The best ideas often come from unexpected places.
"In observational studies, we often have to rely on assumptions that are not directly testable."