This comparative study of two approaches contrasts a schema-based approach to represent a solution approach to solving whole number contextual problems for grades 2 and 3 with the traditional textbook approach. The participants are 9 to 11- year-old Afghani refugee students enrolled in non-public schools administered by NGO organization in Iran. The subjects have difficulty with grade-level mathematics and have been retained in grade at least one year. Subjects were randomly selected from four classrooms in two schools. The schema-based experimental approach is called the Problem Patterns (PP) approach. Students receiving this instructional approach were taught to break problems into data, units, and desired solution, removing irrelevant information, and make a solution model with manipulatives. Control students followed the traditional classroom approach. All classes were taught by the first researcher. Evaluation results showed the PP students had higher achievement and growth scores than the control students. The results also showed the schema building portion of instruction contributed most to the differences in performance of the experimental groups' students.