Analyzing the Undecideds

With a week left in the campaign, all eyes are on the “undecideds” – the 7-10% of voters who say they as yet do not know for whom they will vote. Keep in mind that if current polling averages are accurate, McCain would need to win a significant chunk of these undecideds to make this race competitive, assuming that he can’t peel off existing Obama supporters.  So, how should we allocate these undecideds?  Keep in mind that as late as a day before the 2004 presidential election, fully 3% of people remained undecided.  And pollsters made different decisions on how to allocate those undecideds when projecting the final vote. For example, Gallup indicated that in previous years undecideds almost always break for the challenger. So in 2004 they allocated 2% to Kerry, 1% to Nader, and none to Bush.  Obviously that rule of thumb doesn’t apply this year, since there is no incumbent running.  Assuming that at least 3% remain undecided in the remaining days, how are they likely to break on Election Day?

It would help if we knew who the undecideds are. Fortunately, Pew has provided some demographic information in their latest survey ( Subject to all the usual caveats about polling uncertainty, what do the data show?

Based on demographics , the undecideds look more like potential McCain voters than Obama supporters.  Not surprisingly, the undecideds show a generally lower level of interest in the election than do Obama or McCain supporters. More than half (51%) don’t call themselves either Democrats or Republicans, compared to roughly 28% of voters overall who describe themselves as independents.  And fewer than half (48%) report having voted in the primaries.  So these are not strong partisans.   But here are the key statistics in my view:  the undecideds are “less educated, less affluent, and somewhat more likely to be female than the average voter. Nearly half of undecided voters (48%) say they attend religious services at least weekly, which is same as the proportion of McCain supporters. Fewer Obama supporters (31%) say they attend religious services at least once a week.”  Yep, it’s the bitter, bible-thumpers’ vote – one that I thought was potentially susceptible to the Sarah Palin pitch.  My guess is that a greater number of them will break for McCain than for Obama in the last week.   But it will require almost all of them to go for McCain if he is to have reasonable chance of catching Obama – a very tall order.

A word of caution is in order: because of the small number of independents in the Pew sample, the margin of error – 5.5% – is quite high for this subgroup. So we need to be careful in drawing conclusions about undecideds based on this sample.

How many undecideds will break for McCain, and will it be enough to overcome the 6-8% lead Obama appears to hold right now, based on the average of daily tracking polls?   The answer depends in part on turnout among other groups. I’ll address that in the next post.


  1. Professor Dickinson,

    Whether overtly expressed or not, we knew this was the profile of a significant fraction of the undecideds, for a long while. Having canvassed them in New Hampshire, I can offer first hand substantiation of those descriptors.

    My concern is not about the numbers – it’s a philosophical one, which I had raised in the alumni course you taught in the fall. Given that democracy works best with an informed electorate, how can we expect the will of the people to be served when elections repeatedly degenerate to the whim of the least capable element of the populace – one which is uneducated, ignorant and fundamentalist ?

    I comment somewhat in frustration but I do realize that there are always ways to constructively resolve this problem. The candidate’s campaigns and the frames they use are supposed to educate the undecideds and help them make up their minds. If this is to work, then the % of undecideds should decline monotonically over time since the campaigns began, till by the last week of the elections, you should effectively see very few undecideds. But you don’t. It’s as if the campaigns cancel each other out leaving the undecideds in a fog How can we help these people ? Or more appropriately, what should the characteristics of a system that allows for a robust reflection of an informed electorate be ?

  2. I know you’re skeptical about 538 because the creators admit their political leanings, but I think this analysis on the undecideds is pretty data-driven. Thoughts?

    And in the opposite of data-driven, here’s a great anecdote to suggest one counter-intuitive rationale for undecideds breaking Obama’s way: wanting to be on the “correct” side of history.

Leave a Reply

Your email address will not be published. Required fields are marked *