Parameters and procedures for developing an error-prone model (EPM) to predict financial aid applicants who are likely to misreport on Basic Educational Opportunity Grant (BEOG) applications are introduced. Specifications to adapt these general parameters to secondary data analysis of the Validation, Edits, and Applications Processing Systems (Task 1) are also provided. After briefly discussing three approaches to error-prone modeling that are used at the state and federal levels, four types of validation applicants are identified: exact reporters, overclaimers, underclaimers, and validation dropouts. The research method, automatic interaction detection (AID), is described, and two ways to report the results of an AID analysis (a tree diagram and tabular format) are illustrated. Additional methodological issues are covered: the three study samples (working sample, replication sample, and nonvalidation sample); selecting and coding predictor variables; and the effectiveness of the model as indicated by a contingency coefficient. Flow charts illustrate the steps of the study, as well as the classification of applicants into validation types. Also outlined is a strategy for analyzing the quality control sample, which is derived from the 1978-1979 applicant file. Appendices detail the statistical analysis method. (SW)