Voter targeting – an analytical approach

A common mistake those running for office make is a lack of understanding of voter behavior, which results in their spending unnecessary effort reaching out to potential voters. This article will address both the nature of the problem and what can be done about it.

Understanding voter data

Every election cycle, political scientists bemoan the fact that voter turnout is as low as it is. However, examining voter data shows that the reality of voter turnout is somewhat different.

To illustrate, we will analyze Louisiana voter turnout in the 2012 Presidential race, where over 2 million participated. This was a numerical voter turnout record, although the official statistics produced by the Secretary of State show “only” a 68% voter turnout (turnout was 67% in 2008). Does this mean that approximately a million voters stayed home?

The voter data shows that approximately 388K voters (13% of the electorate) have NEVER voted in any election – there is actually a nonvoter on the voter rolls who registered to vote in 1920. Furthermore, another 518K voters (18% of the electorate) last voted before the 2012 Presidential election. There is actually someone on the voter rolls who last voted in August 1966.

Adding up these two groups of “voters” gets you 31% of the electorate – almost exactly equal to the 32% of the “voters” who stayed home. In other words, voter turnout statistics are artificially depressed by these one million “voters.”


Breakdown of Louisiana voters (by number and %)

Breakdown of Louisiana voters (by number and %)

Implications for campaigns

Since there are a nearly a million voters in Louisiana who are essentially nonvoters, what is the implication for a political campaign? For starters, these voters should NOT be included in any poll, mailout, or phone call, with one exception: 26K voters did register to vote after the October 9 “cutoff” for the 2012 Presidential election. Until the 2016 election rolls around, these voters could presumably show up in a future election, since there is a sizable group (currently about 240K) of people who only participate in presidential elections.

Furthermore, not all elections are the same. Presidential contests typically bring out more voters than statewide elections, and statewide elections typically generate more voter interest than special elections/tax elections. So how should campaigns address this reality?

A way to target voters

JMC Enterprises of Louisiana believes that voter behavior is fairly predictable, and that predictability drives voter targeting. Towards that end, a “voter score” between 0 and 100 is assigned to each voter by measuring his/her election participation against other voters, and depending on the election, you only select voters within a certain “scoring range” (i.e., between 50 and 100).

To illustrate, we chose three elections: the 2012 Presidential election, the 2012 Democratic presidential primary (President Obama was essentially unopposed), and the 2011 governor’s race. Below are the voter scores of an average participant:

2012 GOP Presidential primary: 84

2011 Governor’s race: 76

2012 Presidential election: 61

Since voter score is a measure of a person’s likelihood to vote, it makes sense that a low turnout race like the 2012 primary (12% turnout) would draw a higher scoring voter than the (68% turnout) Presidential election, and that the 2011 governor’s race (with 37% voter turnout) would be in the middle.


In practical terms, this means that political campaigns can use voter score to focus only on likely voters, which will save them time and money from not having to pursue voters uninterested in participating in that election.