statistics

The campaigns of 2008 are now over and many organizations are left with a nice amount of data from their outreach programs. This post will give a brief introduction to the use of regression analysis to look at your data in new ways.

Simple regression allows you to isolate a single variable, in most of your cases probably a contact action.

When you have a lot of data you can group it into sets where all of the variables match except for your chosen dependent and independent variables.

For example, you would create a set of data of 20-year old female voters from the same precinct. You could then look at that data to see how effective your contacts were with that specific demographic. Once you have created a few data sets you can identify some new relationships. It's possible that a specific contact action was more effective with women than men, or with students than young professionals.

Regression analysis allows you to restrict your attention to a single explanatory variable, which may allow you to see something you might have missed otherwise.

The polls are open in New Hampshire (indeed, in some towns they are already closed), and around 7 or 8pm we'll start to get some results. It's pretty much in the bag that Obama will carry the state, the questions are by how much, and what will we see from young New Hampshire voters? Will youth turnout in New Hampshire triple, as it did in Iowa? Will Obama once again beat out Clinton among Millennials by 5-1? Or will Clinton's last minute appeal to young voters cut into his lead?

New Hampshire By the Numbers
According to CIRCLE, who calculated their projections using 2007 Census data for 17-28 year olds (roughly today's 18-29 year olds), voters under 30 make up 20 percent of New Hampshire's population. 38% of those young voters are students (16% high school, 22% college). In raw numbers, there are 195,031 eligible young voters in New Hampshire, 43,215 college students, and 8,910 high school students. These are the baselines against which tonight's turnout will be measured.

In 2004, while participation in the Iowa caucus among young voters quadrupled, it increased only slightly in New Hampshire, rising to 14% from 13% in 2000 (pdf). Like in Iowa, New Hampshire has election day registration, a process which is known to bump youth turnout by as much as 14%. With Obama bringing so many new people into the system, this could also be a factor tonight and will be something to watch. The Union Leader is already reporting on potentially record-breaking turnout, and a high number of voter registrations at the polling places.

The Candidates
Reports are plastered all over the pages of our major media that Hillary has spent the last 5 days reaching out to young voters, particularly on-line. Hillary has introduced some interactive Facebook programs, and spent the week uploading content like mad onto her YouTube page, but it's difficult to see how this cuts into Obama's lead among youth. The Obama campaign didn't spend the last year just making Facebook groups and uploading videos to YouTube. Online activism is about more than content. It's about creating a community and transferring the online action off-line. Peer-to-peer, on-the-ground organizing is still the most effective form (pdf), among voters of all ages. It also doesn't help when your surrogates are out trashing young voters.

According to Rasmussen Reports, the only polling outfit where I could find not only cross tabs on young voters, but cross tabs that break out youth by the 18-29 age range put the race for New Hampshire youth in Obama's favor by about 12 points:

Obama: 38%
Clinton: 26%
Edwards: 17%

The margin of error on this is probably quite high (due to a small sample size), and I think that this will probably turn out to be conservative. The polls close in just a few hours and we'll soon find out.