dings are reported by Alexander and Wartella (1979) in a replication of this study. Reeves and Greenberg (1977) and Reeves (1979) report that when children are asked to compare a variety of television characters, third-, fifth-, and seventh grade children use similar traits in the comparison. In both studies the researchers used a multidimensional scaling system (2) to determine the underlying dimensions children use to differentiate or compare one character with another. They report no age differences in their study. They determined that children at each age level--third, fifth, and seventh grade--utilized four main attributes in differentiating characters: humor, physical strength, attractiveness, and activity. Reeves (1979) attempts to resolve the lack of age differences in these multidimensional scaling studies with age differences found in studies of children's free descriptions of television characters and real people. He argues that differences between the studies may reside in the two methods used or the tasks employed (a comparative versus descriptive task); that is, the differences may be methodological. The two methods may require that children focus on different aspects of people in complying with the researcher's requests. However, if the study differences are not just methodological, Reeves offers an alternative theoretical explanation: Children may learn fairly early (by third grade) that television characters are relatively simple people and only simplified discriminations among them need be made. Thus, while Reeves and his colleagues find no age differences in the underlying dimensions children use to compare television characters, other descriptive studies (as noted above) have found age differences in how children describe TV characters. The difference in findings from these research approaches has not been resolved. There is support from other research for the finding that children only gradually come to develop an u...
