Margin of Σrror

Margin of Σrror -

What early polling can – and can’t – tell us about 2013′s key US elections | Harry J Enten

In races like the New York mayoral election, most candidates’ lack of name recognition makes early polling highly unreliable

Polls taken early in the election season are a most interesting phenomenon. The media loves them, while some think they are a waste of time.

The record shows that early surveys can tell us a lot about where a race will end up in the case of midterm House and even Senate elections. But in the 2013 elections, early polling is, at best, a very rough guide, and at worst, can tell us absolutely nothing.

Let’s take a look at the history of polls from early to late in the campaign for past contests in the races being held this year. The Massachusetts special election upcoming later this year may or may not feature Republican Scott Brown. The polls at this point have him ahead of almost all his opponents, except for, perhaps, Governor Deval Patrick.

The issue here is that none of his opponents is really well-known, which is why I wouldn’t dismiss any of them at this point. The plurality of voters have not formed an opinion on possible Democratic candidates like Mike Capuano, Ed Markey, or Stephen Lynch. The same holds true for possible Republican candidate Bill Weld.

As for the history, at this point in the 2010 Massachusetts special Senate election, Scott Brown was down 30pt to eventual Democratic nominee Martha Coakley. In fact, he was still down by 30pt with only two months to go. Yet he went on to win by 5pt.

Some might say 2010 was an exception, but just look at another special election, a year later, for West Virginia governor. The only pre-primary poll in that race had eventual Democratic candidate Earl Ray Tomblin ahead of Republican candidate Bill Maloney by 33pt. Tomblin would only retain the governor’s mansion for the Democrats by a little less than 3pt.

The bottom line is special elections can turn on a dime, especially when name recognition is low.

Another place where name recognition may be affecting early polling is in New Jersey. I already discussed how current polls that show Chris Christie well ahead of his Democratic opponents are likely inflated by a post-Sandy bounce. One would expect that his approval rating will eventually fall back to a still impressive mid 50s level.

That’s exactly where polls put another incumbent Republican Christine Whitman in 1997. Whitman led then relatively unknown state senator Jim McGreevy by about 15pt in early data. Whitman’s approval rating and polling lead held into the final month; but then, McGreevy rapidly closed the gap and only lost by 1pt.

Pre-Sandy polls put Christie ahead of declared Democrat Barbara Buono by a similar 16pt. Christie actually led by only 6pt over possible Democratic candidate Richard Codey. Christie is, no doubt, the favorite, but New Jersey is a Democratic state – and the last Republican with a good lead nearly lost.

Right next door to New Jersey is the New York City mayoral race. Early surveys have the probable Republican candidates down by 50pt to a generic Democrat in this Democratic bastion. Maybe that will hold, but I’m betting it won’t.

You don’t really have to look far to understand that polling in New York City mayoral elections is about as reliable as the subway after midnight. They were well off in 2009 and that was not an anomaly.

Polls had Democrat David Dinkins ahead of Republican Rudy Giuliani by 20pt two months out in 1989, and by 14pt in the closing weeks. Dinkins won by 2pt.

Republican Mike Bloomberg overcame an early 40pt deficit because of the 9/11 attacks to beat Democrat Mark Green in 2001. Democrat Freddy Ferrer was ahead of Bloomberg by 8pt in March of 2005. By November of that year, surveys had Bloomberg leading by 30-40py. Bloomberg took the race by 19pt.

And, of course, we don’t even know who the Democratic nominee will be for mayor this year.

Finally, in the great Commonwealth of Virginia, early gubernatorial surveys have Democrat Terry McAuliffe barely ahead of Republican Ken Cuccinelli. As in Massachusetts, however, over 45% of voters have no opinion of either candidate. There is also talk that Republican Lt Governor Bill Bolling may run as an independent, which could really throw this race for a loop.

Early surveys in the last two Virginia gubernatorial elections were off the eventual margin by 10-15pt. In 2005, Democrat Tim Kaine trailed Republican Jerry Kilgore by 5-10pt through the summer of 2005. It was only in the final months that he pulled ahead and won by 6pt. In 2009, Republican Bob McDonnell was ahead of Democrat Creigh Deeds by mid single digits through the early fall. He ended up winning by 17pt.

Thus, in the four marquee races for 2013, early polls should be taken with a big grain of salt. In all four cases, we have examples in the past 15 years of early polls being anything from 15pt to upwards of 30pt off.

That’s not to say the early polls won’t be right this time. It just means that if they are, there will definitely be some luck involved. © 2013 Guardian News and Media Limited or its affiliated companies. All rights reserved. | Use of this content is subject to our Terms & Conditions | More Feeds