I actually think that their method may be useful. I'm in a Hyundai now because my Ford Aerostar was totalled in a head-on with another Aerostar. It was an unusual demonstration of selective crumpling. Many car accidents are between two vehicles, both of which would have hopefully crumpled to progressively absorb the impact and save the occupants. However, CU's method simulates what will happen when the car slams into a concrete wall, and as we all know, some car accidents involve one vehicle and a large, unyielding, solid object. So, I'd say that CU's test is rather crude, but quite valid.
Let me add that CU sometimes has test standards that can seem off-the-beam to a person who has experience in the particular realm of the test, and can come up with results that cause me to shake my head in disbelief. However, there are reasons, some unfortunate, for their behavior.
My own beef with them has to do with the reliability of consumer electronic products and the cost of ownership. In other words, how likely is it to fail and especially, how much will that repair cost. However, I think that they don't dare comment upon these concerns because they've already had to fight off Bose, and it was only the US Supreme Court that saved Consumers Union in the end. That's not been the only lawsuit. They've got to be completely objective in ways that will stand up in court.
For cars, I think that the repair charts are misleading. For example, a car can be shown to need lots of repairs to the electrical system. But in reality, let's say that the issue was that the heater motor would fail due to an engineer having specified the wrong fuse. Replacing the fuse (a $1 part) fixes this, but it will show up on the CR repair chart as a black dot because it happened to a lot of people. But another car that has a wiring harness that can burn out when stressed by only the cigarette lighter when used with the headlights will show up as having good electrical systems, but the repair will run $3,000.
Yes. I was pretty surprised to read their recommendation of the '04 Sonata because it's seemed that their descriptions of Hyundai cars have been so ho-hum. But on the linked page, I think that CU explained their context quite well. It seems to be a good job.
Richard