This paper contends that the survey research approach is limited for generating substantive data to improve college teaching, because it is impersonal, lacks opportunities for probing, suppresses the idea of discourse by offering fixed-choice and yes-no questions, disregards respondents' social and personal contexts of meaning, and is dependent on the competence of coders as ordinary language users. In effect, the role of language is ignored in collecting faculty evaluation data. Use of the focus group approach is presented as an alternative. Three in-class faculty evaluation survey instruments were completed by 468 university students, and focus group interviews were conducted with 86 students. Survey instruments had varying numbers of fixed-choice and/or open-ended questions and covered instructor performance, testing, and course objectives. Surveys revealed that faculty members were doing a good job, leaving unanswered the question of why students were dissatisfied with some of their professors and why students were not achieving. The focus group approach was more helpful in linking the conceptualization of the problem to the collection and analysis of data; delineating connections between causes and effects in student achievement; and specifying students' need for affection, empathy, or belonging to a subculture. (JDD)