Page 47 - ITU Journal - ICT Discoveries - Volume 1, No. 2, December 2018 - Second special issue on Data for Good

ITU JOURNAL: ICT Discoveries, Vol. 1(2), December 2018

USING ADVERTISING AUDIENCE ESTIMATES TO IMPROVE GLOBAL DEVELOPMENT STATISTICS

Ingmar Weber , Ridhi Kashyap , Emilio Zagheni
2
1
3
2
3
1 Qatar Computing Research Institute, University of Oxford, Max-Planck Institute for Demographic Research

Abstract – The United Nations Sustainable Development Goals (SDGs) are a key instrument in setting the
agenda around global development until 2030. These goals come with a set of 232 indicators against which
countries should monitor their progress with respect to the SDGs. Existing data sources to measure progress
on the SDGs and global population trends however are often (i) outdated, (ii) lacking international
comparability, (iii) lacking appropriate disaggregation, or (iv) missing completely. These problems are
often especially acute among less developed countries. In this paper we describe how anonymous, aggregate
data from the online advertising platforms of Facebook, LinkedIn and other services can be used in
combination with existing data sources to improve global development statistics. We illustrate the process
of using and validating such non-representative data through two case studies looking at (i) Internet access
gender gaps, and (ii) international migration statistics.

Keywords – Facebook, global development statistics, online advertising, SDG monitoring.

1. INTRODUCTION To address the challenges related to “data poverty”
and to potentially leapfrog the development of
The Sustainable Development Goals (SDGs) are a set traditional data acquisition infrastructure several
of 17 global goals set by the United Nations in 2015 researchers and organizations have explored the
to serve as guiding principles to set the course for a use of “Big Data” approaches for improving
better future. The goals include aspects of development statistics and spark a data revolution
economic, social and environmental development. [5].
To give the goals a practical meaning, they come
with a total of 169 targets, which, in turn come with The sources of big data most commonly used
232 indicators to serve as a global “benchmark” to include satellite data [6-8], call detail records [9,10],
monitor progress on the SDGs. web search data [11,12] and public social media
1
posts [13-15].
However, even before the introduction of these
indicators that countries are expected to monitor, Here we describe a new data source: anonymous
many countries are lacking even the most basic online advertising audience estimates provided by
development statistics. In a 2014 publication, the companies such as Facebook, LinkedIn, Google,
World Bank reports that 16 countries did not Twitter and others. In a nutshell, companies
participate in the 2010 census round, leading to providing targeted online advertising support
census data which is at least 10 years old [2]. Even potential advertisers in their campaign planning
when data is available, data quality is often poor, with information on how many of their users match
which basically renders it useless [3]. Issues of certain targeting criteria. This serves as a kind of
limited data availability and poor data quality are real-time census over the user base of the
particularly pronounced in less developed considered services, providing aggregate answers
countries, which at the same time have a lot to gain to questions such as “how many male Facebook
from proper data use [4]. users from Germany currently live in Geneva?”.

1 However, not all of the indicators already have agreed-upon
measurement criteria [1].

© International Telecommunication Union, 2018 25