It has been suggested that breastfeeding may have seemingly protective effects on children’s behavioral development. However, existing methodological problems may be resulting in an overestimation of these effects. For example, there remains large debate as to whether or not breastfeeding is causally related to behavioral outcomes given the ethical concerns with randomization into breastfeeding groups and the subsequent reliance on observational studies. Without randomization, statistical equivalence between groups cannot be assumed. This raises the possibility that any benefits of breastfeeding on children’s behavioral outcomes within observational studies may in part be attributable to selection bias. This case study involves an in-depth look at a secondary analysis of data with a national birth cohort in Chile and how the use of a quasi-experimental statistical approach (i.e., propensity score matching) helps to remove some of the selection bias inherent in observational studies—getting us one step closer to addressing the issue of casual paths between breastfeeding and children’s behavioral problems. Practical issues with the estimation of models and matching algorithms when using propensity score matching are discussed. Some of the methodological advantages and disadvantages regarding the use of secondary data analysis with cohort studies are also addressed.