The story Adam has at beernews.org that Belgian brewery Duvel will produce Brewery Ommegang beers — otherwise made in Cooperstown, New York — may seem familiar. Duvel briefly brewed Ommegang beers five years ago when demand outstripped capacity.
Back then I took samples of Ommegang Abbey Ale brewed both in Belgium and New York to a local brewpub and asked customers there on a Sunday afternoon — most of whom, it turned out, had never tried anything from Belgium or even inspired by Belgian beers — if they could tell the difference. I repeated the experiment twice more, including with friends as interested in wine as beer, then hard-core beer savvy types who viewed the test as a challenge.
I’ve added the story that resulted to the library.
***SPOILER ALERT-SPOILER ALERT*** For those who don’t want to bother with the click the results are summarized below.
******
******
******
******
******
Quick summary of the results: Just over half (that’s chance) correctly picked out the odd beer in a “triangle tasting,” and 57 percent preferred the Cooperstown version.
My only real contribution to this branch of science was this photo from the time of the two Ommegangs. Any chance that the travel alone was the critical factor?
Do you mean 1/2 or 1/3 picked out the odd beer?
1/2 picked out the odd beer. Then when told which was which I asked which they preferred.
Alan, I’m guessing travel wasn’t a big factor but time might have been. The samples in New York were actually a little older. I bought them when I heard what the plan was, cellared them (about 52F), and hauled them out when the first samples of the Belgian version reached the states.
I’m embarrassed to say I’ve never heard of this triangle tasting. It makes the old researcher in me wonder a few things. Mainly, it makes me wonder if you always get about 50% for beers that look the same. Obviously, if asked to distinguish the kriek from the straight lambics, tasters might score higher. I’m recalling that study recently that showed that people had essentially zero sense of wine. Five bucks or five hundred it was all the same to them.
It encourages me to want to do some of this triangle tasting, and soon.
But, then again, maybe I’d prefer to just rest in the illusion that I actually know what the hell I’m talking about.
Jeff – Rest assured, trained tasters sitting on daily panels at commercial breweries do a lot better than 50%. And that includes toughies, like discerning between a brand bottled 3 days ago and the same one bottled yesterday.
Interesting, Stan, but I’m trying to figure out how interesting. If you know a friendly statistician, I’d love to hear how statistically relevant your results might be.
Given the sample size, the settings, etc. I wouldn’t say you’d be submitting a paper based on my results as an academic paper.
However, to Jeff’s question, you put two beers from Newark and one from Houston in front of a tasting panel at A-B and the panelists can’t pick out the outlier then everybody is happy.
My point was that you would expect 1/3 to guess correctly by chance, so 50% is better than chance (but as you note, with a small sample, it might not be significantly better than chance).
Stan, I’m sure they’re testing freshness–but based on my own experiences with brewery-fresh beer, I bet this isn’t all THAT tough. Freshness is such a huge factor in the way most beers taste. There are a few exceptions, but probably fewer than craft drinkers believe.
Stephen, I’d say a sample size of three falls below significance (by at least an order of magnitude). On the other hand, the experiment produced results far more significant than, say, one blogger writing a post about his own experiences. I’m convinced!
NF – Sorry, I’m probably not doing a good job of explaining how a triangle tasting works.
You get three glasses (better if they are opaque so you aren’t influenced by appearance), two of which contain the same beer, the other something different (maybe a little, maybe a lot0. You try to pick out the one that is different. Chance on that is 50%.
A fine parlor game.
So, this really isn’t a big deal… but the point I am trying to make is that the chance of picking out the one that is different in a triangle test is 1/3. There are three beers. One is the odd one out. If you choose at random, the chance of picking the odd one is 1 in 3. So it sounds like your punters were doing slightly better than chance.
+1 NF
In this scenario “chance” will be to pick one right out of three unknown = 1/3 chance. A result of 50% (1/2) indicates a significant difference between the beers (if the number of testers are large enough).
Stan, I’m afraid you have to think this over again…. 🙂
(or check http://en.wikipedia.org/wiki/Discrimination_testing )
You learn something new every day. It appears the way conducted the test was the same as a duo-trio.