widely, and these include measures of the type (e.g., solitary, parallel, coordinated) and complexity (e.g., exploring things, constructing things, pretending things) of play (see Ladd and Price, 1993, for a review), as well as assessments of how well children appear to be getting along (e.g., prosocial exchanges, aggressive exchanges, withdrawn behavior) and the emotions they are expressing (e.g., positive, angry, sad). Observational measures are not appropriate, however, for the study of enduring relationships since they capture only brief episodes of interaction. Researchers have also asked groups of children whom they like and dislike, which can be done effectively with children as young as age 3 (Coie and Dodge, 1983; Newcomb and Bukowski, 1983). Children are classified as popular (many “like” and few “dislike” nominations from their peers), rejected (many dislike and few like nominations), neglected (few of either kind of nominations), controversial (many of both kinds of nominations), or average. These classifications and the continuous measurement of liking and disliking can be used to explore the emotional, behavioral, and cognitive competencies that influence peer acceptance. These ratings may not generalize, however, beyond the group tested. For example, a child who isn't popular at preschool may be popular in her neighborhood. Unfortunately, we know little about how variation in acceptance across different peer groups affects children.
Peer status is not written in stone, even when assessments are focused on a child's standing in the same group over time. Among preschool children, the number of liking and disliking nominations a child receives at one time accounts for only about 25 percent of the variation in the number of nominations she receives even a short time (3 weeks) later (Olson and Lifren, 1988). Classifications based on such measures are also only modestly stable, with popular, rejected, and average classifications typically being more stable than controversial and neglected classifications (Newcomb and Bukowski, 1984). These methods are also probably culturally bound. Making decisions about who you like and don't like may make sense to children in cultures in which common topics are deciding who does and doesn't get to come to one 's birthday party, who does and doesn't get invited over to play, and who can and cannot be “my friend today.” For children from cultures that encourage them to like all the children in a group, however, asking such questions may make little sense.
Despite these limitations, these so-called sociometric measures have yielded important findings. This is especially true with regard to children who end up in the rejected classification. Most of the work on peer rejection comes from studies of school-age children, so extrapolation from these studies must be done with caution when considering younger children. By the early school years, peer rejection is clearly a risk factor. Rejected children are overrepresented among adults with psychiatric problems (Cowen et al., 1973), among children who do poorly at school (Coie et al.,