Wednesday 25 November 2015

Data-driven Microtargeting in the 2015 General Election


By Dr.  Steve Patten

If there was any doubt, the 2015 general election campaign confirmed the arrival of the era of database politics. All of Canada’s major parties now rely on massive databases, data analytics and predictive modeling, and data-driven microtargeting to maximize their opportunities for electoral success. More than ever before, parties are able to derive intelligence on the electorate from polling and data mining, and this research informs party strategy, including the crafting of messages that are likely to win the support of key segments of the electorate.
But, parties have also built their own voter databases, sometimes called voter management systems, and these databases are used to identify those individuals who are likely supporters or could be persuaded to become supporters. The process of targeted communication designed to influence and mobilize identified voters is known as microtargeting.

 The backbone of party databases is the electronic voters list – containing the name, address, gender, and date of birth of each eligible voter – provided by Elections Canada. The parties merge this list with their membership and donor records, and then employ a range of techniques to gather and input information on voters’ cultural background, occupation, policy concerns, and more. The Conservatives led the pack with the development of their database, the Constituent Information Management System (CIMS), in 2004. In the 2006 and 2008 general election the CIMS database was effectively employed in battleground constituencies where a centrally coordinated voter contact programs were used to identify and get supporters to the polls.

By 2011 all three major parties had roughly similar databases, but the Conservative database contained considerably more personal information on voters, and it was employed most effectively. Thus, as they prepared for the 2015 election, the NDP and Liberals overhauled their databases, known as Populus and Liberalist, and invested heavily in training local campaign teams to collect and employ data in voter persuasion and mobilization. Both parties develop in-house analytics operations, with the Liberals spending three times what they had invested in in data and data analytics in 2011.

As an illustration, the Liberal Party’s 2015 central analytics team employed their research to develop a predictive model that identified the personal characteristics of voters who were, first, highly likely to vote and, second, highly likely to vote Liberal. The analytics team employed this model to construct a six-tier ranking system that guided the voter identification and get-out-the-vote (GOTV) efforts of local campaigns. In one Alberta-based battleground constituency the local campaign team found that there was a 60% chance that a visit or telephone call to a tier 1 voter would result in the campaign identifying a supporter they would want to mobilize on election day. The corresponding results for tier 2 and 3 were in the 35-40% range, and numbers dropped off after that. Thus, the decision was made to focus the canvass campaign on households with tier 1 through 3 voters.

Using an app designed for smart phones and tablets, Liberal canvassers were provided with the addresses (or telephone numbers) of the tier 1-3 voters they were to contact. After speaking with the voter, the canvasser would then use this app to input information about this voter, including whether or not they indicated support for the Liberal candidate. Once uploaded to the central Liberalist database, this information would be available to guide future communication with that voter.

The Liberalist software’s functionality includes a capacity to generate letters to be mailed to voters or send email or text blasts to specific groups of voters. Most campaigns used these functions to intensify their GOTV efforts, but they were also used for fundraising and persuasion. Canvassers in some local campaigns were armed with a variety of centrally produced issue cards, and information extracted from Liberalist determined which card they would leave with the voter.

Although Canada’s parties are still learning how to make the most of their databases and voter management software, there is no doubt that microtargeting has made for more efficient GTOV efforts, and even influenced the results in some battleground constituencies. As the scope and detail of the information in databases expands, and parties become more proficient at employing microtargeting in voter persuasion, highly personalized targeted campaign messages will rival the importance of the messaging of the national campaign and leaders tour, making campaign communications less and less transparent. In fact, data-driven microtargeting shifts the focus of partisan campaigns from the work of public persuasion and the building of a national consensus toward what could be described as manipulative exercises in private persuasion. Concerns have also been raised about the fact that party databases are not governed by either of Canada’s two core privacy laws. The loss of transparency, the manipulative character of targeted persuasion, and privacy concerns suggest data-driven microtargeting is not making a positive contribution to Canadian democracy.

This blog post was initially prepared for a post-election analysis project sponsored jointly by UBC Press and Samara. The complete collection is available online:
http://www.ubcpress.ca/CanadianElectionAnalysis2015/

This and related posts are available on Samara’s Blog:
http://www.samaracanada.com/samarablog?category=political-news



Steve Patten is an associate professor of political science at the University of Alberta. His research and teaching are driven by an interest in the ways particular ideological agendas and approaches to politics and governance shape the quality of democracy in Canada. Most recently, he is co-editor (with Lois Harder) of Patriation and Its Consequences: Constitution Making in Canada (UBC Press, 2015) and author of “The Politics of Alberta’s One-Party State” in B.M. Evans and C.W. Smith, eds., Transforming Provincial Politics: The Political Economy of Canada’s Provinces and Territories in the Neoliberal Era (University of Toronto Press, 2015).

No comments:

Post a Comment