Questions to Ask About Early Voting

Published on

Early voting and vote by mail are about to end and the reports so far seem to indicate that these early and mail voters will exceed the numbers from 2010. This has lead a number of analysts to try to crunch the numbers and figure out which candidate has the advantage from early voting, something that can’t be done unless you know the voting history of each voter. The campaigns and the state parties have this data, pundits typically do not, so the only thing you can do is put together a list of good questions to ask these campaigns/parties and hope it will give you useful information. Let me explain.

I vote in every election, I am a habitual voter. The sun will rise in the east, the Cubs will not win the World Series and I will vote. No campaign should spend any money or effort trying to get me out to vote, that would be like spending money to remind me to breathe. Last week I voted early, as I have in every election since early voting became legal. I didn’t vote early because of any candidate’s GOTV operation, I voted early because that’s what I do.

If you want to measure which campaign’s GOTV operation is performing the best you need to be able to separate out which voters are habitual voters vs. which voters are infrequent voters and early/mail voting GOTV targets. The campaigns have each defined a universe of infrequent voters that are very likely to support their candidate if they make it to the polls and they’re working like crazy to GOTV these voters with either early vote or vote by mail. In order to correctly measure which side is performing best for early/mail voting you need to isolate and measure these efforts.

The campaigns divide voters into three categories 1) will vote for my candidate, 2) will vote for the opponent, 3) don’t know. The size of the first two groups is much bigger than you would probably expect. In the old days they would determine their support group (run universe) from the list of people who’ve pledged support (their plusses) along with other potentially telling factors like the partisan strength of their precinct or their partisan primary voting history and other significant demographic characteristics. These days the national parties are paying to develop modeling scores for their key races which take all of the available information about each voter and uses it to develop a number that measures the likelihood that the voter will support the candidate relative to all the other voters. They can use this info to determine which voters are very likely to support one candidate or the other, although it’s a little bit harder to predict what happens with the 3rd party candidates.

Once you understand all of these variables it’s clear to see why you can’t just look at raw early/mail vote totals and draw conclusions from them. Here are the questions someone will need to ask of the campaigns:

  1. Of the total early/mail voters how many are you certain are voting for your candidate, how many for your opponent?
  2. What percentage of the total early/mail vote universe are habitual voters? How many are you certain are voting for your candidate, how many for your opponent?
  3. What percentage of the total early/mail vote universe are the infrequent voters targeted by early/mail vote GOTV efforts? How many are you certain are voting for your candidate, how many for your opponent?

Also, if you can get them to show their work that would be even better.

What Early Vote Numbers Do (and Don’t) Mean

Published on

Early voting in Illinois began just a few days ago and in some other states around the country it has been going on for as many as a few weeks now. Most states have some form of early voting with some combination of early vote, vote by mail and/or no-fault absentee voting. The county clerks in each jurisdiction are busy taking record of which voters have already completed a ballot, the clerks must do this to ensure that the same voter doesn’t show up to his/her polling place on election day and cast a 2nd ballot, and so we are starting to see some numbers reported both in polling data and in actual numbers voted in media reports.

For example this poll in Time Magazine shows that among early voters in Ohio 60% are supporting President Obama while 30% are supporting Governor Romney, and according to this data from the Iowa Secretary of State 542,000 people have requested an early ballot and 376,000 have already returned their ballot. Here is a good site that is trying to aggregate as many of the available statistics and analysis on early voting.

While any candidate would strongly prefer to have any and all of their supporters vote as early as possible to ensure their vote is heard and counted it’s important to understand that some early votes are more interesting and significant than others. Consider two types of early voters, one we’ll call Habitual Voters. These are voters who were 100% certain to vote and couldn’t wait to cast their vote. The other type we will call Sporadic Voters. These are voters who were likely or very likely to support your candidate but for whatever reason they don’t always make it to the polls. Sometimes they forget, maybe they work multiple jobs at strange hours or perhaps they travel a lot. For whatever reason some voters are just sporadic voters.

Sporadic Voters are the voters that most interest campaigns during early vote. A well organized campaign should be taking advantage of the convenience offered by early vote to help their likely supporters take advantage of this convenience and make sure they cast their ballot. A good GOTV (get out the vote) operation should help a campaign by increasing their vote total among supporters by encouraging sporadic voters to vote.

However the early voting data reported by polling firms, county clerks and state boards of elections can’t distinguish between Sporadic Voters and Habitual Voters. That sort of data just isn’t easily available publicly.

But the campaigns know.

In Illinois, for example, each county clerk (or municipal election authority – there are 8 of them) are required to send a list of voters who have early voted to the State Board of Elections within two days. That data is then aggregated by the State Board and made available to the campaigns who will update their voter files with that data. The campaign voterfiles are sophisticated enough that they can be easily queried to see which voters are Sporadic vs. Habitual Voters and they can keep score on which types of voters have voted and which candidate they are supporting (at this point in a presidential campaign the campaigns know who each voter is voting for with a pretty high degree of accuracy).

Both the Republicans and the Democrats have released memos taking credit for their progress on early voting. This memo from the RNC doesn’t mention any data relative to sporadic voters, but here’s their take:

In the battleground states with available data, Republican AB/EV activity is strong. In addition to raw Republican versus Democrat turnout numbers, there are two key metrics by which we can measure this. First, we can calculate the party’s share of AB/EV activity as compared to the party’s share of voter registration. The data show the percentage of AB/EV activity from Republicans is greater than the percentage of registered voters which are Republican, indicating higher turnout rates among registered Republicans than among registered Democrats. For example, Republicans are outperforming our share of voter registration in absentee requests and early votes by 5.6 points in Florida, 8.73 points in Ohio, and nearly 12 points in Pennsylvania.

Second, we can measure the party’s share of AB/EV activity as compared to its share in 2008. In most cases, the data show Republicans making up a larger share of early voters this year than they did four years ago. Democrats make up a smaller share, giving Republicans an important advantage. Across the eight states, Democrats are underperforming their share of 2008 AB/EV votes cast by a net 5.85 percentage points, while Republicans are over-performing their share by 2.13 points, yielding a net swing of +7.98 percentage points for Republicans.

I’d prefer a more data driven analysis, but that’s their take.

On the other side of the coin is this memo from the Obama campaign field director. I wish they’d make their daily spreadsheets available by state and give us more data but they do have some analysis on Sporadic Voters (defining sporadic voters as voters who early voted in 2012 but didn’t vote at all in 2010):

Non-midtermvoters: Across nine battleground states, Democrats have a 19.7 point advantage in ballots cast among non-midterm voters. More than half (51.5 percent) of non-midterm voters who have voted already are Democrats, while fewer than a third (just 31.8 percent) are Republicans.

  • For example, in North Carolina, 51.5 percent of those who have already voted are Democrats, compared with just 25.1 percent who are Republicans. That’s a major advantage. And among these non-midterm voters who have voted in North Carolina so far, 87 percent of them are youth (under 35), African-American, Latino, or new registrants (registered after the 2008 election).

This is more useful and interesting data but it’s no guarantee of success. Hopefully reporters who are working on early voting stories are asking the campaigns to provide data (not just spin) on the performance and partisanship of sporadic voters who have voted early.

Now having said all this it’s important to keep in mind that the data is not perfect and it’s the larger counties with larger staffs that provide the most reliable data. For example, in Illinois every county clerk is supposed to send their early voting data to the State Board of Elections but I’m certain that many of the smaller counties just don’t have the capacity to meet this obligation daily. So the most reliable data is generally coming from the areas with large populations and if you think population and/or population density correlate to specific candidate performance then you should take that into account when trying to understand the data.