Recently, a number of quantitative studies have explored systematic relationships between sociocultural organization and patterns of language use, based on the assumption that speech behavior reflects the basic categories of the community's social structure. Such systematic links were arrived at by correlating average group scores for sociolinguistic variables with one or more non-linguistic variables. However, the fact that group practices may not reflect the behavior of individuals within the group brings into question the validity of the measurement. A study of phonological variation within a group of 33 Qatari male speakers is used to examine such discrepancies between aggregated figures and individual figures. It is suggested that variable rules are an attempt to account for variability within a group, but that much of the variation is still idealized. The system of analyzing variation may be improved, it is proposed, if the focus is shifted to the function of language use as a symbol of personal identity. While the sociolinguistic technique of comparing average group figures has been useful in illustrating the systematic nature of many unexplained linguistic phenomena, a technique is still needed to explain variation within each group. (MSE)