You're misunderstanding the methodology. They tested a sample and multiplied to get the overall infection rate, under the assumption that their sample was representative.