I’m not disagreeing on most, but I am pointing out that no good, independent sources of data exist. If we were to reframe it as how to get accurate data on make/model reliability, such as trying to quantify their reliability in an unbiased way, we quickly discount most sources. Or, ask the reverse question- how could we show that Tesla is reliable? Again, no good sources.
NHTSA tracks safety concerns, or at least certain types of them. They don’t track things like how often ignition coils fail, or the radio, or the water pump, or the plastic trim breaking. All of these are part of reliability, and what CR at least attempts to do.
CR isn’t biased, but they are bad in other ways due to (sometimes highly) flawed methodology. They also only offer very broad info on failure data, and what they consider. I recently learned that they listed the Bolt as having excessive numbers of failures in the battery, because they counted every single recalled vehicle as a failure point. Not biased, but still bad (or at least, not very useful) data.