Princeton Election Consortium

A first draft of electoral history. Since 2004

The voter influence measure

August 8th, 2008, 10:37am by Sam Wang


A reader James asked about the extreme values that the “Voter Influence” measure was taking on. His comment and my reply are here. This quantity (seen in the right sidebar) expresses the marginal influence that individual voters have on the final outcome probability. Per his feedback, I’ve modifed it…

In 2004, a very close race, I used NJ voters as the measure – the “jerseyvote.” However, jerseyvotes are worth little and fluctuate wildly. Also, 2008 is politically rather different, at least for the moment: measuring voter power makes the most sense if there is an overall swing nationwide so that local changes make a substantial difference. Remember the goal: to identify where local resource allocation (such as GOTV) will have the most effect.

Today I’m making some changes to the MATLAB and python scripts to reflect the discussion. Assuming they stick (which is chancy, I’m soloing today), they will be:

1) Voter influence is now calculated under the assumption that overall, state polls have shifted by a constant amount, one that makes the win probabilities 50-50. This change is equal and opposite to the Popular Meta-Margin, i.e. a swing of approximately 3% toward McCain. It’s equivalent to assuming that some large-scale campaign event occurs in the very near future to make things a toss-up. (This is one case in which taking non-independence among states may affect the result.)

2) Voter influence is now normalized to the day’s most powerful voters – today, Nevada. What to call the units – centinevadas?

Tags: 2008 Election

No Comments so far ↓

There are no comments yet...Kick things off by filling out the form below.

Leave a Comment