Benford's Law for Phone Numbers

[Rohini Shastri]

I understand that Benford's law aren't used for "non-natural" numbers such as telephone numbers because they communicate systematic information.

However in a number such as this

888-999-4346
where 888 conveys info about the province, and 999 about the city, can the last 4 digits be used to detect fake telephone number entries?

[wbuchanan]

If you are talking about US phone numbers they work slightly differently:

area code/toll free/toll code - exchange code - customer/receiver

exchange codes typically identify regions within the area code region. The area codes are only mutually exclusive based on the age of the area code and phone number generation date; for example when a new area code is added to a state it will split an existing area code and new phone numbers will receive the new area code while existing numbers would retain the previous area code. Exchange codes are also used to denote cellular service providers; in Rhode Island the 499 exchange code originally denoted a T-Mobile account, but subscribers can transfer their phone number for cell providers (my 499 number started with T Mobile and I’ve transferred service to Verizon and then AT&T since then).

[Rohini Shastri]

Yes that's correct.
I think my question will be: Is there anyway to know with phone numbers that there may be fake data entry from surveyors

[wbuchanan]

The only thing I could think of would be constraints on the number of responses per number.

[Nick Cox]

This is perhaps too obvious to spell out, but I think you need to know much more about real phone numbers before you can start to guess fairly at what is fake. The assumption that all possible 4 digit numbers are equally likely seems especially strong. I'd guess wildly that for many, perhaps most, exchanges the last 4 digits are not filled with all the possible cases 0000 to 9999.

[Carole J. Wilson]

See Robert Picard’s response to a recent thread: https://www.statalist.org/forums/for...tterns-in-data