As will be seen below, a critical question raised by this method is the estimation of error scores (in the event of false matches between individuals who by coincidence have the same code) and error estimation of the resulting population estimate. We note that these questions would be greatly simplified by attaining a code for more phone number digits. In our case, however,
pre-testing found that asking for more than 3 digits raised suspicion among our research subjects and equally importantly, questions about the assurance of anonymity by our Institutional
Review Board. Given these concerns, a method capable of producing and bounding an estimate within a range of confidence estimates seems particularly important.