Why are phone numbers modeled as integer or long values instead of strings or text in treehouse courses?

From previous studies I have always read or been advised that unless very tight validation is used, or unless there is some mathematical function that needs to be carried out, phone numbers should be modeled as strings or text. I cannot think of any cases where I would need to perform simple math or aggregate functions on one or more phone numbers. Any examination for substrings seems more practical on text anyway. I think this is also more practical for users, as single field phone numbers are often entered with special characters, such as .-,()+ etc.

There are cases where I could see phone numbers being modeled into multiple fields. For example, the North American Numbering Plan has 3 parts: NPA, NXX, and subscriber number. For example, in the number (555) 555-7788, the first 555 is the NPA, the second 555 is the NXX and the 7788 is the subscriber number. There are definite cases where it is valuable to evaluate these separately, especially in the telecommunications industry. However, there is little numeric "meaning" behind these numbers—the values would likely be used for ordering and filtering—and I can't think of any cases where they would need to be anything but strings.

Chris Ramacciotti & Andrew Chalkley , any insight? Just want to make sure I am not missing something.