By Andy Hunn, Chief Operating Officer
Today, Resonate announced the release of its 2014 Voter Motivation Landscape study conducted from our vast, ongoing research. The study dissects survey data that ultimately defines 10 key voting populations that will determine the outcome on Election Day.
We undertook this project to see what our proprietary research would tell us about today’s electorate if we allowed the data to speak for itself. Moreover, what would happen if we suspended conventionally held norms and beliefs that preordain how to slice the voting population, like party affiliation and demographics? In a world of Nate Silver and “big data” – and given the massive volumes of data we have at Resonate – could we help clients gain a new, deeper understanding of voters, one that benefits people and candidates alike? The answer is yes. And here’s how…
The first step in the process required us to determine the best form of data clustering that would let the data tell its story. Without delving too deep into the statistical wizardry, we implemented “k-means clustering,” an analytical technique that assigned people to groups most similar to one another, while each group still remained distinct from the others.
After testing these groups, or segments, using a variety of inputs – like party affiliation, demographics, and ideology – the math revealed that optimal results were achieved when using “issue positions” as the determinants of the segments. In other words, when the models used ALL available inputs, they chose specific stances on issues as the most accurate method to define the voting population.
In today’s increasingly fragmented electorate, this makes a lot of sense. Knowing a voter supports gun rights, opposes offshore drilling, or is persuadable on same sex marriage reveals more about who that person is at their core. And, consequently, for whom they may cast their vote. It goes much deeper than voter file data that, at best, holds party affiliation and their voting record – a record of attendance, not who they actually voted for.
Although rational, this study is also quite profound due to its unique approach and the implications of its findings. The sheer volume of research analyzed is 3 to 4 times the size of typical study samples. The samples of significant studies like Pew’s Beyond Red vs. Blue and NBC/Esquire’s The New American Center are only 3,029 and 2,410 respectively. Using data as granular as issue positions like we did requires much bigger sample sizes, of which ours totaled 8,053.
While they still express an ideological spectrum, Democrats and Republicans are present within each voter segment. But since party affiliation wasn’t used to force the composition of the segments, we have gained new dimensions of understanding into these voter populations. Most significantly, relationships between issues that matter most to voters were unveiled. Furthermore, we see that conventional voting blocs such as “women” and “Hispanics” simply don’t exist outside superficially aggregated data. These “blocs” are as diverse and complex as the entire electorate itself.
We are launching the Voter Motivation Landscape Microsite to provide all with access to this rich granularity in our research. We want everyone to learn and to discover more about the voting electorate and what they truly care about. If we can reach and engage voters more effectively and speak to them in ways that compel them to get involved, we all win.
Take a look around. Absorb the implications of dissecting the electorate in this new way. Let us know what you discover.