The double-blind audio data (e.g. codec) tests conducted by researchers and sound experts will be rapid comparisons of the same short segment. Typically you will pick a few seconds out of a 30 second track, and rapidly switch back and forth in order to easily detect differences.
When we compare gear, we often do the same. 10-30 seconds of the same segment of music, repeated back and forth as quickly as possible between the two pieces of gear. It's the most reliable way to identify real differences.
In any event, you can only do this for so long before you need to take a break. Unless the differences are so large and obvious, it can take a lot of concentration to go through this and people get tired.
When we compare gear, we often do the same. 10-30 seconds of the same segment of music, repeated back and forth as quickly as possible between the two pieces of gear. It's the most reliable way to identify real differences.
In any event, you can only do this for so long before you need to take a break. Unless the differences are so large and obvious, it can take a lot of concentration to go through this and people get tired.