By creating and analyzing matched samples, researchers can simplify their analyses to include fewer covariate variables, relying less on model assumptions, and thus generating results that may be easier to report and interpret. When two groups essentially "look" the same, it is easier to explore their differences and make comparisons based on group membership. The optimization approach, which is presented in this report, was born out of a need to produce matched samples of students from two groups: one who had participated in a particular advanced high school course or set of courses, and one who had not. Propensity score matching produced matched samples of students who not only were unbalanced in terms of student sex, race/ethnicity, and parental education levels but also were significantly different on average test scores. Therefore, an approach to match students one-to-one where sex, race/ethnicity, and parental education levels were identical and test scores were very close (if not equal) was developed in an attempt to create, by force, a balanced sample of students. This report describes the exact matching approach developed, related data considerations, preparation and requirements, and expansion beyond two samples. SAS code with annotations to be used in future research is also provided. SAS Code for Two Sample Match is appended.