I know this is shocking to people but if a phrase is systematically used by native speakers, it is then part of the language. There is no notion of native speakers being systematically wrong in linguistics. It wouldn't make sense scientifically.
In order to examine natural languages using the scientific method, linguists gather data (i.e. native speakers' spoken or written communication) and then analyze this (i.e. find predictive models of this data). Gathering data, then claiming the data is wrong is epistemologically unfounded. Languages simply are the way they are. This would be like gathering data from Hubble and then deciding photons are wrong because their behavior mismatch with Newtonian laws.
In order to examine natural languages using the scientific method, linguists gather data (i.e. native speakers' spoken or written communication) and then analyze this (i.e. find predictive models of this data). Gathering data, then claiming the data is wrong is epistemologically unfounded. Languages simply are the way they are. This would be like gathering data from Hubble and then deciding photons are wrong because their behavior mismatch with Newtonian laws.