Nonresponse bias in mail accounting surveys: A pedagogical extension