The sophisticated analytical techniques available to institutional researchers give them an array of procedures to estimate a causal effect using observational data. But as many quantitative researchers have discovered, access to a wider selection of statistical tools does not necessarily ensure construction of a better analytical model. Moreover, institutional researchers often face real-world challenges of insufficient resources, artificial time constraints, and the pressure that comes from knowing the potential ramifications of administrative decisions based on their findings. Within this context, institutional researchers are continuously looking for effective and efficient techniques that produce unbiased results. This is particularly true when the question of interest asks whether participation in a curricular or co-curricular program or activity (for example, a freshman transition course, working during college, or a living/learning community) improves or inhibits student success. The simplest way to answer this question might be to compare the outcomes (grades, persistence, and timely graduation) between participants and nonparticipants. However, students are rarely randomly assigned to programs and instead select whether or not to participate. The researcher cannot unequivocally claim that the program effect is attributable to the program experience or to an amalgam of factors that may have influenced the student's decision to participate. A variety of statistical methods, such as structural equation modeling, regression discontinuity, and analysis of covariance, attempt to adjust for nonrandom selection to allow the researcher to accurately estimate the relationship between the "treatment" and the outcome. Increasingly, institutional researchers recommend using propensity score methods as a means to adjust for factors that influence selection into programs. In this chapter, the authors examine the advantages and disadvantages of using propensity score matching models in institutional research. Using longitudinal and pre- and posttest data, they further discuss conditions under which propensity score matching models might be useful in institutional research. (Contains 1 table.)