third party as a protest, because I wanted to record the fact that I was not satisfied with the two-party choices (also, the winner of my state's electoral votes was never in doubt). If you want to know your options for recording a protest vote, you'll need a different FAQ.

Is it rational to vote?

Yes. Voting for president is one of the most cost-effective actions any patriotic American can take.

Let me explain what the question means. For your vote to have an effect on the outcome of the election, you would have to live in a decisive state, meaning a state that would give one candidate or the other the required 270th electoral vote. More importantly, your vote would have to break an exact tie in your state (or, more likely, shift the way that the lawyers and judges will sort out how to count and recount the votes). With 100 million voters nationwide, what are the chances of that? If the chance is so small, why bother voting at all?

Historically, most voters either didn't worry about this problem, or figured they would vote despite the fact that they weren't likely to change the outcome, or vote because they want to register the degree of support for their candidate (even a vote that is not decisive is a vote that helps establish whether or not the winner has a "mandate"). But then the 2000 Florida election changed all that, with its slim 537 vote (0.009%) margin.

What is the probability that there will be a decisive state with a very close vote total, where a single vote could make a difference? Statistician Andrew Gelman of Columbia University says about one in 10 million.

That's a small chance, but what is the value of getting to break the tie? We can estimate the total monetary value by noting that President George W. Bush presided over a $3 trillion war and at least a $1 trillion economic melt-down. Senator Sheldon Whitehouse (D-RI) estimated the cost of the Bush presidency at $7.7 trillion. Let's compromise and call it $6 trillion, and assume that the other candidate would have been revenue neutral, so the net difference of the presidential choice is $6 trillion.

The value of not voting is that you save, say, an hour of your time. If you're an average American wage-earner, that's about $20. In contrast, the value of voting is the probability that your vote will decide the election (1 in 10 million if you live in a swing state) times the cost difference (potentially $6 trillion). That means the expected value of your vote (in that election) was $600,000. What else have you ever done in your life with an expected value of $600,000 per hour? Not even Warren Buffett makes that much. (One caveat: you need to be certain that your contribution is positive, not negative. If you vote for a candidate who makes things worse, then you have a negative expected value. So do your homework before voting. If you haven't already done that, then you'll need to add maybe 100 hours to the cost of voting, and the expected value goes down to $6,000 per hour.)

By the way, here's proof that voting is patriotic: If you think of the value of your vote to the country, $600,000, then obviously voting is the right thing to do. But if you think of the value to yourself, by dividing the $600,000 by 300 million Americans, then the benefit of voting is less than a penny, and the rational choice would be not to vote (since it costs you an hour of your time). So anyone who votes is patriotic, and anyone who doesn't is selfish (or irrational). Are you a rational patriot?

You can see a similar argument in a 23 page pdf article by Andrew Gelman.

This also suggests that election financing--all the contributions to pay for all the ads, etc.--is also a bargain. The total cost so far for the 2012 presidential election has been about 2.5 billion dollars. For that, we may get a return of several trillion dollars; a 100,000% return on investment, maybe more! The total spent on the presidential campaign plus all the house and senate campaigns will be about $5.8 billion; less than America spends on potato chips each year.

Who do you think will win?

I think the poll aggregators have it down to a science, and Obama will win. I agree with the poll aggregators that the chance that this year's polls are systematically biased against Romney is low, somewhere in the 2% to 15% range, but I can't be more precise than that.

Here is my state-by-state prediction. It is the same as the 2008 map except that I predict FL, IN, and NC will move over to the Republican side, In putting FL on the Republican side I agree with Sam Wang and disagree wtih Nate Silver and Drew Linzer.

Why do you support Obama?

First, let's look at what Obama has done. We'll focus on the campaign promises he made and kept:

If that's not enough, Washington Monthly has a list of Obama's Top 5o Accomplishments.

Next, let's compare Obama's vision to Romney's.

Jobs. One of Romney's best talking points is that he has "a plan to create 12 million new jobs." This sounds great; in recent months we've been adding about 150,000 jobs per month, so it would take 80 months to get to 12 million at this pace; Romney promises to get there in the 48 months of his term. But Moody's Analytics and the Macroeconomic Advisors already predict a gain of 12 million jobs regardless of who is president, so this doesn't favor one candidate over another. On the other hand, a study by the (left-leaning) Economic Policy Institute finds that the Romney plan would result in a loss of 1/2 million jobs.

Taxes. Romney promises to cut taxes to everyone by 20%, then make up for it by closing loopholes. The nonpartisan Tax Policy Center reports that that is impossible; this would cut $86 billion in tax revenue from people making over $200,000 a year, and it is not possible to make up for that in loopholes; therefore Romney must either increase the deficit or raise taxes on the middle class. Or perhaps eliminate popular deductions such as mortgage credit or charitable giving deductions. After quite a bit of back, and forth, I believe this analysis still holds. As David Frum says ""Romney's tax cut plan doesn't work. I'm a Republican, I support Romney, etc. But you can't cut that much in such a stagnant economy and expect to break even. Even with a deductions cap, it just won't happen."

As stated above in the section on taxes, there is little or no correlation between highest marginal tax rate and economic performance. This seems sensible to me. I know that my business experience is not typical of all of America, but in my experience, job creation is a partnership of entrepreneurs who create new products, and consumers who buy the product. Although I know a dozen billionaires and over a hundred millionaires, I never once heard one express the idea "I was going to start a business, but with the marginal tax rate at 39% instead of 35%, I guess I won't." Rather, every entrepreneur I know is interested in whether their venture will have a 1000% return or a 10,000% return -- the tax rate is irrelevant. I think Obama's policies fit the world as I see it: we need small businesses and consumers to grow together. Obama is concentrated on helping them; Romney seems to concentrate on making the rich richer.

Civil Rights. Obama repealed the "Don't Ask Don't Tell law," thereby allowing gays to serve openly in the military. Public opinion supports this: 77% agree with the repeal. The Chairman of the Joint Chiefs of Staff agrees, and as DOD Fed Globe reports, "In November 2010, the Defense Department published a comprehensive report on the effects of repeal. The report included the results of a survey of military personnel on bases throughout the United States and overseas, including 400,000 service-members and 150,000 military spouses. Overall, 70 percent of military personnel thought that integrating gays into the military would be positive, mixed, or of no consequence." Romney opposes same-sex marriage and civil unions, and in the past has supported Don't Ask Don't Tell, and has opposed the rights of gay couples to adopt or bear children. In addition, Obama instructed the Department of Justice to use the criteria of "heightened scrutiny" to examine discrimination against gays, an important distinction, recently upheld by the Second Circuit Court of Appeals. Americans are increasingly coming to accept that sexual orientation is not something we should be discriminating against (just as in the 1960s and 1970s they came to accept color of skin) but Romney opposes equal rights with respect to this issue.

For women's rights, it is hard to know where Romney stands, because he has reversed many of his positions today from when he was serving in Massachusetts, and he has refused to answer some questions. For example, Obama's first legislative action was signing the Lilly Ledbetter Fair Pay Act, supporting equal pay for women. Romney hasn't said whether he supports it (although he has said that he will appoint justices like Samuel Alito, who wrote the original Ledbetter opinion that limited the right to equal pay for many women. We do know Romney wants to defund Planned Parenthood, thereby eliminating access to medical help for millions of poor women. We know that he said during the campaign that he wants to make it possible for health practitioners to deny access to contraceptives to women. As Massachusetts governor, he vetoed a bill that would give rape victims access to emergency contraception. During the Republican primaries, he stated that he would like to reverse the 1965 case Griswold v. Connecticut, the case that established the right to marital privacy in the use of contraceptives. This makes it seem like he wants to roll back progress by 50 years. However, in the second presidential debate, Romney reversed his position, saying "Every woman in America should have access to contraceptives." So now it is not clear what his position is. It is likely that Romney's policies would increase the number of abortions performed in America, although it is hard to pin down an estimate of how much: probably over 10,000 additional abortions if Romney's policies are instituted. Recent news shows that teen pregnancy rates are at an all-time low in the US, largely because girls have access to effective contraception. The GOP platform states a goal of making all abortion illegal, but it is not clear exactly what Romney's policy is. He has said that he would nominate Supreme Court justices that would overthrow Roe v. Wade, thereby making it possible for states to make abortion illegal.

The 47%. One of the biggest stories of the campaign was a recording of Romney speaking to supporters and saying:

"There are 47 percent of the people who will vote for the president no matter what. There are 47 percent who are with him, who are dependent upon government, who believe that they are victims, who believe that government has a responsibility to care for them, who believe that they are entitled to health care, to food, to housing, to you name it. Forty-seven percent of Americans pay no income tax ... I'll never convince them they should take personal responsibility and care for their lives."

I think almost everyone agrees that people should take responsibility, that people should not expect government to do everything for them. That is a true point, but a trivial, uncontroversial one.

But the statement hurt Romney, and it should (even after he said it was completely wrong). How was it wrong? It showed that Romney has a wrong view of how America works, and of what lives are like for those who aren't millionaires. Romney says that using a government service means you have no personal responsibility; actually, 96% of Americans benefit from government services; of the 4% who have not yet, most will as they grow older. Romney says that if you don't pay income tax, you have no personal responsibility. Most of the people who don't pay income taxes are children, or retired people, or serving in the military. Of the remainder, most are low-wage earners. A calculation shows that for a family of four with typical deductions and one parent working, the family would pay payroll taxes, but would only pay income tax if their total income exceeds $45,000. Suppose the worker in that family has a job at minimum wage, $7.25/hour. He or she would have to work over 119 hours per week, 52 weeks per year before having to pay income tax. Romney is saying that someone who works 119 hours per week (17 hours per day, 7 days a week) has no personal responsibility. That is completely wrong, but he can't make it right just by apologizing. It appears that that is what he really thinks, and that's not what I want in a President.

Peter Norvig peter@norvig.com