I have access to the original article. If anyone else has a subscription to PNAS you can see it here:
http://www.pnas.org/content/early/2012/01/02/1114999109.abstract
Otherwise you'll be limited to the abstract. I won't be able to post the article, as I'm pretty sure that violates many, many rules. Discussing it probably isn't a problem though.
The study was conducted at the 8th International Violin Competition of Indianapolis (which is apparently a big deal. Personal disclaimer being I haven't touched a violin since the 3rd grade so I know little of big deals in the violin world). They had 21 participants, most of whom were participating in the competition either as competitors, adjudicators, or symphony members. The overwhelming majority had advanced degrees (masters+) in music.
There were 6 violins that the participants were asked to review. 3 of them were considered "new" (between several days to several years old). The other 3 were considered "old" (2 Stradiveri, one from his 'golden age' and 1 Guarneri). The combined value of these 3 violins was 10 million dollars, approximately 100 times that of the new violins. Participants were visually and aromatically blinded (goggles and scented chin rests). Researchers were also blinded, making this a classical double blind experiment.
For their first test, they do a number of interesting metrics. Most notably they asked the participants to rate playability, projection, tone color, and response. Then they also asked which instrument they would prefer to take home. Throughout this entire evaluation period, the participants had access to all 6 violins, so they could compare between them.
The ratings were pretty much everywhere. The figure that they present doesn't seem to indicate much by way of trends for the most part, or at least anything the participants consistently agreed on. The real outliers in this was one new violin, which was consistently rated better than the rest in all categories, and the old stradiveri, which was consistently rated worse than the rest in all categories except projection. The worst projection rating was given to a new violin.
In the second evaluation they paired and old and a new violin, and gave each participant one minute to play each one of a pair. They would then choose one violin that they preferred.
As for results in the second evaluation, participants pretty much performed at chance, neither preferring any set (old or new) of violins over another. Except for one: the older Stradiveri was preferred consistently less when compared to any of the new violins (this effect was huge. 15-18/21 preferred the new to the old).
There was another figure that I didn't discuss, mainly because I don't think it adds much besides another way to look at the data from the first evaluation. If anyone has any other questions, I'll be happy to look at the paper again.
This was a cool link OP. Thanks for sharing. I'll let you guys make your own conclusions, but to me, this looks like a fairly well-done, controlled study.