The VoterProb® algorithm
How does it work?
While the algorithm is proprietary, we look at two types of voter behavior:
Recency
Votes that are cast more recently are obviously worth more than those in the past. Many other systems simply assign the same value to all elections, meaning that if you missed one election ten years ago, out of ten elections, you’d have a 90% chance of voting. That’s not the right way to do it.
In fact, missing a more recent election is a red flag, especially for a super voter. Sometimes, voter files are not culled for those who died or moved, resulting in spurious voter records. If a voter doesn’t vote recently, VoterProb® dings that person’s score accordingly.
Frequency
Voters should – and do – get credit from VoterProb® for every election in which they voted. That means even ten years ago, the score gives credit for votes cast. It’s just that simple.
Micro-targeting…
VoterProb® is a form of micro-targeting. It helps identify likely voters, and is especially powerful when combined with other forms of voter information. While there are databases easily available that contain information on consumer behavior, demographics, issues of importance, and other types of data, VoterProb® adds the invaluable probability of whether that voter will even cast a ballot.
While a voter may be micro-targeted for contact by a campaign, political organization, or special interest group, it’s pointless to follow up unless that voter is expected to vote. And, VoterProb® provides an easy to understand probability of just how much effort would be required to get that vote to happen.
…Proven on the Macro level
While VoterProb® is mostly used to target key groups of voters, it is also useful in predicting voter turnout. That’s because every person’s vote is assigned a (usually) fractional value. A perfect voter has a score of 1, or 100%, a lackadaisical voter may have a score of .50, or 50%, and a disengaged voter may have a lowly .01 or 1% chance of voting. These fractional vote values are then added up to predict voter turnout, which is the macro level. In general, voter predictions by VoterProb® are routinely over 94% accurate, proving that micro-targeting with these probabilities makes sense.