Unlikely social network findings

The methodologies used to determine social network effects may be subject to large inaccuracies. Two re­searchers, in Boston and New Haven, United States, have shown apparent network effects for features they consider unlikely to be networked.

The study included data for 4,300–5,400 adolescents, subsamples from the National Lon­gitudinal Study of Adolescent Health. Each par­ticipant nominated a friend, and information about headache severity, acne severity and self-reported height was obtained in 1994–1995, 1995–1996 and 2000–2001. There were significant inter-friend cor­relations for all three measures. A friend with acne increased the risk of acne by 47%, and a friend with headaches increased the risk of headaches by 62%. A friend’s height also influenced a person’s height. After adjustments for environmental con­founders, these correlations became smaller and non-significant.

Apparent social networking connections should be interpreted with caution.

Cohen-Cole E, Fletcher JM. Detecting implausible social network effects in acne, height, and headaches: longitudinal analysis. BMJ 2009;338:28–31; Steptoe A, Diez Roux AV. Happiness, health, and social networks. Ibid:1–2 (editorial).