Small networks and high isolation? A reexamination of American discussion networks

Recent findings from the General Social Survey suggest that discussion networks have shrunk, and social isolation has increased, since 1985. These data have been threatened by possible artifacts, however, and may be unreliable. This paper reports on a nationally representative experiment that estimates discussion network size and evaluates several proposed explanations for the shrinking networks finding. It finds that modern discussion networks have decreased in size, but that social isolation has not become more prevalent. It also disconfirms several explanations for shrinking networks, and validates a method for increasing the number of names provided in response to name generators.

► Nationally representative experimental examination of American discussion network size.
► The average American has roughly two discussion partners.
► Social isolation has been overestimated in previous research.
► Estimates of social isolation may be inherently unreliable.
► Tests a method for improving name generator responses.