The notion that "ditching your fat friends" will lessen your own propensity to become fat is either stupid, or else implicitly assumes that obesity is viral.

The NEJM analysis implies that having friends who are overweight (and, importantly, thinking of them as friends rather than acquaintances) increases one's risk of obesity through (the authors theorize) changes in one's perceptions of social norms. If this is conscious - that is, if one actively and knowingly becomes more accepting of overweight / obesity through friendship - then "ditching your fat friends" seems pretty silly, as you could simply consciously decide not to let yourself gain weight.

If this effect is unconscious, on the other hand, then contagion is the proper metaphor, as obesity spreads by person-to-person contact without the victims conscious participation.

Either way, Saletan's conclusions are off base.