Elections and Candidates with Big Data
OPINION |

Elections and Candidates with Big Data

IN THE US, THE POLITICAL PARTIES' STATISTICAL INFRASTRUCTURE IS DECISIVE AT ELECTION TIME. GEORGE SOROS RECENTLY GAVE $2,5 MILLION TO THE DEMOCRATS TO IMPROVE THEIR DATA ANALYSIS IN AN ATTEMPT TO KEEP PACE WITH THEIR POLITICAL ADVERSARIES

by Tommaso Nannicini, Dept. of Economics, Bocconi
Translated by Alex Foti


In many countries, data gathering and statistical analysis are indispensable components of electoral campaigns. In Italy as well there has been much talk of big data, but large-scale databases alone, especially if obtained via online polling, are not enough. You need smart data, and this means gathering data in an intelligent and targeted fashion, possibly with various experimental approaches, in order to understand how to design electoral marketing for higher efficiency depending on the competitive environment at hand. The US is leading the process, but many other countries are learning the lesson.

For those interested in the subject, The Victory Lab, a book by journalist Sasha Issenberg, describes decades of experimenting in the industry: decades of trial and error, of successful and less successful collaborations between politicians, consultants, and academics. A mix of innovation by individual visionaries and hard-headed political party choices, like when in the 1980s the Democratic National Committee decided to invest in a unit for the gathering and analysis of political data. This approach eventually led to the redesign of traditional campaigning methods, such as door-to-door propaganda, the sending of postcards, and making a lot of targeted phone calls. These tools are now guided by data, rather than political intuition.

A few months ago, George Soros made a $2.5 million donation to an organization affiliated with the Democratic Party which is specialized in data analysis. The aim is to create a statistical infrastructure that can be used by democratic candidates in the upcoming mid-term elections, thus counterbalancing the efforts of rival republicans in this direction.

In the US, it’s clear to everybody that knowing how to employ data is not a sufficient condition for winning an election, but also that not knowing how is a sure-fire recipe to lose one. What about Italy? Local spin doctors study US best practice, but the lack of data, the quickness of campaigns, and the candidates’ cultural lag are a drag on the process. The country seems to be in the same situation the US experienced twenty years ago, and room for innovation is ample.

In a study I conducted with Chad Kendall and Francesco Trebbi (“How Do Voters Respond to Information? Evidence from a Randomized Campaign”, American Economic Review, forthcoming), we made a field experiment. In collaboration with the incumbent mayor of the city of Arezzo, Giuseppe Fanfani, who ran for re-election in 2011, we did a research study that asked various groups randomly chosen at different polling stations to respond to various electoral propaganda techniques (flyers, phone calls), as well as direct messages from the candidate, so to make more rigorous estimates of their effect on voters. We found that personalized propaganda techniques (such as volunteers’ calls) and messages revolving around the candidate’s competence had a stronger effect, especially among less polarized groups of voters.

Summing up: in Italy, there is plenty of room to make electoral campaigns more effective by gathering big data and analyze them rigorously. We need new forms of collaboration between politicians, researchers, and creative people. Words for the wise.

Latest Articles Opinion

Go to archive
  • Occupational Downgrading Puts Mental Health at Risk

    Immigrants tend to be overqualified for the lowskilled, highrisk occupations that they tend to take up. A research study shows that this fact increases the risk of psychological disorders

  • With Increasingly Flexible Work, the Old Stakes Are of Little Use

    New work time arrangements and forms require new ways of management and new protections. And therefore, new forms of labor relations and union organizations

  • Trust That Can Be Trusted

    There is social trust and institutional trust. And if the former is more stable to external shocks, the latter is more sensitive to them. The two can also diverge, as during the pandemic, when mistrust in US institutions towards emergency management corresponded to an increase in social trust

Browse the magazine in digital format.

View previous issues of Via Sarfatti 25

BROWSE THE MAGAZINE

Events

Mon Tue Wed Thu Fri Sat Sun
        1 2 3
4 5 6 7 8 9 10
11 12 13 14 15 16 17
18 19 20 21 22 23 24
25 26 27 28 29 30 31