Running it in different cities skews the results - would have been better to run multiple trials in the same city. I'm also not sure that counting applicants a

Running it in different cities skews the results - would have been better to run multiple trials in the same city. I'm also not sure that counting applicants as a proportion of the unemployed in a city makes sense. I'm not unemployed, but I am looking for a new job. I also technically don't live in the city/district I work in. That's not to say that those techniques won't work as filters, but the percentages quoted are basically just made-up numbers.