Don't be misled into thinking "early voting" reports necessarily, or even very closely, correspond to actual votes cast!

The mainstream media is trying to make you think it knows exactly how the early voting is going, but it's assuming that registrations equal actual voting decisions, which is very misleading (as I argued in a Thursday guest-post at HughHewitt.com).

---------------------------------

[Copied here for archival purposes on November 5, 2008, from the post linked above at HughHewitt.com.]

(Guest Post by Bill Dyer a/k/a Beldar)

I was stunned to find a post on ABC News' Political Radar blog entitled Democrats Lead in Early Battleground Voting. I checked my calendar: Nope, it's still October. How, then, can ABC News report that "the early numbers in key battleground states show Democrats are outperforming Republicans"?

The answer is: This kind of report is grossly and dangerously misleading. And whether intended or not, it may have the practical effect of misleading people into thinking that their votes don't matter when in fact those votes may become crucial. This is dangerous misinformation that you should help combat, friends and neighbors.

What you're reading in these reports isn't based on political opinion polling — not even "exit polling." Instead, it's based on guesswork — specifically, a guess that in those states which record voter affiliation (typically either from votes cast in primaries or affirmative acts of registration by voters before the primaries), the top-of-the-ticket general election votes, as actually cast, will match up precisely with those registrations when the actual votes are counted (starting on November 4th after the polls close).

But in Texas, for instance, where early voting started this week, I am absolutely certain that many tens of thousands of dedicated Republicans crossed over in primary season — whether as part of Rush Limbaugh's "Operation Chaos" or just out of sheer cussedness — to vote for Hillary Clinton and against Barack Obama in the Democratic Primary. They have absolutely no intention of voting for either in the general election. But if they vote early, then based on their having voted in the Democratic Primary, they'll be counted in "early voting" news reports like this one as "Democrats" and they'll be assumed — wrongly — to have voted for Barack Obama.

Besides such "strategic cross-over voters," this reporting also will misreport "PUMA" votes from real Democrats who nevertheless choose not to vote for Obama. And it will misreport genuinely undecided voters who nevertheless voted in the Democratic primary, perhaps because they wanted to have a voice in down-ballot races, and ended up breaking their indecision by the time of the general election in McCain-Palin's favor.

Some states have continuing registrations that persist until the voter takes some affirmative action to change it. But in no state does your voter registration prevent you from voting for the opposite party in the general election! Some folks may still be registered as Democrats even though they haven't voted for one since Jimmy Carter in 1980, and they may have actually cast their early vote this time for the GOP ticket again, but they'll show up in these reports as yet another Obama vote.

And some people do actually change their minds to vary their actual votes from their registration status at the last minute. With no incumbent running and one candidate in this race being such a comparatively unknown quantity, there are substantial reasons to suspect that may be even more true in this election season than in most.

Moreover, there are all sorts of well-demonstrated reasons why early and absentee voters' actual votes, when they're actually counted starting on November 4th, may vary substantially from the trends set by those who vote on Election Day itself. But these numbers are even more unreliable as a basis for projecting final-vote results than even an accurate count of the early votes would be, if somehow the actual votes could be counted before Election Day (which they can't be). So these reports of early voting have at least a two-fold inaccuracy built in.

And this isn't just a failure to get the totals right, it's an affirmative mis-mark, where one voter being wrongly presumed to have voted in conformity with his/her registration  will show up in the totals both as one too many votes for one slate and one too few for the other. In what was assumed to have been a 100-to-100 tie, in other words, the actual count will turn out to be 101-to-99 (and a two-vote margin of victory) when the mistaken assumption is backed out and the actual vote is properly counted.

USA Today's comparable article at least contains this explanation, buried several paragraphs down:

Election records in many states show whether voters are affiliated with a political party or are independent. States that must comply with the federal Voting Rights Act, such as Georgia and North Carolina, also show racial breakdowns. No votes will be counted until Nov. 4.

But many reports don't have even that explanation. And a casual reader might well be misled into thinking that these are hard numbers being reported, instead of guesswork.

Please, please — whatever your preferences in this election! — don't be misled into thinking that you're seeing anything but guesses yet when the press start quoting these numbers. Only one party (and its adoring press) is pushing the "it's over, don't bother" meme this year, but don't fall for that tripe regardless of your own voting intentions. If you make the mistaken assumption from these sorts of reports that your vote doesn't count, then you may regret that for the next four years — as voters of both parties in the Florida panhandle did when they dropped out of election lines in 2000 after the networks had prematurely, and wrongly, "called" Florida for Gore after the polls closed in the part of Florida in the Eastern time zone. Treat your vote as if it could decide the election.

— Beldar