Well, over the years, I ran 2 sets of Internet "blind" tests on this.
Back in Feb 2013, I did high-bitrate MP3 vs. FLAC - no significant preference among 151 respondents.
June 2014, comparison of 24/96 vs. straightforward dithered 16/96 - 140 respondents again no significant preference.
These comparisons are all done in the tester's homes using their own equipment, with the test "advertised" on audiophile websites over 2 months. Music files used were of good dynamic range and "true" 24-bit source (for the 16-bit vs. 24-bit test). Furthermore, I looked at age stratification and did not see evidence of ability to identify a difference.
Despite all the subjective claims, there really has not been good evidence than perhaps other than in a lab environment with specialized test tones, that in "real life", there is a significant difference between high-bitrate MP3 encoding and the same source 16/44 lossless. Nor is there evidence that 24-bit imparts any special quality over a reasonably dithered 16-bit version of the same thing.
(Tests and results on my blog: archimago.blogspot.ca)