In an effort to nearly replicate Bidwell and Kasarda (1975) and to analyze the efficacy of an adapted version of their model to predict high school retention, all 66 school districts in Louisiana were surveyed to determine how many freshman students in 1971-72 graduated in 1974-75 and why. An alternative model was devised which differed from the Bidwell-Kasarda model in that it included some new exogeneous variables, a new dependent variable (retention instead of achievement), and a reordering of the causal sequence within the model. Variables were categorized as follows: environmental conditions (school district size, fiscal resources, disadvantaged students, parental education, percent nonwhite, and percent rural); structural conditions (pupil-teacher ratio, administrative intensity, and professional support component); staff composition conditions (certified staff qualifications); and dependent variable (retention). Results indicated the model worked well in accounting for variance in two of the intervening variables, pupil-teacher ratio and administrative intensity, but for the other two intervening variables, professional staff and qualifications of teachers, and for the dependent variable, retention, the model accounted for only a small amount of the variation. It was concluded that neither this model nor the Bidwell-Kasarda model accounted for much variation--21% and 25%, respectively. (JC)