Under the first sense for heterogenous, the OED writes:
A less correct form of heterogeneous adj. (In mod. use prob. repr. the pronunc. /hɛtəˈrɒdʒɪnəs/ given by some speakers to heterogeneous: cf. homogenous adj. (homogeneous adj. ¶).)
Labelling the word as “less correct” suggest a proscriptive approach to usage. Checking the British National Corpus produced 13 instances of heterogenous compared to 278 of the allegedly better form heterogeneous. A similar distribution is found by doing a full-text search of the OED itself: 11 tokens of heterogenous and 206 of heterogeneous. So we clearly have to do with variant spellings here.
The OED revision process started at M, and hasn’t got to H yet, so it’ll be interesting to see what will happen to this word when it is revised.
Any other examples of pre- or proscriptivism in the OED?
Bridging the Unbridgeable 
How many of the 11 tokens of “heterogenous” in the OED, and the 13 in the BNC, can be attributed simply to mis-typing? The 1st volume of the OED 2nd edition ends with the spelling “bazouki”, which is described as a ‘mis-spelling’ of “bouzouki”. The online version of OED doesn’t include the spelling as a headword. Again, is this because it is simply a typo?
Hmm, yes, I see your point. Interesting how the word got removed for the online version. One that didn’t is “mamalone”, which is marked as “Probable editorial misreading of mamalouc”. It was already corrected in the 1995 edition of Jane Austen’s letters (and a fourth edition of the letters came out in 2011). Furthermore, the Jane Austen mis-quotation is the only instance, so why keep it in if it is an error that has already been corrected?