In 2006, Harcourt found the evidence supporting the broken windows theory might be flawed. He reviewed the study Kelling had conducted in 2001, and found the areas that saw the largest number of misdemeanor arrests also had the biggest drops in violent crime.