Using an Online Sample to Estimate the Size of an Offline Population

Abstract

Online data sources offer tremendous promise to demography and other social sciences, but researchers worry that the group of people who are represented in online data sets can be different from the general population. We show that by sampling and anonymously interviewing people who are online, researchers can learn about both people who are online and people who are offline. Our approach is based on the insight that people everywhere are connected through in-person social networks, such as kin, friendship, and contact networks. We illustrate how this insight can be used to derive an estimator for tracking the digital divide in access to the Internet, an increasingly important dimension of population inequality in the modern world. We conducted a large-scale empirical test of our approach, using an online sample to estimate Internet adoption in five countries (n ≈ 15,000). Our test embedded a randomized experiment whose results can help design future studies. Our approach could be adapted to many other settings, offering one way to overcome some of the major challenges facing demographers in the information age.Electronic supplementary materialThe online version of this article (10.1007/s13524-019-00840-z) contains supplementary material, which is available to authorized users.

Quantifying population contact patterns in the United States during the COVID-19 pandemic

“…There are surprisingly few existing estimates for the rate of contact in the US before the COVID-19 pandemic 6 – 8 ; here, we compare our estimates to contact patterns estimated from a probability sample of US Facebook users in 2015 (ref. 9 ) (see Supplementary Fig. 7 for a comparison of available pre-pandemic estimates of contact patterns).…”

“…We estimated the relative reduction in R 0 , assuming (1) that contact patterns in the population before physical distancing became widespread were equivalent to the 2015 study 9 and (2) that disease-specific parameters remained unchanged over the course of the survey period (see “Methods”). We find 73% (95% CI: 72–75%), 57% (95% CI: 53–61%), 48% (95% CI: 43–53%), and 36% (95% CI: 29–42%) declines in the implied R 0 in Waves 0, 1, 2, and 3 respectively, relative to the pre-pandemic period.…”

Monitoring global digital gender inequality using the online populations of Facebook and Google

“…Further extensions of this work could analyze these online gender gaps as outcomes to understand what factors explain their variation, and these data could be routinely collected to examine how they vary over time. These online platforms could themselves be used to field surveys and estimate digital inequalities, using approaches such as that in Feehan and Cobb (2019).…”

“…5 In the same database, data on gender gaps in specific digital skills are available for even fewer countries than data on more general internet use measures, with coverage of under 50 countries for all indicators. Routine survey data collection on individual-level ICT use within households is expensive, and while some population censuses are able to collect information on internet or mobile availability at the household level, intra-household inequalities are not captured in these data sources (Fatehkia, Kashyap, and Weber 2018;Feehan and Cobb 2019).…”

Quantifying population contact patterns in the United States during the COVID-19 pandemic

“…(2011) conducted a telephone survey of four counties in North Carolina and found that respondents reported an average of about 10 speaking interactions per day. AndFeehan and Cobb (2019), based on a probability sample of US Facebook users, found an average of about 12 conversational contacts per day. These previous estimates, though not exactly comparable, are suggestive of a decline in daily interpersonal interaction of about 70% (from an average of about 11 to an average about 3).…”