Why we need national polls even when it's swing states that decide the election

http://www.guardian.co.uk/commentisfree/2012/oct/03/national-polls-swing-state-election

Version 0 of 1.

I've been asked what the point of national polls is. The question was posed because, with the electoral college system, this election is going to be decided in the swing states. Indeed, most voters have pretty much no chance at casting the decisive vote.

But here are five reasons why I believe national polls are and should remain an important part of any election coverage.

1). National polls can tell us what the real issues are

You know with the back-and-forth of the horse race coverage, it's sometimes hard to forget that elections are ultimately about issues. After the election occurs, politicians are going to have to go to Washington and try to solve the problems facing this country.

Politicians should know what Americans' positions on healthcare, medicare, economic stimulus, the environment, etc are. We often elect individuals for many different reasons than they think. Few would think that we elected President Obama in 2008 because the majority agreed with all of his position stances. Likewise, it's difficult to believe that Republicans took the House in 2010 because Americans were upset with President Obama's supreme court picks.

Without polls on issues with the national electorate, politicians could make all sorts of claims and there couldn't be any factual repudiation. Now, politicians may still make inaccurate statements with regard to what elections mean, but at least we have the numbers to know they are overstepping their mandate.

2). People in the non-swing states matter, too

We may have an electoral college, but that doesn't mean voters in non-swing states don't matter. The only reason safe blue and red states aren't "important" is because their outcomes are decided. Another way to look at it, however, is that these non-swing states are actually the vast majority of the country. Yes, Ohio is important in this election, but politicians shouldn't kowtow to their interests only. Presidents are the leader of all the United States, not just the chosen 10 or so swing states.

3). The swing states may not be the swing states

It's silly to think that the states that were swing states last time around will be the same this time. Especially early in the campaign, national polls probably tell a better story of where the electorate is. Had we just relied on the old idea that Missouri was the ultimate bellwether, as it once was, we would have thought that John McCain was in good position in 2008. This year, we believe Ohio is the ultimate swing state, but perhaps it won't be, given President Obama's solid lead there.

4). National polls are usually more accurate than state surveys

Does anyone really believe that Mitt Romney is down by anywhere close to 10 points in Florida, as Quinnipiac found? There are so many national polls that outliers don't move polling averages as much, making them more accurate and reliable.

The national polls are also often conducted by better firms than the statewide polling is. Anyone ever heard of Gravis Marketing, before 2012? I hadn't. I don't mean to pick on Gravis (although they have had a huge Republican house effect), but they are typical of many state pollsters. They tend to have small sampling periods, don't call cellphones, and just don't have a track record of any kind. I know when I read a national poll from ABC/Washington Post or NBC/Wall Street Journal that I'm getting quality data from tested sources.

That's not to say that we should dismiss state polling. In fact, I spend most of my days looking at state data. It does seem that President Obama may be doing better in the swing states than the national numbers indicate. Still, national polling can provide a nice check on whether state numbers are out-of-whack.

5). The chance that the winner of the popular vote differs from the state vote is just not large

It should be recognized that a national poll is pretty representative of the swing states at large. In any given election, there's only a 3-5% chance that the electoral college winner will differ from the national vote winner. That chance is pretty much zero if the national polls continue to show Obama leading Romney by 3-4 percentage points. Even if the national margin were to shrink to 1 percentage point, the chance that the electoral college winner would be different is only 25%.

Even if we did get inside a 1-point national vote, it's not particularly clear that swing state polls would provide much clarity. We would know what states would provide the winning margin. Yet, polling margin of errors are large enough that once we get this close, we really wouldn't have that much of a better idea about who would win. Remember in 2000 when many thought Al Gore would lose the popular vote and win the electoral college, but the opposite happened?

Conclusion

National polls are an important complement to state data. They don't tell the whole story, but they give us a lot of important information regarding both where the electorate is and why it is where it is.