Similarity and Perceived Similarity of Personal Constructs as Influences on Friendship Choice

It is argued here that one derivable prediction both from within and outside Personal Construct theory is that similarity in the content of construct systems should underlie friendship. A further prediction is that friends can perceive the similarity which exists between them. Accordingly a mixed‐sex group of 40 student teachers was given a sociometric test and the Reptest, and then asked to provide for each construct a list of people who shared it. The Reptest analysis produced two mean similarity scores for each subject: one from observable similarities with friends (‘friendship pairs’) and one from similarities with non‐friends (‘nominal pairs’). Significant differences were found between these means. It was also found from the last test that subjects tended to overestimate the similarities between themselves and their chosen friends. These results are discussed as strong evidence not only for the similarity hypothesis when a Kellian view is taken of personality, but also for the usefulness of using Personal Construct theory as an experimental and hypothetical rationale for studies in social psychology.