Michael P. McDonald

2012 Election: A Failure to Mobilize the Youth Vote

2013-05-10T16:44:44-04:00

Among the most dramatic findings reported in the Census Bureau's Current Population Survey (CPS) is a large decline in turnout rates among young people, particularly those who were first eligible to vote in the 2012 election. In contrast, older persons' turnout rates remained steady, or even increased. Given the large disparities in support for Obama among younger and older voters, Obama's smaller margin of victory in 2012 was thus partially a turnout story, as the electorate's composition was older, and more favorable to Republicans in 2012 compared to 2008.

Younger people are less likely to vote, a pattern evident in all election surveys. These younger people may thus benefit from campaigns' mobilization efforts to register and encourage them to vote. These mobilization efforts, like much else in presidential campaigns, are concentrated in the closely fought battleground states -- which I define here as Colorado, Florida, Iowa, Nevada, New Hampshire, North Carolina, Ohio, Virginia. The turnout rate decline from 2008 to 2012 may thus be expected to be lower among the battleground states, particularly for younger people.

Failing to Mobilize the Youth Vote

Below, in a methodology section, I argue that the adjusted citizen-voting-age population (CVAP) turnout rates, in the far right column under "CVAP Turnout Rate (Adj.)", the most reliable to assess patterns in youth vote. I provide the unadjusted turnout rates under "CVAP Turnout Rate" as a reference, but I will only discuss the former. Turnout rates in 2008 and 2012 are presented for various age categories in both battleground and non-battleground states. I discuss overall national rates here.

The key statistics in the table is that turnout rates among the youngest people, for whom the 2012 election was their first election, declined in both the battleground states and the other states, by 6.2 percentage points and 8.8 percentage points, respectively. Conversely, turnout rates for older people actually increased between 2008 and 2012, even more so in the battleground states.

Since 2004, younger and older people hold increasingly different preferences over Democratic and Republican candidates. Obama's lower national vote margin in 2012 compared to 2008 can thus be understood as a turnout story, with an important Democratic constituency removing themselves from the electorate and an important Republican constituency reemerging.

The Obama campaign appears to be only marginally successful at counteracting youth disengagement through their mobilization efforts targeted at the youngest people in the battleground states. A significant challenge thus lays before Democrats in 2014 and beyond: they will need to double their mobilization efforts to target not only young people entering the electorate since 2012, but also those who failed to participate in 2012. And for Republicans, there is an opportunity: abstaining young voters have not staked a partisan attachment yet and may still be persuaded to vote for Republican candidates if first mobilized by Republicans.

A Survey Methodology: A Digression into Non-Response Bias

Aggregate voter turnout statistics compiled from official election results demonstrates the turnout benefits that accrue to battleground states. Between 2008 and 2012, the presidential turnout rate for those eligible to vote declined 1.8 percentage points in the battleground states (from 66.2 percent to 84.4 percent). In the other states, turnout rates declined 4.0 percentage points (from 62.1 percent to 58.1 percent).

The aggregate statistics do not reveal turnout patterns among age groups. For that, we need to examine survey data. Of course, there are errors in survey data. Some are well-known, such as the margin of error that arises from statistical sampling. Less well-known issues that survey researchers wrestle with are over-report and non-response bias. Over-report bias is a phenomenon whereby more people report voting than the official statistics indicate. While it may be tempting to think of these people as lying, by trying to portray themselves in a favorable light to the interviewer, the actual mechanics are not perfectly understood. A potential related issue may be over-report bias, that those who choose to participate in a survey are generally more participatory people.

The Current Population Survey is generally regarded as a high-quality survey. Its primary purpose is to calculate the monthly unemployment rate. The Census Bureau goes to great lengths to ensure that those who are selected to be in the sample respond to the survey, to mitigate potential non-response bias. However, people may still choose not to answer specific questions, particularly those in the voting and registration supplemental questionnaire, which is administered in the November of a federal election year. Indeed, the supplement may not even be administered to a respondent.

Non-response is a serious issue, with 12.8 percent of CPS respondents failing to respond to the vote question in 2012, which was slightly lower than the 13.8 percent in 2008. While the higher response rate in 2012 is welcome, that does not necessarily mean that all respondents improved their responsiveness in the 2012, as we shall see. Furthermore, the CPS's practice is to include non-respondents as having not voted, even those who were never administered the supplemental questions. That's right. A person who was never asked the voting question is marked as having not voted. A more reasonable approach would be to exclude these people from the turnout rate calculations.

With this in mind, over-report bias and non-response bias are clearly evident among the CPS voting questions. For example, in 2012 the presidential turnout rate among those who were eligible, according to the actual vote tallies, was 58.7 percent, compared to 63.6 percent, according to the CPS. The CPS over-report bias is more severe when non-respondents are removed, with 70.9 percent reporting having voted.

Even though over-report bias may be present, the CPS still provides valuable information. We can see, for example, the expected pattern of turnout decline among battleground states and other states, with the CPS indicating a 1.4 percentage point decline among battleground states (1.0 points when non-respondents are removed) compared with 1.8 percentage point decline using the official election results. The decline among non-battleground states is 1.9 percentage points for the unadjusted CPS turnout rates, 3.4 percentage applying the non-response adjustment, and 4.0 points using the official statistics. With the official election results as a validation point, the CPS turnout rate adjusted for non-response thus appears to be more valid than the unadjusted turnout rate.

2012 Turnout: Race, Ethnicity and the Youth Vote

2013-05-08T16:59:54-04:00

An important election survey that reveals patterns in voting and registration is the Census Bureau's Current Population Survey November Voting and Registration Supplement, or CPS for short. The 2012 CPS reveals insights to major stories about the election divined from the exit polls: the changing face of the electorate and the role of young people in determining the outcome of the presidential election. As I suggested previously, the increasing diversity of the 2012 electorate was a partially a turnout story, with non-Hispanic Whites modestly withdrawing from the electorate. The CPS further documents how it is also a story of the inevitable trend of increasing diversity of the country. Perhaps the most revealing new finding is a dramatic decrease in the youth vote, which has important ramifications for future elections.

The Changing Face of the Electorate

Among the major stories that emerged from the media's 2012 exit polls was how the non-White share of the electorate increased by two percentage points from 2008. The exit polls are, of course, only a survey of voters. The Census Bureau's CPS survey illuminates how turnout played a factor in the increasing non-White composition of the electorate, as suggestively evident in the aggregate election results.

According to the CPS, non-Hispanic White citizen voting-age population turnout rate decreased from 2008 to 2012 from 64.4% to 62.2%, or 2.2 percentage points. In raw terms, 1.3 million fewer non-Hispanic Whites voted in 2008 than 2012. While the country is becoming more ethnically diverse, non-Hispanic Whites are not experiencing a negative growth rate, rather other races and ethnicities are outpacing non-Hispanic White population growth. Comparatively, African-Americans experienced a small turnout rate increase -- although, I discuss below how this increase is an artifact of non-response bias. Even though the Hispanic turnout rate decreased 1.9 percentage points, 1.4 million more Hispanics voted in 2012 compared to 2008.

Thus, part of the changing face of the American electorate is a turnout rate story, with a small decline in non-Hispanic White turnout rates compared to other races. Part of it is also a true change in the composition of the country's citizenry, as the country becomes more diverse.

The Youth Drop Out

Republicans may feel encouraged for their future election prospects, in that fewer non-Hispanic Whites voted in 2012 than 2008, since Whites tend to vote Republican. However, in reality there is both good news and bad news for Democrats here, and an opportunity for Republicans.

The good news for Democrats is that the decline in non-Hispanic White turnout was primarily due to younger people abstaining. Turnout among all 18 to 24 year olds dropped 7.3 percentage points between 2008 and 2012. Indeed, this decline is evident across all races and ethnicities. Non-Hispanic White turnout among these youth dropped 7.4 percentage points; it dropped 6.7 percentage points for African-Americans; and it dropped 4.5 percentage points for Hispanics. Conversely, turnout rates increased for persons age 65 and older -- by 1.1 percentage points for persons age 65 to 74 and by 2.2 percentage points for persons age 75 and older -- a pattern that is again replicated among other racial and ethnic groups.

Lower youth turnout in 2012 is good news for Democrats from one perspective because Obama won in 2012 despite the withdrawal of this important Democratic constituency. Part of Obama's lower vote margin in 2012 compared to 2008 is due to lower youth turnout. However, there is certainly much for Democrats to be concerned about. Academic studies find that voting is habit-forming. Once a person has voted they are more likely to vote again, perhaps because they have learned how to navigate the system. The Obama campaign failed to mobilize the youth cohort who first became eligible to vote in the 2012 election, and Democrats will thus need to work harder in future elections to activate these people by registering them and helping them vote. These young people also present an opportunity for Republicans because they are the most persuadable potential voters, in that they have not yet become as fixed in their partisanship as people tend to do as they age. If Republicans can contact these young people first, they may be able to sway them to vote Republican.

Non-Response Bias: Did Black Turnout Really Exceed non-Hispanic White in 2012 for the First Time?
The short answer is that African-American turnout first exceeded non-Hispanic White turnout in 2008, not 2012. It did again in 2012.

The long-answer for polling nerds, involving non-response bias, follows.

The Census Bureau took a welcome step towards greater transparency in the 2012 Voting and Registration report regarding potential non-response bias -- or those who fail to participate in a poll or answer questions. Past Census Bureau practice has been to count persons who were not administered the voting and registration supplement, refused to answer a question, or did not know an answer as being a "no" response to questions such as if a person was registered or voted. While this practice continues in the 2012 report, for the first time the Census Bureau reports non-response in the detailed tables accompanying the report. While it has been possible to identify and remove non-responders using the raw individual-level data, the new practice enables more casual users of these data to make non-response adjustments.

Why does non-response matter? As I have discussed previously, non-response to the voting and registration supplement has been increasing over time and varies across important demographic groups -- such as race and ethnicity -- which can lead to erroneous conclusions when making temporal comparisons of registration and turnout rates. The good news is that the 2012 CPS non-response modestly declined from 2008 to 2012 -- from 13.8% in 2008 to 12.8% in 2012 -- reversing the upward trend.

The decline in non-response was mostly centered among African-Americans, of which 18.0% did not respond in 2008 compared with 15.0% in 2012. In comparison, non-Hispanic White non-response declined from 12.8% in 2008 to 12.2% in 2012.

The non-response has an important implication for a major finding reported on p.3 of the Census Bureau report:

Blacks voted at a higher rate (66.2 percent) than non-Hispanic Whites (64.1 percent) for the first time since the Census Bureau started publishing voting rates by the eligible citizenship population in 1996.

Keep in mind, the Census Bureau counts persons who did not respond to the survey as not-voting. I provide corrections to the tables presented above by removing those who did not provide a response t the voting question from the turnout rate calculations.

When non-respondents are removed from the turnout rate calculations, African-Americans first exceeded non-Hispanic White turnout in 2008 -- 78.9% to 75.5%. They did so again in 2012, 77.9% to 72.8%. Also of note is that African-American turnout rate is the only group to exhibit a turnout rate increase between 2008 and 2012 -- by 1.5 percentage points -- when counting non-respondents as having not voted, but decreased by 1.0 percentage points when non-respondents are removed. In the latter case, African-American turnout simply declined by a more modest amount than other races and ethnicities.

These adjusted numbers may help resolve another incongruity in the CPS survey data. The Census Bureau reports that the overall number of voters increased from 131.1 million in 2008 to 132.9 million. This can't possibly be correct since my tabulations from official election results show the overall turnout declined from 132.7 million to 130.7 million. The overall unadjusted turnout rate decreased by 1.8 percentage points while the adjusted turnout rate decreased 2.9 percentage points. The decline in the adjusted turnout rate is more in line with the decline observed among the administrative records, of 3.5 percentage points.

Summary

The big news from the CPS will be that African-American turnout exceeded non-Hispanic White turnout. But I contend that really that happened first in 2008. The more important political story may be the dramatic decline in youth turnout, as it poses a challenge to Democrats and an opportunity for Republicans. There is much more that will follow in the coming days -- from voter registration to early voting -- so stay tuned!

Who Registers to Vote Online?

2013-04-29T14:02:08-04:00

A sensible election administration reform is quietly sweeping the nation. According to the National Conference of State Legislatures, 18 states have implemented or recently adopted online voter registration, either initiating a new registration or updating an old one. Twelve other states have legislation winding its way through the legislative process. The reform is bipartisan in that both Democratic- and Republican-controlled state governments have adopted it, from Arizona to Maryland.

Legislators are attracted to online voter registration because it offers substantial election administration savings. Arizona, the first state to adopt online voter registration in 2002, reports that over 70 percent of registrations are now conducted online. The old paper system cost 83 cents to process each registration form, compared to 3 cents for the online system. The online system is more reliable than the paper system, reducing data entry errors that can disenfranchise voters and introduce other election administration costs when communications -- such as absentee ballots -- from election officials to voters are sent to a bad address. With state and local governments strapped for cash, online voter registration can reduce election administration costs by millions of dollars while simultaneously improving the integrity of the system. And for those who are concerned about fraud, federal law requires first time registrants to provide identification before they are allowed to vote.

So, who uses online voter registration? Which party may benefit? Data provided by the Maryland State Board of Elections illuminates patterns of online voter registration use. Keep in mind, Maryland is just one state, so these observations may not be applicable elsewhere.

A substantial number of people -- 124,819 -- used Maryland's online system from the date the system went live on July 1, 2012 through the most recent date where I have data, April 11, 2013. Among these, 39.8 percent used the online system to initiate a new registration and 60.2 percent updated an existing registration.

A simple comparison of all registered voters to those using the online system reveals that online registrants tend to be less often registered as Republicans -- by 8.5 percentage points -- and more likely to register as unaffiliated or a minor party -- by 8.0 percentage points. The partisan implications are likely a consequence of young people being much more likely to use the online system. For example, only 18.8 percent of registered voters are age 18-29, compared to 41.6 percent of online registrants. Women are also slightly more likely to use the online system compared to all registered voters, by 1.0 percentage points.

Young people are less likely to be registered to vote and might thus be more likely to need to use the online system. A better comparison to understand online voter registration usage patterns is to examine those who registered for the first time using the old paper system and the new online system. Ideally, those who updated their registrations using the two systems would also be compared, but because of data limitations only those who used the online system to update their registrations can be identified. There were 49,712 people who used the online system to initiate a new registration and 75,107 who used it to update an existing registration. There were 197,056 new registrants who used the old system to initiate a new registration since the online system went live on July 1, 2012.

New registrants using the online system are more likely to register as Republicans -- by 3.4 percentage points -- compared to new registrants using the old system. The age distribution among online users is slightly older, too, and just slightly more male. The well-known digital divide is likely a factor. Notably, the online pattern is evident despite another well-known divide that might work against it, where rural areas have less broadband access than urban areas. Fewer new registrants used the online system in more rural, Republican areas of the state -- 8.1 percent of new registrations using the old system originated from eleven counties in the panhandle or DelMarVa peninsula, compared with 6.3 percent using the online system.

Those using the online system to update an existing registration are strikingly more likely to be female -- by 5.7 percentage points -- than those who use the system to initiate a new registration. They are also strikingly more likely to be a registered Democrat -- by 13.0 percentage points -- and tend to be older. Reflection of those who need to update their registrations may at least partially explain these patterns. While some are people who have recently moved, some are also women who are recently married, adopted their husband's last name, and need to change their last name.

Overall, these patterns suggest that the partisan implications of online voter registration are unclear. Registered Republicans appear to more often use Maryland's online system to initiate a new registration while Democrats use it to update an existing registration. We, of course, do not know how these people would have behaved if online voter registration was unavailable. However, another important lesson appears to be that online access has not matured to the point yet where an online system can substitute for good old paper.

Reassessing Colorado's Turnout

2013-04-14T15:17:49-04:00

Colorado legislators are contemplating changes to their election laws this week that may result in higher turnout. A report issued by Colorado Secretary of State Scott Gessler figures prominently in the deliberations, which triumphantly touts "the most successful (election) in Colorado history," implying that no changes are needed.

However, claims in the report that Colorado's voter turnout increased compared to 2008, while turnout declined elsewhere are not supported by the facts. I reassess the report with the best available data -- my United States Elections Project -- which the Colorado Secretary of State's office analyzed selectively. With Colorado considering changes to its election laws, policy makers should make their decisions with the most accurate information.

The Colorado Secretary of State issued their "2012 General Election Review: A Colorado Success Story" on February 7, 2013. My voter turnout statistics disseminated through the United States Elections Project are cited as the source for the 2008 election and preliminary reports from the New York Times are cited as a source for the 2012 election, as are my statistics -- but only where they are compared favorably.

Analyzing these statistics, the Secretary of State asserts that Colorado's turnout rate increased by 1.8 percentage points, while turnout rates nationally and among swing states decreased, by 6.8 and 1.2 percentage points, respectively.

The Secretary of State's office did not need to analyze preliminary 2008 reports from the New York Times, which can be subject to error as they are often projected from incomplete election night reporting. By the time the Secretary of State's office issued the report, my 2012 turnout rate statistics were available and a picture of Colorado's and the nation's turnout had emerged. Indeed, on p.2 of the report, United States Elections Project 2012 turnout rates are cited to claim -- inaccurately -- that Colorado had the highest voting-eligible turnout rate as a percentage of total ballots counted.

To provide accurate information for policymakers, I recreate the analysis in the Colorado Secretary of State's report using the current best turnout rate calculations (details regarding the construction of these statistics are below).

Contrary to the Colorado Secretary of State's self-congratulatory claims, Colorado did not appear to experience a turnout rate increase from 2008 to 2012. Instead, the state had a modest decline. Colorado still has much to be proud about, if we peel back the hyped claims. The state's participation decline was less than many other states, likely due to the presidential campaigns expending considerable resources to mobilize voters in this pivotal battleground state. Likewise, Colorado did not have the highest turnout rate as claimed in the report -- Minnesota retained its perennial crown with a VEP total ballots cast turnout rate of 76.1 percent, compared to Colorado's 71.1 percent. Interestingly, Minnesota has Election Day registration, a form of which is among the reforms Colorado policy makers are contemplating.

As Colorado policymakers consider reforms, I urge them to listen to their local election administrators since they know best how to efficiently run elections to serve their voters. New laws should not be contemplated because one side or the other may be advantaged by perceived higher or lower turnout. Going beyond the rhetoric and popular perceptions, it is often surprisingly difficult to determine which party benefits from an election law change. There are numerous counter-intuitive examples where an election law has the opposite effect from what is widely believed. Local context appears to matter most: the capacity for the parties to take advantage of a new law to mobilize their supporters. The test for a new election law should not be partisan advantage, it should be whether or not it improves the voting experience and improves the reliability and efficiency of election administration.

Notes about statistics in the tables

In the tables, 'VEP' refers to voting-eligible population and 'VAP' refers to voting-age population. In the "Voter Participation" table rates are calculated as the presidential vote divided by the VEP. I follow the Colorado Secretary of State's report in identifying the swing states as CO, FL, IA, NC, NH, NV, OH, PA, VA and WI. Notably, three swing states experienced increases from 2008: IA, NV, and WI (as well as DC).

Although Wisconsin has not reported total ballots cast yet, the state's VEP turnout rate for presidential vote exceeds Colorado's VEP turnout rate for total ballots cast. Wisconsin's VEP total ballots cast turnout rate thus logically exceeds Colorado's and I have ranked Colorado behind Wisconsin accordingly.

Turnout in the 2012 Presidential Election

2013-02-11T13:09:57-05:00

The reelection of Barack Obama was tarnished by a lower voter turnout rate than 2008, dropping from 61.6% of those eligible to vote to 58.2%, or a decrease of 3.4 percentage points. Here, I place the 2012 turnout rate in historical perspective and analyze where turnout decreased, and where it increased. These turnout patterns illuminate how participation is stimulated in battleground states, how much the devastation of Hurricane Sandy depressed turnout, and provide insights to the much-discussed rise in the minority share of the electorate.

The Big Picture

The 2012 election is placed in recent historical perspective in the graph below, which plots from 1948 to 2012 presidential turnout rates for those eligible to vote. From this perspective, the decline since 2008 is not alarming. Modern second term presidents successfully seeking reelection most often experience a participation decline. The only post-World War II presidents who enjoyed a turnout increase were Ronald Reagan and George W. Bush. Dwight Eisenhower, Richard Nixon, Bill Clinton, and Barack Obama combined for an average 4.3 percentage point drop in voter participation from their first election to their second.

Perhaps a turnout decline is to be expected for successfully reelected presidents. Presidents who fail to win reelection are defeated when a majority of citizens desire a new direction for the country, while citizens approve the job performance of those who win reelection. It is easier to mobilize people against than to approve a political figure. The two modern presidents who sought reelection and failed were Jimmy Carter and George H.W. Bush. The 1980 election saw a slight 0.6 percentage point turnout decline from 1976 and the 1992 election saw a dramatic 5.3 percentage point increase from 1988. (Gerald Ford also sought election, but did not formally seek reelection in his role of replacing Richard Nixon following the resignation.)

Furthermore, while the turnout rate declined in 2012, participation did not experience a wholesale collapse. The 2012 turnout rate is higher than all turnout rates from 1972 to 2000, a time when pundits and academics alike lamented a decline in voter turnout. The 2012 turnout rate of 58.2% is just slightly higher than the 1992 rate of 58.1%, arguably one of the more exciting recent elections due to the uncertainty surrounding the three-way race between George H.W. Bush, Ross Perot and Bill Clinton. Despite a dip in the road, American citizens continue to be politically engaged.

Minnesota led the states, again, with a turnout rate of 75.7%. Hawai'i was again the laggard, with a turnout rate of 44.2%. These two states are perennially in their respective positions. Minnesota puts voters first with voter-friendly policies known to increase turnout, such as Election Day registration. Perhaps Hawai'i can be forgiven for its poor showing since the state is far from the mainland: it rarely gets any attention from the presidential campaigns and many potential voters know who the presidential winner is before their polls close.

The Value of the Battleground

Voters in three states, plus the District of Columbia, were actually more engaged than 2008, albeit with a increase of less than a percentage point. Putting DC aside for a moment, these three states were Iowa, Nevada, and Wisconsin. An obvious commonality is that all three are 2012 battleground states. Indeed, voter participation in every battleground state declined less than the national average. The exception is Pennsylvania, although only the Romney campaign belatedly claimed the state was up for grabs. States that dropped off the battleground in 2012 were also among those that lagged even further behind the national decline. Indiana, Michigan, Missouri, New Mexico, and (Pennsylvania) all fell off the map.

Battleground states should reasonably have higher turnout. The presidential campaigns and outside organizations buy up airtime to disseminate their commercials, they recruit volunteers to talk to voters, and the candidates themselves often come to town. These mobilization efforts motivate voters who perceive that their vote will matter to participate in the election.

One might take from this that the national turnout rate would increase dramatically if the Electoral College was replaced by a direct popular vote for president, something I favor. However, I caution that the turnout gains might be smaller than expected. What is not known is how much of the simulative effect of battleground status is due to increased campaign activity, and how much is due to voters believing their vote matters more in a close election. If it is primarily the former, then campaign activity -- and the amount of money spent on presidential elections -- would have to increase dramatically to reach more voters in the currently sleepy corners of the country to appreciably increase turnout when electing the president by a national popular vote.

Home Field Advantage

The state that experienced the largest participation drop-off from 2008 was Alaska, with a 9.1 percentage point decrease. The polarizing figure of Sarah Palin comes to mind as a reason why Alaskans may have been excited to vote in 2008, but were disinterested in 2012. There may be something deeper going on, however, with a national candidate fostering local interest. Paul Ryan's home state of Wisconsin was among the three states that experienced a turnout rate increase. Two states closely associated with Romney -- Massachusetts and Utah -- had the smallest declines from 2008 to 2012. Similarly, John McCain's home state of Arizona and Barack Obama's home state of Illinois increased more than the national average from 2004 to 2008. (Sorry Joe Biden, Delaware did not). While scholars find a negligible vote share gain for candidates' home states, perhaps the real effect is people who both like and dislike the favored son or daughter are energized to participate, thus cancelling each other out.

Hurricane Sandy

The depressive effect of Hurricane Sandy on participation is evident in the three hardest hit states of Connecticut, New York, and New Jersey. All were among the ten lowest decliners, with an average decline of 5.6 percentage points. It is difficult to say with certainty what would have happened if Sandy had taken a different track. If these three states' decline kept pace with the national average, nearly a half a million more people might have voted in these states. While that is a lot of people -- and I do not wish to diminish these peoples' unfortunate troubles -- they contribute only two tenths of a percentage point to the national turnout rate.

Changing Face of America

Among the most striking demographic changes in the electorate from 2008 to 2012 is that the media's exit polls indicate the share of the electorate that calls itself "White" decreased from 74% in 2008 to 72% in 2012. Census Bureau demographers forecast that by about 2050, Whites will no longer be a majority of the population (they will still be the largest racial group), primarily due to the country's increasing Latino population. It stands to reason, then, that the exit polls are the proverbial canary in the coalmine, the first to document the coming changes to the American electorate.

Only those who voted are asked to participate in an exit poll. These polls cannot truly reveal if the country's overall citizenry is changing or if the face of voters is just different from 2008. The best source of information on the demographic composition of the electorate and associated turnout rates will be the Census Bureau's Current Population Survey 2012 November Voting and Registration Supplement. This survey will be released sometime in the Spring of 2013; in the meantime information like the exit polls fill the void.

The aggregate state turnout rates also provide clues to the changing face of the electorate. A striking pattern is that states with large African-American populations experienced less turnout decline than other states. Every state (including the District of Columbia) with an African-American voting-age population greater than twenty percent fared better than the national average decline, with the exception of Georgia. (A simple correlation between state turnout rates and African-American voting-age population from the 2010 census is .33, and this does not control for the other factors already mentioned.)

Although the exit polls report the percentage of the electorate that is African-American remained steady at 13% from 2008 to 2012, some rounding error may be present, whereby the 2008 African-American percentage was rounded up to 13% and 2012 rounded down. As for Latino populations, the correlation is negative (-.20), meaning that larger state Latino populations are associated with larger turnout rate declines. Hispanics may still compose a larger segment of these states' electorates than 2008, but they are likely not contributing as much as African-Americans to the changing demographic composition of the 2012 electorate. Unfortunately, Asian-American populations are too small to draw any meaningful aggregate patterns from.

I suspect that at least part of the story of the changing face of the 2012 electorate is that Whites turned out at lower rates than 2008. The Current Population Survey will ultimately provide confirmatory evidence. Before Democrats become too giddy at the small change in the demographic composition of the 2012 electorate, which is undoubtedly slowly occurring, it is further instructive to remember that minorities' turnout rates drop-off to a higher degree in midterm elections. According to the Census Bureau, White turnout rates decreased 21 percentage points from 2008 to 2010, while African-Americans decreased 25 percentage points and Latinos decreased 23 percentage points. In the short term, the drop-off of minorities -- and young people, too -- in midterm elections poses a challenge to Democrats who wish to control congressional majorities and to win the many state offices that are more frequently elected on the midterm calendar.

Conclusion

Turnout declined in the 2012 presidential election from 2008. Only a small portion of this decline can be attributed to Hurricane Sandy, which depressed turnout in the Northeast. Perhaps the decline is not surprising since the 2008 turnout rate was the highest since 1964, or nearly half a century. The 2012 turnout rate is still higher than the entire 1972 to 2000 period. While turnout declined for Barack Obama's reelection, this is typical for modern reelected presidents. Since World War II, only two of six presidents who were successfully reelected did not experience a turnout decline from their first to second election.

Looking deeper at the state numbers, a handful of states bucked the national decline with an increase, and others were more resistant to the downward pressures -- I'm talking primarily about the battleground states. The 2012 election thus serves as yet another reminder that Americans living in battleground states experience a different quality of democracy than those elsewhere. While America's changing face leaves an impression on the state level turnout rates, an outstanding question for Democrats in the 2014 election and beyond is whether minorities -- particularly African-Americans -- will continue to be as highly engaged as they have been in the last two presidential elections.

Correction: To my utter embarrassment, I previously included John F. Kennedy among those presidents who had been reelected.

Learn more about American elections in my forthcoming book from Oxford University Press, The Art of Voting. In the meantime, the numbers and other election resources can be found at here.

How Did People Vote and Who Did They Vote for?

2013-01-13T23:16:50-05:00

You know who won the 2012 presidential election. What you may not know is which candidate did best among the different ways that people vote.

A few states report their election results by how people voted: mail ballots, in-person early voting, Election Day voters, people who cast a provisional ballot, and even overseas voters. Often, this reporting is possible because these states report their results in separate precincts for each type of voting. There is no consistency among states as to how they report these election results; no state breaks down their results into all the voting method categories.

These statistics are reported in the Table: Obama Vote Share by Voting Method. The numbers can be a little confusing at first because of the varied way in which states report their statistics. So, let's start with describing the different voting methods.

Definitions

Election Day Voters are persons who cast a vote at a polling location on Election Day.

Early Voters: Mail and In-Person are votes cast prior to Election Day -- either by mail or in-person -- where a state does not further distinguish between these two voting methods. To give further perspective, the number of all early votes cast as a percentage of all votes is provided. States are also identified if in-person early voting is permitted and if voters must provide a valid excuse to vote a mail ballot.

Early Voters: In-Person are votes cast prior to Election Day in-person, where a state also reports election results for early voters who cast mail votes. Most states permit in-person early voting at special polling locations, but some do not.

Early Voters: Mail are votes cast prior to Election Day by mail, either where this is the only method of early voting permitted or where a state also reports election results for early voters who cast in-person votes. In most states where mail balloting is the only early voting option, voters must also provide a state-approved valid excuse, such as being absent from their residence or otherwise unable to travel to the polls -- except for the all-mail ballot states such as Oregon and Washington.

Provisional Ballots are votes that are usually cast when there is a problem with a person's voter registration or another issue at the polls. A few states also use them for other purposes, such as Florida and Ohio which use provisional ballots to process address changes on Election Day.

Overseas and Military Voters are also known by the acronym UOCAVA voters for the federal Uniformed and Overseas Civilian Absentee Voting Act -- that is, overseas civilian and domestic and overseas military voters. Two states report separate election results for UOCAVA voters -- Hawaii and Maine.

Analysis

These statistics reveal that that Barack Obama's support varied across voting methods. While the directions of support are not consistent across states, patterns emerge that inform why we see battles over policies such as voter identification and early voting.

Early Voting: In-Person

A major legal battle in the 2012 election was the Obama campaign's attempts to overturn Republican state government laws in Florida and Ohio enacted since 2010 aimed at reducing the number of days in-person early voting is allowed. Under the radar -- perhaps because it happened in non-battleground states -- Republican-controlled local electoral boards in states like Indiana voted not to allow in-person early voting at satellite voting locations, even although it was offered in 2008.

Why? In the eight states where election results are reported for in-person early voters Obama fared better than Election Day voters by an average of 5.1 percentage points. New Mexico is the only state where this is not true. Furthermore, in the three states where all early votes are combined into one category, Obama fared better among early voters than Election Day voters by an average of 8.7 percentage points. Arizona is the only state where Obama fared worse. The Census Bureau will report later in 2013 results from a large-scale survey that may illuminate why Arizona and New Mexico are exceptions, but only by a little, with Obama doing better among Election Day voters by 0.5 and 1.0 percentage points, respectively.

These statistics also confirm the pre-election surveys that showed Obama winning the early vote, since where in-person early voting and mail voting are both allowed, voters tend to favor voting in-person over mail balloting by large numbers. The largest percentages of early voting shown in the table above are in states that allow in-person early voting.

Early Voting: Mail

Another action by Republican state governments is also stripped to its partisan ramifications when election results for mail ballots are compared to other voting methods. Republicans have enacted photo identification requirements for in-person voters -- either on Election Day or early -- but have exempted persons who vote by mail from these requirements.

In the nine states that report election results for mail early voters separately from other voting methods, Obama fared worse than Election Day voters by an average of 1.5 percentage points. Obama fared better among these voters only in Arkansas, Hawaii, and Maryland by 3.3, 2.2 and 8.4 percentage points, respectively. Maryland's pattern is perhaps explained in that the state combines mail ballots and provisional ballots into one category; and we will see that Obama did exceptionally well among provisional voters.

Early Voting: The Case of Iowa

Iowa's experience with early voting merits special attention. Obama won the state's early voters by 13 percentage points. Although the state does not distinguish between in-person and mail votes, many if not the majority of these votes were likely cast by mail, not in-person. Consider that the state reported 689,661 votes cast early as of November 6 (election officials continued to accept ballots postmarked on November 5 and received by November 13). By the first day of the in-person early voting period on September 27, 203,081 persons had already requested a mail ballot.

Iowa's Democratic Party has a tradition of mobilizing Democrats to vote early. Even in 2004, John Kerry won the Iowa's early and special ballots precincts with 58.5 percent of the vote, while losing the state by a little under ten thousand votes. In the aftermath, the Kerry campaign accused Iowa Democrats of spending too much effort on early voting mobilization and too little on Election Day.

Iowa's tale tells us is that the campaigns and voters can adapt to the electoral rules. This was evident in Florida, where the Obama campaign anticipated the reduced early voting hours would lead to long lines for in-person early voting and on Election Day. So, the campaign encouraged their supporters to vote by mail. While Florida does not report election results by method of voting, the partisan registration enjoyed by Republicans dwindled from about fourteen points in 2008 to four in 2012. (Florida had long voting lines, but they could have been worse without more people opting to vote by mail.)

Photo identification laws, by exempting mail voters, tend to relieve Republican voters of the identification burden. The irony is that mail balloting is the voting method known to have the highest incidence of vote fraud, which is often given as the reason for identification laws (keep in mind, even here it is extremely rare). Completely tongue-in-cheek, one may wonder why Republicans are encouraging more vote fraud by adopting photo identification laws that exempt mail voters, and thus encouraging more people to vote-by-mail. But to take the tongue out slightly, with Democratic campaign tactics changing to embrace mail balloting, conservative pundits like John Fund have begun questioning vote-by-mail, noting in his book Stealing Elections, that "Absentee voting used to be available only to those legitimately out of town or physically unable to vote in person...as the arguments against promiscuous absentee or early voting mount, it is time to rethink this trend." Expect more of this rhetoric aimed at mail balloting in the years to come.

Provisional Ballots

Provisional ballots were mandated for federal elections by the 2002 Help America Vote Act. They are often used when people have a problem at the polling place that prevents them from voting, such as a voter registration problem.

One of the ways that this can happen is when a voter tries to vote in the wrong precinct. In some places -- predominantly urban locales -- voting for multiple precincts may be held in the same polling location, such as a school gym. This can lead to a problem known as "right church, wrong pew" where a voter is directed by an election worker in these super-polling locations to the wrong precinct. The voter does not appear on the voting list and is handed a provisional ballot, which does not count in some states when a person votes in the wrong precinct. Not surprisingly, this was litigated in Ohio, with a federal court ruling that voters should not be disenfranchised for being directed by an election worker to the wrong precinct.

Why might policies be enacted to make provisional ballots less likely to be counted? In the four states that report election results separately for provisional ballots, Obama won 17.2 percentage points more vote share compared to the overall election results.

UOCAVA Voters

UOCAVA refers to the Uniformed and Overseas Civilian Absentee Voting Act, and is a general shorthand for overseas civilians or overseas and domestic military personnel who vote an absentee ballot. Federal laws like UOCAVA and the MOVE (Military and Overseas Voter Empowerment) Act require election officials to treat these voters differently than domestic civilian voters, for example, by sending ballots earlier with an option for the voter to receive their ballots electronically. Some states even allow UOCAVA voters to return ballots electronically! Most states add UOCAVA voters with their other mail ballots.

In the two states that report separate election results for UOCAVA voters -- Hawaii and Maine -- Obama performed an average of 10.5 percentage points better than the overall electorate.

This may come as a surprise since "UOCAVA" is usually considered a synonym for military voters, which are generally considered to favor Republicans. However, these statistics make sense if there are a greater number of overseas civilians than military personnel in these states. North Carolina provides evidence in support. Although the state does not report election results for UOCAVA voters, the party registration of military and overseas civilian voters is tracked. For military voters, 40.9 percent register as Republicans and 30.9 percent register as Democrats. Conversely, only 22.7 percent of overseas civilians register as Republicans and 43.3 percent register as Democrats. It may thus be that UOCAVA voters in Hawaii and Maine are more often overseas civilians than military personnel.

Summary

In the 2012 general election, Democrats tend to vote early in-person and by provisional ballots, while Republicans tend to favor mail balloting, with Election Day voters usually falling in-between.

Policymakers who wish to expand or curtail a method of voting assume that campaigns and voters will passively react to a new law -- say to reduce in-person early voting hours -- by not voting. Some of the states that buck these trends suggest otherwise. Voting behavior and campaign tactics can change to meet new election circumstances. Democratic campaign tactics in Florida and Iowa demonstrate that Democratic voters will vote by mail when encouraged to do so. Indeed, although not analyzed, Oregon and Washington hold their elections entirely by mail, and Obama won both states.

Working with election administrators, I know that rural election officials are the most stressed: usually a part-time county clerk runs elections on a shoestring budget. Instead of trying to play games with election laws by making it more difficult to vote -- spending literally tens of millions of dollars we don't have to implement photo identification laws and to defend restrictive voting laws in court -- Republicans would be better served by making rural voters' election experiences better, thereby increasing the turnout of their base, rather than giving a rallying cry to their opponents by trying to reduce Democratic turnout.

Democrats Leaving Mail Ballots on the Table

2012-11-03T14:48:05-04:00

Both figurative and literally, registered Democrats are leaving requested mail ballots on the table. Democrats are returning the mail ballots at a lower rate than Republicans, leading to wide disparities among the ballots that have been cast compared to the ballots that are still sitting on kitchen tables across the country.

Many states require mail ballots cast by domestic civilians to be returned to election officials by election day, although some -- notably Ohio -- allow mail ballots to be postmarked by the day before the election and will continue to accept them until Nov. 16.

Persons who requested a mail ballot but attempt to vote in-person on Election Day should bring their mail ballot with them to the polling place, if they still have it. If they do not, most states require these individuals to cast a provisional ballot to ensure that they do not vote twice, once by mail and once in-person. (This is the easiest form of vote fraud for election officials to detect, so don't even think of trying it.)

Here are some numbers that will give the Obama campaign heartburn in the key states of Iowa and Florida.

In Florida, 406,634 registered Democrats have not returned their mail ballots compared to 362,920 Republicans. In comparison, registered Democrats have returned 700,970 mail ballots compared to 781,043 Republicans. Thus, even though Republicans outnumber Democrats in returned mail ballots by a wide margin, more Democrats have yet to return their ballot.

In Iowa, 40,601 registered Democrats have not returned their mail ballots compared to 21,224 Republicans. Iowa is not reporting their ballot status in the same way as Florida, but we know that among all ballots cast -- both by mail and in-person -- registered Democrats outnumber Republicans 261,166 to 198,130. And while that may look like a comfortable margin, keep in mind that Iowa Republicans have historically voted in-person on Election Day in large numbers; John Kerry won the Iowa early vote in 2004, but lost the state.

It is not out of the question that a person with a mail ballot sitting on the counter may think of themselves as a voter even if they have not filled out the ballot. If so, when surveyed they may respond in the affirmative that they have voted even if they have not. The polling within these states may thus favor Romney slightly more than we may think.

More early vote statistics can be found here.

Follow me on Twitter @ElectProject

Early Voting Mesmerizing

2012-10-26T21:27:36-04:00

With more than 10 million people already casting a ballot, early voting is on pace likely to exceed the 41 million cast in 2008. Next week will be crunch time as millions of people across the country cast ballots.

Everyone wants to know who is winning. In my commentary below I explain why I conclude Obama has narrow leads in Iowa and Nevada, Romney has a narrow lead in North Carolina, and in the remainder the early vote is not providing a clear direction yet. There is still ample time for conditions to change, but it will be increasingly difficult to do so as more votes pour in.

There are instructive patterns other than the horse race. The patterns of early voting in Florida and Ohio suggest that Obama supporters are successfully overcoming limitations in early voting enacted by those states' Republican governments.

To date, early voters have been primarily high propensity voters; strong partisans who know who they will vote for. In the coming week, the early voting volume will pick up substantially, and we will see more of the lower to moderate propensity voters who are more persuadable enter the early electorate. In some of these close states, there is still time for the candidates to gain advantage though early voting as we reach the critical home stretch.

The data for this analysis can be found here, where I track the early vote in real time, as much as that is humanly possible. (The blog title is a wink to those who tell me they keep my website open during the day. Thank you to all the well-wishers!)

IOWA

Iowa continues to firm up as more votes come in, now with 423,586 voted. Registered Republicans have been making gains on Democrats during the early voting period in terms of party registration. But on Monday of this week Republicans peaked among absentee ballot requests, at 30.9 percent. By Thursday, their share of ballot requests dropped to 30.8 percent. The Democrats have been losing ground, too, starting the week at 44.9 percent and ending it at 44.0 percent. The gain is among those who do not registered with a party, something I discuss below.

Romney needs to make up more ground than he has among the early vote -- or have a great showing on Election Day -- in order to win Iowa. Obama won the state by 9.5 percentage points in 2008, when registered Democrats were 46.9 percent of early voters and Republicans were 28.9 percent. Closing the early voting gap by a couple of points is not enough, especially since the level of early voting will likely exceed 2008. The Iowa early vote thus confirms the polling showing an Obama lead, perhaps smaller than 2008.

NEVADA

Currently, registered Democrats lead Republicans by 45.8 percent to 37.3 percent statewide. A tidy comparison to 2008 is unfortunately impossible. In 2008, the state reported only in-person voting without party registration. So, to infer something about the state of play in Nevada, we can examine the geographic distribution of support across counties, with the Democratic stronghold of Clark -- home of Las Vegas -- figuring most prominently.

So far in 2012, 263,782 Nevadans had voted in-person. In 2008 -- for the same number of days prior to the election -- 199,412 Nevadans had voted early in-person. In 2012, Clark County comprised 68.8 percent of those who had voted and in 2008 they were 68.4 percent. So, the geographic composition of the Nevada early electorate look similar to 2008, when Obama won the state by 12.2 percentage points.

The Nevada early vote thus points in the same direction we've seen in the polls, which consistently show Obama leads.

NORTH CAROLINA

The North Carolina early vote looks good for Obama at first glance. Over 1.1 million North Carolinians have voted, and registered Democrats are 50.3 percent of these voters while Republicans are 30.8 percent.

Republicans have improved their standing. During the entire early period in 2008, Democrats were 51.4 percent of early voters and Republicans were 30.1 percent. And at the same point in time prior to the election in 2008, Democrats were 55.9 percent of early voters and Republicans were 27.2 percent.

It is possible that in North Carolina -- or any state -- these changes are simply a rearranging of furniture as people vote at different times. But, with the polls showing a narrow Romney lead on average in a state that Obama won by a razor-thin margin of 0.3 percentage points in 2008, the most reasonable conclusion is that the early voting confirms Romney has a small North Carolina lead.

Unlike Iowa for Romney, there is more opportunity for Obama to turn things around. The early voting volume is much higher than 2008, with 210,909 more people voting at the relative same point in time. So, his current lead, although smaller than 2008, may be more meaningful to the election outcome. Also, the state's "one stop" voting, where an unregistered voter can register and vote at an in-person early voting location, offers Obama the opportunity to covert more non-voters -- from a polling perspective -- to already voted.

COLORADO

As of Friday, 626,097 Coloradoans have voted. The party registration looks good for Romney with 36.1 percent registered Democrats voting to date and 39.0 percent Republicans. In comparison, in 2008, registered Democrats outnumbered Republicans 37.7 percent to 35.9 percent, when Obama won the state by 8.9 percentage points. The early vote thus points to a closer election than 2008, which corroborates the polling showing a close race.

Florida and Ohio: The vote suppression dog that didn't bark?

I hesitate to draw conclusions from the early voting in Florida and Ohio, other than to say that the election is close. These states are hopelessly confounded by the campaigns' mobilizing activities. But those activities inform us about the much discussed and litigated attempts by Republican state governments in these states to curtail early voting.

In Florida, 1.13 million mail ballots have been cast with 1.49 million ballots yet to be returned. The volume of early voting will thus likely exceed 2008, when 1.85 million mail ballots were cast.

Democrats trail Republicans 39.3 percent to 44.5 percent among mail ballots. At first blush, this is good news for Romney, but Florida Republicans usually win the mail ballots by much wider margin, by 12 percentage points or higher, and Democrats make up ground in in-person early voting.

The Obama campaign is encouraging their supporters to vote during the mail ballot period -- either by mail or in-person at election offices, what is called a "counter vote." We will have to see if the increased levels of voting offset the reduction in in-person early voting, starting on Saturday, Oct. 16, but will now be unavailable the following weekend. While mail balloting levels are high, keep in mind that 2.66 million Floridians voted early in-person in 2008, with more available days and hours.

We are not getting the best data out of Ohio. The state has what is known as "bottom up" election administration where the local counties are preeminent. There is no central reporting of mail ballots as can be found in many other states. So, we are left with deciphering a dated and stale Ohio Secretary of State report of statewide early vote activity and collecting local data where it is available (and these two sources conflict). In a state without party registration, the best that can be done is look at the voting patterns across the state.

Cuyahoga County is providing easily accessible data. Election officials report that despite a curtailment of weekend early voting by seven days so far -- yes, although the U.S. Supreme Court allowed weekend-before-the-election early voting to continue, there has been no weekend early voting to date -- the number of persons voting in-person in Cuyahoga so far is 23,954, compared to 20,334 for the same number of days prior to the election in 2008.

Although we do not have apples-to-apples comparison with mail ballots, Cuyahoga and the other urban counties are poised to exceed their 2008 early voting levels. But so, too, are the rural counties. Many voters thus appear to be taking advantage of mail ballot application that election officials sent to every registered voter in the state. Furthermore, both campaigns are fully engaged in a Ohio ground war encouraging their supporters to vote early by any means. Ohio continues to look to be a close state.

Attention Now Turns to Independents

Until now, the typical early voter has been a person who fits the profile of a likely voter: someone who follows politics closely and is certain who they are going to vote for. In the next week, more independent voters will enter the electorate. States that are predominantly voting by mail so far provide evidence in their partisan registration statistics.

In Florida, unaffiliated voters are 16.2 percent of persons who have voted, but are 21.0 percent of persons who have requested but not returned a ballot.

In Iowa, unaffiliated voters are 22.7 percent of persons who have voted, but are 25.1 percent of persons who have requested but not returned a ballot.

In Maine, unaffiliated voters are 26.1 percent of persons who have voted, but are 34.0 percent of persons who have requested but not returned a ballot.

In Maryland, unaffiliated voters are 12.0 percent of persons who have voted, but are 17.1 percent of persons who have requested but not returned a ballot.

In North Carolina, unaffiliated voters are 19.0 percent of persons who have voted, but are 24.6 percent of persons who have requested but not returned a ballot.

Cuyahoga County, Ohio permits us to look at this from another perspective in another report. Although Ohio does not have party registration, election officials record the last party primary an individual voted in. Among persons who have already voted, 21.9 percent are persons who have never voted in a primary, but are 39.7 percent of those who have requested but not returned a ballot.

Why are there more unaffiliated, lower propensity voters among those who have asked for a mail ballot, but have not cast one yet? People vote when they decide whom to vote for. What these statistics indicate is that there remains a pool of undecided voters, perhaps larger than many polls suggest.

The early voting ground war will shift in the coming week away from banking votes of people who would likely have voted anyway and towards voters who are less tied to the parties and are lower propensity voters. Keep in mind, too, that This will be a critical time for the Obama campaign, since many polls show Obama faring better among registered voters than among likely voters. If these voters are holding their ballots because they have yet to come to a decision on who to vote for, the Romney campaign cannot ignore the potential to sway some of these voters into voting for their candidate.

More early vote data available here
Follow me on Twitter @ElectProject

Early Voting Spinning

2012-10-19T15:34:55-04:00

Now with over 3 million people voting across the country, the campaigns are spinning the early vote. Here are my takeaways so far:

Iowa is firming up for Obama. The early vote confirms the polls showing an Obama lead, but by a narrower margin than 2008. Romney still has time to make up ground, but with over 300,000 people voting the clock is certainly ticking.

The early vote indicates that Florida and Ohio are close. We knew that already.

North Carolina started in-person early voting and the early vote numbers shifted dramatically towards registered Democrats. This is expected, as more registered Democrats voted early than Republicans in 2008 and 2004. A point of interest is that North Carolina allows unregistered voters to register and vote all in "one stop." This has the potential of shifting likely voter models since unregistered voters don't make the likely voter screen.

My detailed analysis follows, based on the raw data I report here.

IOWA

Iowa continues to be the state of most interest because of its high volume of early voting. Now over 300,000 people have voted, with registered Democrats outnumbering Republicans 49 percent to 31 percent among voters, and a lesser 46 percent to 31 percent among absentee ballot requests.

The Iowa early vote looks good for Obama. However, in 2008 Democrats had a party registration lead of 47 percent to 29 percent (as of Nov. 1, 2008). If the Romney campaign can convert all the absentee ballot requests, they will narrow the 2008 Democratic advantage ever so slightly, when the actual election results show that Obama won the Iowa early vote 60 percent to 38 percent and won the state 54 percent to 45 percent.

The Iowa early vote thus appears to confirm recent polling showing a narrower Obama lead over Romney than his victory over McCain in 2008. Romney is playing catch up in this critical battleground state.

FLORIDA

The campaigns are engaged in a back and forth spin battle about the Florida early vote. It goes something like this:

Romney campaign: Of the 620,187 mail ballots cast as of Thursday, registered Republicans lead 45.0 percent to 39.5 percent. Advantage Romney!

Obama campaign: Not so fast. Republicans vote by mail in Florida. In 2008, registered Republicans had a 12 percentage point lead among mail ballots (according to my data). Republicans should be doing much better. Advantage Obama!

Complicating matters is that the Obama campaign is encouraging their supporters to vote what is called a "counter vote." Counter votes are like in-person early voting, where a voter goes to an election office, requests a ballot and votes on the spot. It is not quite the same as Florida's in-person early voting soon to be allowed at satellite early voting locations as it is counted as a mail ballot.

With Democrats traditionally favoring in-person early voting and with Democrats lined up at the election office counter to do essentially the same, the mail ballot numbers are now confounded by changing Obama campaign tactics. So take a deep breath on Florida. We won't have a clearer picture of Florida until we are further into the in-person early voting period. At best, the early vote confirms the polling that the Florida election will be close.

OHIO

Ohio is where Republicans are spinning the most, and unfortunately some reporters are buying it. Take CNN, which headlines "Republicans point to early vote gains in Ohio."

The primary source for this story is the Romney campaign, which is promoting party registration statistics to back up their claims. Only at the bottom of the story, does CNN's Peter Hamby write that Ohio does not have party registration. "Party" in Ohio is a record of the last party primary an individual voted in. Worse, Peter Hamby reports this as a he-said-she-said story, noting that it is the Obama campaign who points out Ohio does not have party registration, something he could have easily discovered on his own.

Let's deconstruct another easily verifiable claim in this CNN report.

In Cuyahoga, Republicans only make up about 12 percent of registered voters. Ballot numbers through Tuesday of this week, however, signal that almost 22 percent of early voters in Cuyahoga are Republican.

Advantage Romney!

But not so fast, let's look a the actual numbers that Cuyahoga so helpfully posts online.

As of Thursday's report, there are 124,967 Cuyahoga registered voters who most recently participated in a Republican primary, or 13 percent of all registered voters. There are 17,133 such persons who have voted, or 21 percent of all voters. So far the story is mostly true; perhaps it is based on an earlier Cuyahoga report.

But what about the voters who last participated in a Democratic primary? They are 343,392 of all registered voters, or 37 percent. 49,720 of these folks have voted, or 60 percent. Comparably, a larger percentage of "Democrats" have voted early in Cuyahoga than Republicans, compared to their base registration statistics.

Advantage Obama!

Before we draw that conclusion, let's understand what is really going on here. 2.4 million Ohioans voted in the 2008 Democratic primary, compared to 0.5 million in the Republican primary. Over the course of four years, some of these people were purged from the voter rolls. In 2012, 1.1 million Ohioans voted in the Democratic primary, and 1.2 million voted in the Republican primary. I suspect that there were a good number of Democrats who crossed over and voted in the Republican primary just because it was the more interesting race from the presidential perspective. In Ohio, all of these folks are now labeled Republicans. "Party" is so hopelessly confounded in Ohio that it is next to meaningless to divine who is ahead.

The Cuyahoga numbers do reveal something about early voters. They are highly participatory people who tend to vote in primaries. There are 458,193, or 49 percent, Cuyahoga registered voters who have no record of voting in any primary. Only 15,835 have voted so far, or 19 percent. Let me put this another way, people who vote the earliest are people who just generally vote.

I thus take the same conclusion from the CNN story:

Few in either party question the Obama campaign's sophisticated ground game and most expect Democrats to bank more votes before Election Day. But Republicans have vastly improved their turnout effort in Ohio from the dog days of October 2008.

Just because Romney has a better ground game than McCain, that does not mean that Obama has not stepped up efforts, too. The Cuyahoga numbers show evidence that Obama's campaign is at least keeping pace with Republicans, if not outpacing them.

Looking across the Ohio counties, it appears that early voting is up everywhere across the state. Both campaigns are hard at work through the extended early voting period. Ohio is ground zero for this election, but we already knew that.

NORTH CAROLINA

North Carolina began in-person early voting on Thursday, and oh what a difference a day makes. In one day, over 150,872 people voted in-person, which is the Democrats preferred method of early voting in North Carolina. The party registration numbers were upended. As of Wednesday, registered Republicans outnumbered Democrats 52 percent to 27 percent and as of Thursday, Democrats outnumber Republicans 47 percent to 35 percent. But before we call North Carolina for Obama, registered Democrats had healthy early vote leads in 2008 and 2004.

North Carolina has an innovation unavailable elsewhere. In-person early voting is called "one stop" voting in the state because a person can register and vote all in one stop at an early voting polling location. Over 100,000 people took advantage of this in 2008. Unregistered voters don't even make registered voter poll screens, much less likely voter screens. It will be worthwhile to watch if one stop voting moves the North Carolina polling as early voting progresses.

More early vote stats are available here, updated as fast as I can.

You can follow updates on Twitter.

Early Voting Rising

2012-10-12T19:01:03-04:00

More than 800,000 people have already voted in the 2012 general election. A clearer picture of the potential 2012 electorate is emerging in some states where election officials are providing information on who has voted.

Early voting -- both mail and in-person -- is on pace to exceed 2008 levels, when about 30 percent of all votes nationally were cast prior to Election Day. The 2008 levels may be exceeded even further in states such as Iowa and Ohio, where early voting has been brisk. As a corollary, with no collapse in early voting, there is no indication so far that overall turnout, both early and Election Day, will be substantially lower than 2008.

In two states where party registration of early voters is available -- Maine and North Carolina -- slightly more registered Democrats have requested ballots as of a comparable date in 2008. Conversely, fewer registered Republicans have requested ballots. In Iowa, both registered Democrats and Republicans are voting above their 2008 levels, with a greater increase among Democrats.

Ohio presents a more complex picture. Early voting is up quite a bit over the 2008 level. This is likely because election officials sent absentee ballot applications to every registered voter. Ohio does not have party registration, so county-level data provides a better picture of the partisan implications. While levels are somewhat higher in rural counties, they are up in urban counties, too. In-person early voting is a preferred choice of urbanites, so the differences observed so far may narrow as in-person early voting continues. While the Obama campaign has been concerned about the limitation of in-person early voting during the weekend in Ohio, there is little indication so far that this has negatively influenced in-person voting. Still, the heaviest period of in-person voting in 2008 was the weekend prior to the election, so caution should be taken when projecting out from this lighter period of early in-person voting.

A more detailed state-by-state analysis follows.

IOWA

Iowa sticks out as a sore thumb in early voting. Already, more than 200,000 people have voted in Iowa, a much larger share of the potential 2012 electorate than any other state.

The number of registered Democrats who have voted is about twice that of registered Republicans, 53 percent to 28 percent, with the remainder unaffiliated. Bet let's not be so fast in calling Iowa for Obama. In 2008, Democrats outpaced Republicans 47 percent to 29 percent among early voters. Among the nearly 400,000 ballot requests in 2012, Democrats outnumber Republicans by a narrower margin: 48 percent to 30 percent. Republicans have been steadily making up ground among the votes cast and the ballot requests.

Romney needs more Republicans to vote early or needs Election Day to be vastly different, since Obama won the state by nine percentage points in 2008. An early vote electorate that looks like 2008 would therefore pose a difficult uphill climb for Romney. If these early vote numbers fail to firm up further for Romney, Iowa may be the first state where the campaign will have to make a tough strategic decision as to whether or not to continue expending resources.

MAINE and NORTH CAROLINA

Maine and North Carolina provide detailed individual records of registered voters who have requested a ballot and the disposition of the ballot. Using this information it is possible to make a comparison of the number of ballot requested in 2012 to the same point in time as in 2008 (adjusting for the fact that the election was held two days earlier in 2008).

In these states the number of registered Democrats requesting ballots has increased from 2008, while the number of registered Republicans decreased. In North Carolina, the number registered Democrats increased by 695 over 2008, while registered Republicans decreased by 2,521. In Maine, the registered Democrats increased by 427, while registered Republicans decreased by 1,922.

These are just small deviations, so there is ample time for Republicans to catch up. However, these statistics do not square with surveys that find that Democrats are less engaged than Republicans in this election. This is critical since engagement is often a factor in who is a likely voter.

OHIO

Republicans have crowed that the number of Ohio-registered Republicans voting early has increased from 2008. The problem with this assertion is that Ohio does not have party registration. "Party" in Ohio is a record of the last party primary a voter participated in. Naturally, with only a contested 2012 Republican presidential primary, the number of "registered Republicans" in the state increased. Because party statistics are so deeply confounded in Ohio, I prefer to examine patterns of early voting across counties.

A significant change in Ohio is that election officials mailed an absentee ballot application to every registered voter. This appears to have stimulated interest in mail balloting.

In Hamilton County, home of Cincinnati, it is possible to make an apples-to-apples comparison at the same point in time to the 2008 election by examining data provided by the county election officials. The number of ballot applications is currently at 84,999 compared to 60,005 in 2008. A 42 percent increase.

Compare Hamilton to a rural county like Defiance. So far, there are 3,021 ballot applications in 2012, compared with 1,935 applications in 2008. A 56 percent increase.

Where urban counties may make up ground in the coming weeks is among in-person early voting, which is more prevalent in urban areas. Of course, another much-reported Ohio change is the cancellation of weekend early voting hours. However, so far this has not resulted in a lower volume of early voting compared to 2008 in some of the larger counties. In Cuyahoga, the number of in-person early voters in 2008 was 8,400 while in 2012 the number is 10,440 -- despite no weekend or Columbus Day voting. Still, I would caution that the heaviest volume of early voting is yet to come the week before Election Day. The change in the early voting law may thus depress some votes, but perhaps with so many people voting by mail, the effect will be offset by increased mail balloting which will also reduce Election Day congestion at the polling places.

More state-by-state early vote statistics available here.
Follow me on Twitter: @ElectProject

Early Voting Status Report

2012-09-27T14:30:32-04:00

Early voting is now underway.

All states were required by federal law to deliver military and overseas ballots last Saturday and as of Thursday, Sept. 27, all but twenty-one states have reached their deadlines to start early voting for domestic civilians. Within some of these latter states -- notably Ohio -- some local election officials have already begun mailing ballots to domestic civilians who requested them in advance of their state's deadline.

I call any ballot cast prior to Election Day an early vote, be it by mail or in-person at a special polling location.

So, what do the numbers tell us so far? Well, not a lot. But that hasn't stopped people from drawing wild conclusions from seemingly-surprising numbers in the few states were statistics are available.

Is Obama crushing Romney in Iowa? Registered Democrats comprise 64 percent of the 185,747 requested ballots so far, compared to 13 percent of registered Republicans. While an Iowa Republican operative grimaces at these numbers, it is important to understand that Democrats usually do very well among Iowan early voters. In 2008, registered Democrats were 50 percent of Iowa's 481,179 early votes compared to 29 percent for registered Republicans. Day to day, the 2012 number is slowly declining as more Republicans request ballots than Democrats, so there is ample time for Republicans to catch up. One of the ways that they might do so is through a state law that allows a hundred people to petition for a special in-person early voting site, which evangelicals are taking advantage of to request Sunday early voting sites at their churches.

Is Romney winning North Carolina? Registered Republicans are 53 percent of the 78,801 requested mail ballots so far, compared to 27 percent for registered Democrats. So, if the election was stopped today, Romney likely would win. Unfortunately for Romney, only 13 percent of all North Carolina early votes were cast by mail in 2008, while 87 percent were cast at in-person polling locations. Democrats overwhelmingly voted early in-person in 2008 and 2004, so once that option becomes available in October, the 2012 numbers will likely dramatically shift in Obama's favor.

Are Ohio Republicans energized to vote early compared to 2012? Yes, but there is an important change from 2008 that likely accounts for this. Party registration is somewhat meaningless in Ohio since it only indicates which party primary a voter last participated in. More meaningful is the distribution of vote among counties. Looking at rural counties won by McCain in 2008, the number of mail ballot requests is up by a higher percentage than the urban counties won by McCain. It is noteworthy that mail ballot requests are higher in urban counties, too, and also had higher levels of early voting than rural counties in 2008.

The big, under-reported change from 2008 is that all Ohio registered voters were mailed an absentee ballot request form this year. People were thus prompted to request ballots, and we might expect this to have a greater effect in rural counties with lower volumes of early voting in 2008. We'll have to wait to see if those requesting ballots were people who would have otherwise voted if they hadn't requested the mail ballot.

If there is one takeaway that I see in the earliest of early vote numbers is that voters are engaged in this election. Surveys are showing similarly high levels of interest in the election, and at least some of the recent polling movement towards Obama and other Democratic candidates is from Democrats who have finally engaged now that the fall is upon us. There is no compelling evidence yet that there will be a dramatic decrease in turnout come November.

For early vote statistics, see here. Follow updates: @ElectProject

Absentee Ballot Applications Down by Nearly Half in NC

2012-09-09T14:46:08-04:00

We are beginning to get the first early voting statistics out of North Carolina, which started mailing absentee ballots on Friday, Sept. 7. Election officials report delivering 21,875 mail ballots as of Saturday, Sept. 8, and two ballots were accepted. Congratulations Daniel and Justin, you are the first two people to vote for president in the 2012 general election!

What do the number of ballot requests tell us so far?

It is difficult to make a true comparison to 2008 since the first day that election administrators mailed ballots in 2008 was on Monday, Sept. 15 and in 2012, it was earlier on Friday, Sept. 7. To make the most consistent comparison as possible, I generate statistics for the first date of mailing of ballots.

The number of absentee ballot applications is down by nearly half from 2008. In 2008, election officials had received 37,539 applications compared to 20,695 in 2012, or 45 percent fewer applications. The number of applications from registered Republicans is down more than Democrats, which are also down. The percentage of registered Republicans declined by 55 percent while the percentage of registered Democrats declined 35 percent. Thus registered Republicans composed 51 percent of the earliest absentee ballot applications in 2008 and 42 percent in 2012.

These numbers appear to confirm a report from Chapman University finding military absentee ballot applications are down from 2008. While news reports primarily discuss problems with the administration of elections, which are significant for overseas and military voters, they fail to consider that military personnel may have been particularly enthusiastic in 2008 to vote for veteran John McCain.

Still, this cannot be the whole story since the majority of requests are from civilians. There were only 3,949 military absentee ballot applications in 2008 and 2,127 in 2012. Whatever is going on is also affecting civilian voters. Pamela Mitchell, Acting Director of the Federal Voting Assistance Program, notes that ballot applications are on par with 2004, the last time there was an uncontested nomination for the incumbent president's party. Perhaps this is the reason, or something else is happening.

It is also worth noting that eventually registered Democrats far outpaced registered Republicans in 2008, especially once North Carolina started in-person early voting. Absentee ballots were only 8.6 percent of all early votes cast in 2008.

The lesson, while these statistics are interesting, caution should be taken when making simple comparisons at different points in time across different elections.

Follow early voting statistics here and updates on Twitter: @ElectProject

UPDATE 9/12: The number of ballot requests has continues to increase, with an updated 2,360 reported on Monday (revised from 1,890) and a preliminary 2,129 on Tuesday. The number of absentee ballot requests is now 28,912.

I previously discussed above an apparent decline in the number of absentee ballot requests in comparison to 2008. Then, at the start date of mail balloting there were 37,539 absentee ballot requests. Now, at the start of mail balloting last Friday, the number was 20,695. As I caution, the start date for requests is not entirely comparable since mail balloting began over a week earlier in 2012 than in 2008 -- but the election was also two days later into November. This Saturday marks the same number of days from the election as when mail balloting began in 2008. If the rate of requests continues on the current pace, it may be that the number of requests will be on par with 2008, including for those among the military, which now stands at 3,003, compared to 2,127 last Friday and 3,949 for the start of mail balloting in 2008.

Early Voting Has Started in NC

2012-09-07T11:39:34-04:00

And we're off!

The first mail ballots have been sent to North Carolina voters. As of Friday mid-morning, election officials have mailed 5,085 ballots to 17,724 voters who submitted requests.

North Carolina is the first state to start early voting by mail. Indiana and Kentucky will be next up on Monday, Sept. 17. In October, some states will open special polling places for persons who wish to cast an early vote in-person.

As the early voting season progresses, I track early voting here. A few states, like North Carolina, provide detailed information about early voters, including their party registration, age, gender and race.

I forecast that 35 percent of all votes will be cast prior to election day, up from 30 percent in 2008. This early in the election season, early voting statistics are merely a curiosity. But, as the season progresses, they will provide a clearer picture of the state of the election, particularly in key battleground states like Colorado, Florida, Nevada and North Carolina, that have high early voting rates.

UPDATE: Two people voted in North Carolina on Friday, marking the first ballots cast in the 2012 general election.

Early Voting in 2012: What to Expect

2012-08-13T16:19:26-04:00

Early voting in recent American elections has skyrocketed, reaching a record thirty percent of all votes cast in the 2008 presidential election, remarkably higher than the twenty percent cast in 2004. All indications are the record will be shattered again in 2012, with somewhere around thirty-five of the vote cast prior to Election Day.

States vary their early voting options. Some states like Indiana and Texas allow persons to vote early at special polling locations. Some like Oregon and Washington, and some local jurisdictions, run all-mail ballot elections. Some like California and Colorado allow persons to request that they vote by mail in all future elections. Some like Ohio allow persons to request a mail ballot for any reason. Then there are a handful of holdouts like Pennsylvania and Virginia have traditional absentee balloting laws that extend early voting only to those who provide a valid excuse. Complicating definitions is that some states like Florida and North Carolina allow both early voting at special polling locations and no-fault absentee balloting. And where mail balloting is the primary method of early voting, voters can vote in-person at an election administration office. (I recommend seeking up-to-date voting information from state and local election officials.)

All states used to have what might be considered traditional absentee voting laws. The laws have evolved since. California was the first to adopt no-fault absentee balloting in 1980. Florida, Tennessee and Texas first opened special early-voting locations in 1996. Oregon adopted all-mail elections by a 1998 voter initiative. When early voting is tabulated by states, the national upward trend in early voting is located clearly among early voting states, although there has been a slight rise in early voting in states with traditional absentee balloting. The upward national trajectory is thus a combination of more states adopting early voting alternatives and increasing use among voters. Washington perhaps best exemplifies the love voters have for early voting. So many people signed up to permanently receive ballots that election administrators decided to dispense with opening Election Day polling places that were costly to run and empty. Colorado is nearing a similar tipping point.

The campaigns have adjusted their strategies to the way people vote. Election administrators track the status of every registered voter -- whether they voted in-person early, and if they have a mail ballot in hand or if it has been returned. The campaigns scratch these voters off their target lists and refocus their efforts to those who have yet to vote. Once the election rolls around and you want the campaigns to stop contacting you -- vote early!

In 2008, the Obama campaign had an aggressive early voting strategy. They made early voting a part of their voter mobilization ground game and juiced early voting by other innovative means. Obama and Biden's travel schedule coincided with the opening of early voting in battleground states, rallies were held near early voting locations, and preferential seating was given to supporters sporting "I voted" stickers. For the cash-strapped McCain campaign in contrast, the candidates only belatedly began mentioning early voting in their campaign speeches the week before Election Day.

The Romney campaign is flush with money and will not repeat the McCain mistake of ceding the early vote to Obama. Already, early voting mobilization was a part of the campaign's primary election strategy. The Romney campaign's early vote forecasts in some key battleground states exceed the upward trend, and likely reflect that unlike four years ago both presidential campaigns will be aggressively mobilizing people to vote early.

State, Romney Forecast, 2008 Percentage
CO, 85.0%, 78.9%
NV, 75.0%, 66.9%
NM, 72.0%, 62.3%
NC, 70.0%, 60.6%
FL, 70.0%, 51.8%
OH, 45.0%, 25.2%
IA, 41.0%, 36.0%
MI, 30.0%, 20.4%
WI, 30.0%, 21.2%
VA, 20.0%, 13.5%
NH, 11.0%, 10.0%

Perhaps the most eye-opening Romney campaign forecasts are Florida and Ohio, two states that have recently been criticized -- and under a cloud of litigation -- for curtailing in-person early voting the weekend before the election. The campaign predicts a twenty percentage point increase in these states. A countervailing factor in Ohio is that the state will automatically send out absentee ballot request forms to all registered voters, where in 2010 only the most-populated Democratic leaning counties did so -- with a sizable mail balloting increase. Both states' forecasts also likely reflect the Romney campaign's commitment to early voting mobilization.

The Florida and Ohio policy changes purportedly narrowly target Democrats since conventional wisdom is the weekend prior to the election is when Democrats, particularly minorities, vote early as part of a church-led "pew to polls" mobilization effort. There is evidence supporting these claims.

However, the evidence is largely based on the 2008 election, which was an unusual election because the Obama campaign invested heavily in early voter mobilization and Democrats, particularly African-Americans, were excited to vote at the first opportunity for the first African-American general election presidential candidate. According to Census Bureau survey data, the normal pattern among early voters is that non-Hispanic Whites tend to make up a larger percentage of early voters that Election Day voters. Only in 2008 did minorities make a greater proportion of the early vote than the Election Day vote (77.0% to 76.5%). 2008 election surveys also confirm that more Democrats voted early than Republicans. The historical pattern reasserted itself in 2010. With the Romney's new commitment to early voter mobilization and with Republicans energized, the early vote will likely be more contested than in 2008.

Is Wisconsin Now a Purple State?

2012-06-06T12:38:25-04:00

Does Scott Walker's victory in the Wisconsin recall election mean that Wisconsin is now in play in the November presidential election?

Short answer is no, for two reasons.

The first reason is that the exit polls show that if Obama was on the ballot, he would have won by a comfortable 51 percent to 44 percent margin.

What the exit polls do reveal for the Romney campaign is that 18 percent of Obama supporters voted for Walker. These persons represent potential persuadable voters that the Romney campaign will target if they decide to invest heavily in Wisconsin. What we do not know from the exit polls is why these voters would have split their vote between Obama and Walker. That mystery would have to be unraveled by additional polling to see if there is an opportunity to change these folks' votes. It very well could be that these voters' attitudes cannot be changed if, for example, they simply dislike recall elections. Still, even as Romney targets these voters, so too will Obama, so it will not be easy to changes these voters' support for Obama.

The second reason is that the recall electorate is different than what will likely be the November general election.

The total number of votes cast in the recall election was about 2.5 million, while about 3.0 million voters were present in 2008. Among those that chose to sit out the recall election are strong Democratic constituencies. In comparison to the 2008 exit polls, persons age 18-29 composed 16 percent of the recall electorate, while in 2008 they were 22 percent. Minorities constituted 9 percent of the recall electorate, and 11 percent of the 2008 electorate.

The presence of these Obama supporters in November will further tilt the balance evident in the exit polls in an Obama direction.

Additional polling would need to be done to determine why these key Democratic groups sat out the recall election, but it comes as no surprise. Like clockwork, young people and minorities vote at lower rates in state and local elections.

Certainly, these state and local elections affect all peoples' lives more directly, but for various reasons these groups do not perceive these elections as being as important as the presidential election. When the hand on the clock ticks to the general election, turnout among these groups will likely rise.

Of course, all of this is a snapshot of the Wisconsin recall electorate as of June 5, 2012. There is still a long way to go in the election. But, if Wisconsin is in play in November, it will likely be because the national mood has shifted towards Romney and there will be major problems for Obama in other battleground states. Wisconsin will perhaps be the least of Obama's worries if that comes to pass.