Extended Data Table 3.

Summary Statistics for Analysis Samples

Panel A presents summary statistics for our primary analysis sample, which consists of individuals between the ages of 25 and 44 as of May 28, 2022 who reside in the United States, have been active on the Facebook platform at least once in the previous 30 days, have at least 100 U.S.-based Facebook friends, and for whom we are able to allocate at least one friend to a setting using the algorithm described in the ‘Variable Definitions’ section of Methods. Panel B replicates Panel A for the subsample used to measure childhood economic connectedness: individuals in the primary analysis sample whom we can link to parents with valid SES predictions (see Supplementary Information A.2 in our companion paper⁷) and who can be assigned to a high school (see Supplementary Information B.1). For each variable, columns 1-7 present means, standard deviations and selected percentiles. Columns 8 and 9 report means and standard deviations for the corresponding variables using the nationally representative 2012-2016 American Community Survey (ACS) for median household income and the 2014-2018 ACS for the remaining variables. Age and gender are self-reported by individuals. English Language refers to whether users set their language to English in the Facebook data and to the share of individuals who speak English only or speak English “very well” in the ACS data. Years on Facebook and Number of Facebook Friends are only observed in the Facebook data. In both panels, we report the number of Facebook friends within our primary analysis sample, as opposed to the total number of Facebook friends. Share of Friends Assigned to a Setting refers to the set of friends that we are able to assign to one of the six settings, as described in Supplementary Information B.2. The remaining variables are obtained from the 2014–2018 American Community Survey (ACS) and are assigned based on an individual’s residential county. The ACS means for these variables are population-weighted averages of the same county-level variables.