This study, exploring why first grade children from Rio Grande do Sul, Brazil, fail in school, utilized computerized techniques of statistical analysis to measure the relationships of various school and family characteristics with student achievement. Four types of schools--urban state, rural state, municipal, and private--were used to test the effect of student achievement as measured by parents, supervisors, school, principal, teacher, and class, on the dependent variables of age, and repetition on pass rates and dropout rates. A model for effects on student achievement was said to include three types: school only, parents only, and the interaction of specific parent and school characteristics. The effects of school alone were found to be low for all types of schools, while 75 percent or more of the explained variance in grades in language was based directly or indirectly on measures of the socioeconomic background of the child, such as mother and father's occupation, father's educational level, and the number of textbooks. Two areas of recommendation focused on were: (1) the development of educational policies to improve student achievement by changing internal characteristics such as the distribution of books and improvement of teacher training, and (2) the need for additional research to refine and test further hypotheses arising through the study. (Author/AM)