Lathan HS, Kwan A, Takats C, Tanner JP, Wormer R, Romero D, Jones HE. Ethical considerations and methodological uses of Facebook data in public health research: A systematic review.
Soc Sci Med 2023;
322:115807. [PMID:
36889221 DOI:
10.1016/j.socscimed.2023.115807]
[Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 10/31/2022] [Revised: 01/10/2023] [Accepted: 02/22/2023] [Indexed: 03/08/2023]
Abstract
OBJECTIVE
Since 2016, around seven in 10 adults in the United States (U.S.) actively use Facebook. While much Facebook data is publicly available for research, many users may not understand how their data are being used. We sought to examine to what extent research ethical practices were employed and the research methods being used with Facebook data in public health research.
METHODS
We conducted a systematic review (PROSPERO registration CRD42020148170) of social media-based public health research focused on Facebook published in peer-reviewed journals in English between January 1, 2006 and October 31, 2019. We extracted data on ethical practices, methodology, and data analytic approaches. For studies that included verbatim user content, we attempted to locate users/posts within a timed 10-min period.
RESULTS
Sixty-one studies met eligibility criteria. Just under half (48%, n = 29) sought IRB approval and six (10%) sought and obtained informed consent from Facebook users. Users' written content appeared in 39 (64%) papers, of which 36 presented verbatim quotes. We were able to locate users/posts within 10 min for half (50%, n = 18) of the 36 studies containing verbatim content. Identifiable posts included content about sensitive health topics. We identified six categories of analytic approaches to using these data: network analysis, utility (i.e., usefulness of Facebook as a tool for surveillance, public health dissemination, or attitudes), associational studies of users' behavior and health outcomes, predictive model development, and two types of content analysis (thematic analysis and sentiment analysis). Associational studies were the most likely to seek IRB review (5/6, 83%), while those of utility (0/4, 0%) and prediction (1/4, 25%) were the least likely to do so.
CONCLUSIONS
Stronger guidance on research ethics for using Facebook data, especially the use of personal identifiers, is needed.
Collapse