The amount of data that exists in the world if properly researched and collated suddenly reveals a lot about a person’s habits. This information is all kinds of things ranging from income, age, race, voting history of a particular area, magazines subscriptions, college education, high school education, where someone works; all of these things tell something about how a person is likely to vote.
For years exit polling had been correlated post elections to help the parties determine exactly who was voting for what party and on what issues were they voting. The evolution of data has been that now the predictions and the models can be run before elections now, not after. With the massive advancements in data collection and data mapping, there’s now enough information built up that parties can actually keep track of who has been active in election cycles and will use the information for campaign
contribution drives amongst other things.
As more and more of people’s personal data becomes available, complicated algorithms can be created to help create a more accurate examination of people’s voting habits and then their past habits become clear predictors of future voting behaviors.