A language is not a single, measurable entity which exists outside of the populations of its speakers. All that exists is the mental representation of that language, however you wish to model it, in the minds of the speakers. Happily, this means that no native speaker is “using” the language “wrong,” since their usage literally defines that language.
External prescriptions which are placed on language usage are literally independent of the language itself. They do not reflect facts about the language. They only reflect judgments about various usages, which in some sense reflect social perceptions of variants which may help to reveal your educational, regional, racial or class background. These are not facts about language, but simply about how language becomes a participant in social dynamics.
My favourite example of the sheer arbitrariness of language value judgements is rhoticity, or whether you pronounce an /r/ after a vowel in words like “car” and “yard”. If you’re British, the most prestigious dialect is RP, which is non-rhotic, and it’s less prestigious to speak a rhotic dialect, such as Lancashire and other Northern English dialects. If you’re American, it’s the opposite: the most prestigious dialect is SAE, which is rhotic, and non-rhotic dialects such as those in New York or Boston are less prestigious.
It’s the exact same contrast between /r/ and lack of /r/, but when you cross a large body of water suddenly the social prestige associated with it is inverted. You can’t pretend that language value judgments are “objective facts” about the language.
See also: More specific social commentary about literacy privilege.