Attempts to explain poor academic performance of Mexican American bilingual students have relied primarily on two causative models: home and school. Both models were evaluated with data from a matched sample of public school bilingual Mexican American children from grades 1-8, in a medium sized South-Central Texas community. The control group students, identified as low achievers, were participants in a compensatory educational program which consisted of a modified behavioral modification schedule and strategy designed to stimulate academic achievement through: the usual token economy reward structure; the use of bilingual tutors and various cultural activities designed to strengthen the self-concept of the Mexican American child; an effort to enlist the family's interest and cooperation in the educational process. These students were matched by age, grade, sex and social class with other Mexican American students not identified as poor achievers. Students in both groups were surveyed and tested in the fall and spring with the Metropolitan Achievement Test, the Piers-Harris Self-concept Scale, and measures of internal/external social power. Teacher's evaluative ratings were obtained and parents were interviewed in both time periods. After one year, experimental subjects had significantly higher reading and math scores and improved self-concepts. Regression analysis indicated higher achievement scores were due to self-concept gains, a result of the interaction of positive family and school expectations. (Author/NQ)