The only bias would be customers having problems with their vehicles or not liking certain aspects of it. There is no weighting or tweaking, it's problems reported per thousand. The people are selected at random, it's a standardized survey form, and there needs to be at least (I think) 100 surveys return to evaluate the product.
It's not some pay to play or pay to win scheme. The OEMs only pay to see the data, and if they win an award they pay to be able to "display" the award in their adds.