For political campaigns, there are two primary goals:
1. Get as many people as possible to support you
2. Get these supporters to come out and vote for you
If the above two are done successfully, then theoretically you should win the popular vote (though sometimes that is not enough). However, conducting political campaigns requires financial backing and staff. Like any true economic model, there is scarcity: you cannot have unlimited campaign finance money and you cannot have unlimited staff to conduct outreach efforts. Thus an outreach strategy that allocates scarce resources for maximal output is the critical third but unstated goal of any campaign.
3. Have an outreach strategy that efficiently deploys campaign finance & resources
Campaigns have time and again utilized technologies and methods to enhance their efficiency. The recent advent of Big Data has made data-driven decision making a reality. Starting from the Howard Dean’s campaign to the recent midterm elections in the US, data-driven election campaigns are commonplace. Using historic data from publicly available demographic profiles (census), past voting records, voter registration records, and aggregate profile of a locality, campaigns are able to build a picture of the target constituency – Project Narwhal of the Obama campaign is a relevant example. However it is not about reaching everyone in the target constituency – as there are likely to be supporters of the opponent as well and efforts to mobilize them could be counterproductive. Targeted and precise outreach efforts are necessary for supporters who are [1] likely to become supporters and [2] come out and vote. For this purpose individual outreach efforts through volunteers and staff are effective but expensive. This is where Big Data and associated analytics can help.
Nickerson and Rogers in their paper – Political Campaigns and Big Data discuss the predictive models campaigns use to plan their outreach strategies and their effectiveness. These models measure three salient outcomes among others: behavior scores (whether the citizen is likely to engage in a political activity), support scores (historic data to understand the targeted citizen’s voting preference), and responsive scores (will the citizen be receptive to the targeted campaign). Using these predictive models, the Obama campaign (both 2008 and 2012) rallied individual voters to its cause. By maximizing the outcomes gained from individual targeted outreach efforts, the Obama showed the power of the use of data and grassroots mobilization of supporters. Success of the Obama campaign hasn’t gone unnoticed and campaigns around the world have adopted data-driven election campaigns.
Data and poles don’t win elections, people win elections!
However, technologies like Big Data and predictive modeling can only get campaigns so far. The real battle is of context: policies, the gravitas and charisma of the candidate, and the political conditions for the elections. As the popular statement goes: data and poles don’t win elections, people do. As witnessed by the recent midterm elections, all the data driven outreach and grassroots mobilization of voters is useless if the political conditions are against the incumbent.
As a Pakistani-American, I try to stay abreast with the developments and political climate in Pakistan. Last year’s election in Pakistan were momentous for many reasons. One of them was the emergence of a third-option – Pakistan Tehreek-e-Insaf (PTI) party led by the charismatic Imran Khan (cricket turned philanthropist turned politician). PTI utilized campaign strategies reminiscent of the Obama 2007 campaign – Tabdeeli (urdu for Change), Hope, and Naya Pakistan (New Pakistan). Much like the Obama campaign, PTI mobilized grassroots level campaigning – signing up new voters, getting the youth to support the cause,
PTI and Imran Khan Election 2013 Campaign Poster
and neighborhood volunteer efforts to get out the vote. Imran Khan and PTI became the darling of the public and news media; election win seemed inevitable – but Imran and PTI lost. All the analytics and grassroots campaigning failed to take into account the context and conditions of the country. PTI was a popular candidate in urban metropolises of the country but they constitute mere 38% of the total population. The rural areas with limited infrastructure, education, and media reach, were still voting for the established and ingrained parties: PML and PPP. Add to the fact that there were cases of election rigging and lack of transparency, PTI was never going to challenge the established parties. Data, analytics, and localized electioneering can only get you that far.
Ali S. Kamil 