In determining the appropriate weight or strength of forensic evidence, one must be aware of both the associative and discriminative power of that evidence. While much of traditional forensic science research has focused on best practices for establishing positive associations, for many forensic tools, the power to discriminate and uniquely identify individual sources is largely unknown and sometimes highly disputed. In statistics, association is a measure of the connection between two variables, more general than correlation. For example, exposure to a chemical could be associated with the occurrence of a type of cancer in that the disease occurs in a greater proportion of individuals exposed to the chemical than to those with no such exposure. Sensitivity and specificity are related to association and discrimination. Tests with high sensitivity must utilize variables that have a strong association with the positive outcomes.