Using R. E. Clark's concept of media attributions, a study examined the evolution of fourth through tenth grade students' perceptions about computers on three dependent variables--preference, perceived learning, and perceived difficulty, over the course of 3 years. Subjects were 339 public school students in Tennessee who completed a self-administered questionnaire. Findings demonstrated clear evidence of novelty effects. Students' judgments regarding preferences for computers declined significantly as did their perceptions of learning from the technology during the 3 years. Perceived difficulty of using computers, which was expected to decline, remained stable. In addition, both gender and age proved to be significantly related to all three dependent variables. Older students were consistently more skeptical about the technology than were younger students, and boys were consistently more positive than girls. These relationships showed no evidence of change over the course of 3 years. The results support critiques of the methodological limitations of the dominant approach to the study of computer effects in learning environment. Reports of short-term experimental applications of the technology have led to misleading generalizations about the computer's instructional potential. (Thirty-three references are included.) (PRA)