The Monkey Cage

Hi. I work at a political polling firm, as per my handle, and I can answer your question. The best answer is "Probably". Campaign pollsters will usually try to take this into account, although not always, depending on the number of early or absentee votes they expect, and the campaigns are certainly aware of it. Whether the public polls are taking early voting into account, I'm less sure of--I certainly haven't seen it mentioned in public polls, but I haven't been watching too closely for it. As noted by Mark Blumenthal at pollster.com, the amount of pollster transparency is limited at best.

Some of the polls (I think SurveyUSA's is one) have indicated that they included people who had already voted early or absentee and reported their vote percentages separately. But I'm not sure if this is universally the case.

In Oregon, where early voting has been going on for a number of years, there are two ways pollster take this into account, depending on how the sample was drawn for the survey:

(1) If the survey is random digit dial (RDD), the likely voter screen question ("how likely are you to vote in the election on feb. 5") is altered to include something like "or have you already voted". Those responding "already voted" are then automatically screened in.

(2) If the survey uses a voter sample file, it is possible to "match back" the respondents to the list of voters reported as turning in a ballot (the Oregon secretary of state does this, for instance). This can be combined with (1), which can yield a cross-tab verifying voter responses (or bad data).

In most current surveys for Florida, this level of details (in particular, the questions asked) are not released.

Does the Oregon SOS keep a running tab of people who have turned in a ballot and make that available?

If they do all I have to say is wow.

Eric

Eric,

Yes. They keep a running tally not only of the number of votes turned in starting the first day they are received, but they make available under certain conditions the voter IDs of those who have voted each day. This may not be completely public, but it was not difficult to obtain. This is the same as other states' SOSs providing who voted data that is used in voter files to note who voted in which election; the Oregon data is just more exciting.

Anon is right. Pollsters can gain access to the early voting lists, although it is easy in some states (e.g. Florida's early in person lists) but harder in others (e.g. Florida's absentee ballot lists).

In general, in most states, the lists of ballots already cast are kept at the county level. In general, in most states, you have to pay for a data pull. And in general, in most states, this will contain just a voter ID number and whether the ballot has been cast or not. You need to voter reg file to do match to names--and the voter reg files seldom have phone numbers.

(Take a breath here).

BUT if a pollster is out doing a RDD pre-election poll, almost all of them now include a "did you already vote" filter.

===

The reader doesn't raise the more interesting question, however: are CANDIDATES doing this polling and therefore have some sense, before election day, of how many votes they've already collected.