That was annoying me the entire time I was reading this, the entire article has an unspoken assumption that their interview process couldn't possibly be responsible for any discrepancy in the data.