On Friday we talked about how undecided voters’ (“Undecideds”) opinions on ten major issues have shifted over the course of the last three months. Today we’re going to focus on what we know about Undecideds’ demographics and personal attributes. Do Undecideds from September look the same as they did in August or July?

(As a reminder, all numbers presented below were weighted by US Census figures for gender, age, race, and region from the 2008 voting population. All findings presented here are statistically significant, with an adjusted p-value less than 0.01. These numbers are derived from samples of 2,361 Undecideds in July, 3,181 in August, and 5,507 in September. The increasing sample sizes are due to an increase in our overall respondent base, NOT an increase in the proportion of voters who say they are “undecided.”)

The most obvious thing to look for when analyzing the collective profile of Undecideds is whether or not they more closely match the profile of an Obama supporter or a Romney supporter. For example, August Undecideds appeared to look more like Obama supporters than Romney supporters. In September, Undecideds seemed to resemble Romney supporters a little more than they had in the past. This might suggest that a larger number of previously-Undecideds moved firmly into the Obama camp.

One thing we know about this election as a whole is that the younger a voter is, the more likely they are to vote for President Obama, and vice versa. So when we analyze Undecideds’ personal characteristics and find things that amount to an age proxy, we can say with some confidence which candidate they’re more likely to support come November 6.

In August, for categories that could be considered age proxies, Undecideds looked more like Obama supporters than they did in July. August Undecideds appeared to be more transient than July Undecideds. 60% of August Undecideds have lived in their current house or apartment for five years or more, vs. 68% of July Undecideds. Conversely, 31% of August Undecideds have lived in their current house or apartment between one and five years, vs. 22% of July Undecideds. In response to the question, “Are you more of a night owl or morning person?” 36% of August Undecideds told us they’re a morning person, vs. 44% of July Undecideds. Lastly, we learned that 39% of August Undecideds own a video game console (for themselves, not for their kids), vs. 30% of July Undecideds.

But what about September Undecideds? Interestingly, there weren’t any age proxy characteristics with statistically significant differences, but for age itself we did see a marked difference. 31% of September Undecideds told us they’re aged between 25 and 44, vs. 38% of August Undecideds, while 62% of September Undecideds were aged 45 or older, vs. 55% of August Undecideds. Unlike what we saw from July to August, in September we saw an age shift towards Romney’s base.

age distribution of undecided voters

Somewhat oddly, we saw large swings in Undecideds’ favorite type of car over the last three months. In July, 31% of Undecideds told us that SUVs were their favorite type of car. In August that number fell sharply to 21%, but then rose again to 29% in September. Opinions on compact cars were even more volatile. In July 13% of Undecideds listed compacts as their favorite type of car, vs. 35% in August and then 14% in September.

Undecided voters and car preferences

These data points match Undecideds’ profile shift toward Obama in August, and then back toward Romney in September. They also tell us that Governor Romney will have a hard time creating his own version of Scott Brown’s memorable pickup truck. It probably wouldn’t have helped Romney overcome his inconsistency issues had he tied his image to the Ford Explorer in August, then abruptly switched to the Focus in September.

So what can be learned from this? Keeping in mind that the proportion of Undecideds has remained fairly static throughout the year — our numbers have consistently put the number at 7% — we’re inclined to say that Undecideds’ recent profile shift toward that of a Romney supporter is bad news for Governor Romney, and good news for the President.

Knowing that polls showed a gradually increasing lead for Obama throughout September, it stands to reason that Undecideds who previously seemed more likely to vote for Obama have started to make up their minds, while one-time Romney supporters are now second-guessing themselves. This underscores how important it will be for Romney to win over the 41% of current Undecideds (see our previous post) who will be heavily weighing the results of the debates as they decide whom to vote for.

Image source