Unassisted Humans Infer Personal Traits from Facebook Group Memberships: An Empirical Study with Implications for Employers and State Entities

The practice of using online social network (OSN) profiles and other internet-based records by third-parties in order to evaluate individuals for various purposes, known as cybervetting, is growing more popular. The United States State Department now requires all non-immigrant and immigrant visa applications to supply OSN profile identifiers operated by applicants on various platforms, including Facebook, Instagram, and Twitter (referred to as the “social media registration” requirement). Employers and recruiters regularly use OSN profiles and related information to screen or monitor employees and job candidates. In these contexts, certain personal traits of individuals may be considered especially sensitive, especially where human reviewers are decision-makers. Visa applicants may not wish to disclose information about themselves that is not explicitly required (such as religious and spiritual beliefs or sexual preference) for fear of discrimination. Similarly, job applicants may wish to keep private certain personal traits (such as race, ethnicity, gender, and age), even if their influence in decision-making would constitute illegal discrimination. The aim of this research is to determine if Facebook Group memberships can disclose users' information that may be considered sensitive, private, and/or legally protected to human reviewers. It is motivated by the observation that the non-hidden Facebook Group memberships of any user are publicly discoverable (with some effort), which may contradict users’ expectations of the privacy of their aggregate group membership information, and therefore not have been treated as a potential source of public data disclosure. We first collected real Facebook profile information from 32 users with diverse demographic backgrounds. We then conducted an empirical study with 63 participants to measure the ability of humans to infer eight personal traits (race and ethnicity, gender, age, religious and spiritual beliefs, relationship status, highest level of education, employment status, and income) of these users based exclusively on their Facebook Group memberships. Our results show that certain traits are more inferable by human reviewers than others. Participants were able to infer race and ethnicity identities of 50% of subjects more than 88% of the time, and gender identities of 50% of subjects more than 70% of the time. We discuss the implications of our findings in the context of current regulations that prohibit employers from requesting sensitive demographic data as well as formal governmental processes that require foreign nationals to disclose their social media profiles.

Name thesis:degree_name

Master

Department dc:contributor.department

Massachusetts Institute of Technology. Institute for Data, Systems, and Society

Grantor dc:publisher

Massachusetts Institute of Technology

Year dc:date.issued

2023

Author dc:creator

• Paeth, Kevin

Advisor dc:contributor.advisor

• Liccardi, Ilaria

Statement dc:rights

• In Copyright - Educational Use Permitted
• Copyright MIT

Licence dc:rights.uri

• http://rightsstatements.org/page/InC-EDU/1.0/

Handle dc:identifier.uri

https://hdl.handle.net/1721.1/150085

OAI identifier oai:identifier

oai:dspace.mit.edu:1721.1/150085