Data and Modeling

Back to Basics, Better Models, Experimental Data

By far the most important and valuable data in a voter file is what everyone already has or can easily get; party registration, voter history, and census and neighborhood data.

These hard political and social facts undergird almost all of the explanatory value found in “microtargeting” models. All the consumer data, and even extra polling, simply pales in comparison.

We have in-house, top-of-field statisticians and data scientists who wring the most out of your core data.

And we add what no other firm can . . . persuasion modeling based on dynamic experimental data that tells you who is most likely to move, when and where.

We are continually conducting internal research and innovation to integrate recent advances, but never lose sight of what is most effective in the glare of shiny new things.

Social Context Matters – A Lot

Where a person lives can tell you a great deal about them. Whether the exact causal mechanism is self-selection into particular neighborhoods because of their desirable social characteristics, or the processes of socialization and conformity that bring people into line with their opinion environments, location information helps to understand and predict the opinions and choices of individuals.

For this reason, survey data and voter file information is always more powerful when it is merged with granular information about a person’s place of residence. At ES this information is used in analysis at multiple levels of geographic aggregation, including census geographies (block groups, tracts, counties), as well as voting precincts, zip codes and real estate-based neighborhood boundaries.

We begin by merging individual voter file records with the latest census demographic and economic data. Then we merge area-based political information to the files, such as precinct level support for a party’s candidates. These low cost sources of information, available for all voters with a complete street address, then serve as a starting point for analyzing a variety of outcomes, including past turnout, party preference, or survey-based measures of opinion.

Missing Movers Means Missing Voters

Americans move, a lot. According to current estimates, people make an average of twelve moves across the life-span. The American Community Survey estimates that more than 15 percent of Americans changed residences over the previous year – 46.6 million people overall.

The U.S. is the most geographically fluid society in the world. While that makes for a dynamic country, it also makes tracking voters difficult. These moves, and the lost data, accumulate quickly over the course of a couple of election cycles.

ES uses alternative name and birthdate matching criteria and voter files from adjacent states and across multiple years to identify and track movers across town, across the state, or to a nearby state from one election-cycle to the next. We identify these “new” registrants as re-registered and pull in the most powerful data there is; voting history from their previous address. Consumer information and modeled scores can never come close to providing the value inherent in hard voter records.

Traditional voter file providers and vendors utilize the National Change of Address (NCOA) Registry from the U.S. Postal Service to identify those who have recently moved. But many movers do not fill out the Post Office change-of-address form.

Our research shows this misses the vast majority of recent movers. For example, we identify well over one million recent movers (2008-2012) on the Pennsylvania voter file; the NCOA captures just 21 percent of them. In California, we identified over 2 million movers over the same period, while the NCOA system identified a minuscule 2 percent.

When a voter moves, she leaves the most important information on her political behavior behind; her voting record. Working off a standard voter file, you don’t know if she has never voted or always votes, what her Party registration was, or whether she voted in a primary election. In short, you lose almost everything that matters for planning voter contact.

New movers also turn out at lower rates than they otherwise should. The outreach of a political organization is especially critical in these circumstances; these are lost voters and lost opportunities of huge potential.

Campaigns and political organizations can no longer afford to miss so much of political reality.

Evolving Strategies

Get in touch today