It probably came from comparing accept percentage of Harvard (6% now but 7% a couple years ago) with a kidnap chance calculated for the whole of children born in the US. Of course applicants are already a selected subset and not representative of all candidates born in the same year. It's a pretty useless comparison.
e.g. 40 kidnaps per year for 10 years = 400. 400 * 700 'times more likely' = 280,000. Divide that by 4mil children born in a year to get 7%.
e.g. 40 kidnaps per year for 10 years = 400. 400 * 700 'times more likely' = 280,000. Divide that by 4mil children born in a year to get 7%.