Adolescent groups, what classroom surveys show

Nomination surveys ask each pupil in a class or school to name classmates according to different criteria. They produce a complete graph of the population surveyed, an heir to the sociometry Jacob Moreno developed in the 1930s.

That exhaustiveness distinguishes them from ordinary surveys: both declarations and their reciprocity are available, which allows structural positions to be measured and not just opinions.

The design changed scale with the large longitudinal surveys, such as the American Add Health study launched in the mid-1990s, which collected the nominations of tens of thousands of pupils in many schools and followed some of them into adulthood.

The literature distinguishes sociometric popularity, measured by how many classmates say they like the person, from perceived popularity, measured by asking who is considered popular.

The two do not coincide. Several studies, notably by Antonius Cillessen and Amanda Rose, show that the second is associated with dominant conduct that reduces the first. A highly visible pupil may thus be rarely named as a friend, and a quiet pupil liked by many.

Network analyses find resemblances between friends on many behaviours. Attributing those resemblances to influence requires ruling out selection: people connect with those who already resemble them. The two mechanisms often operate at the same time and reinforce each other.

Studies exploiting unchosen allocation, notably in boarding schools or by roommate draw, such as Bruce Sacerdote's 2001 study of students whose rooms had been assigned at random, find real influence effects but far weaker than the raw resemblances.

Nomination covers classmates within the school, which excludes outside relationships. Part of adolescent sociability therefore escapes the design, in particular online relationships and those formed in the neighbourhood, in sports activities or in the extended family.

In addition, the school strongly constrains possibilities: a map of classroom relationships measures school allocation as much as preferences. A pupil rarely named in class may have most of their relationships elsewhere, which the graph does not show.

Surveys distinguish rejected pupils, named negatively, from neglected pupils, simply named little. The classification proposed by John Coie, Kenneth Dodge and Heide Coppotelli in 1982 established that distinction. The two situations are associated with different trajectories in follow-up work.

That distinction is regularly lost in coverage, which treats both as one category although research has explicitly separated them since the 1980s. Being rarely named by the classmates of a given class describes a position in a graph, not a characteristic of the person.