Identifying and influencing as many people as possible to vote, is one of the areas political campaigns spend the largest amount of resources on. Campaigns often use statistical tools and social media analytics for deciding who to target and how (also known as ‘micro-targeting’).
To target their ads efficiently politicians, follow a two-step process. First they use analytical techniques to identify people who are suitable to direct their mobilization efforts at (also known as voter targeting). Particularly, they identify people who are likely to support their cause, party or themselves as the candidate and then how likely they are to turn out to the voter polls. This means that politicians need rich voter data and powerful data models that can predict such information with as high accuracy as possible. The data used to conduct these algorithms, come from a variety of sources such as the voters list, polling, door-to-door canvassing, census data, social media and commercial data brokers. Yet, there is no standard protocol for conducting statistical analysis of the public to target them appropriately.
Following that, they craft measures to best motivate these people to vote for them. This includes making decisions on which tactics best translate to mobilization. Micro-targeting, therefore, means that you and your similar-but-different friend might get different campaign literature from the same candidate, encouraging you to vote for their party. An example of this may be the advertisement you see for the party’s website being different than the one your friend sees on her social media channel.
However, the targeted campaign may not be all it appears to be, as a growing body of research (conducted in North America, primarily after the USA 2016 election) has reported very little value of micro-targeting. The majority of the reviewed study suggests that while the public can be accurately identified and placed into groups, and therefore targeted appropriately, like all statistical models, the algorithms used to group these individuals are only about probabilities. Furthermore, as increasing awareness happens, voters can get turned off by ‘targeted marketing’, that may have violated what they believe is private information to achieve— and may punish the campaigns that use it. However, it is worth noting that the privacy laws in Canada which govern the extent to which individual-level data on the Canadian consumer can be sold on the open marketplace are relatively stringent.
The framework for micro-targeting has its pros and cons. The framework generalizes approaches used by political campaigns and is relatively easy to communicate to those familiar with big data analytics and logistical regressions. Micro-targeting allows politicians to create a segmentation of the data which corresponds to building data-driver profiles of groups of potential-voters to enhance the effectiveness to targeting measures.