Language standardisation has variously been described, for example by Einar Haugen, as a process involving the selection, codification, acceptance, and elaboration of a linguistic norm. I'm concerned here with the selection of a norm. Often this consists not of selecting a dialect that is already present in its entirety, but a process called koinéisation takes place - the levelling of differences between dialects.
Apart from regional varieties (dialects), social varieties (sociolects) also play a role. I think I have heard/read repeatedly that linguistic structures/words etc. used by educated speakers are more likely to become part of the emerging standard than linguistic structures used by less educated speakers. This is, I think, because educated speakers play a role as gatekeepers due to their social, economic and political power. In consequence, the varieties used by uneducated speakers are often looked down upon.
However, in the literature on language standardisation (such as Language Standardization and Language Change I could not find any reference to the importance of educated speakers in language standardisation.
Can anybody provide academic references for this? (of course I would also be happy to consider references denying the role of educated speakers in language standardisation)