I watch A LOT of "amateur" science stuff and I think Brülosophy is by far the best for scientific process... Absolutely superb appreciation for experimental design. Way better than a lot of multi billion dollar companies I've worked with professionally.
Is there any data on what percentage of tasters pick certain colors? Technically speaking, the primary colors are red, blue and yellow. By doing red, blue and green instead blue and green are technically closer to each other than red on the color spectrum. So red would be the most different from a visual perspective and we know that color and visual appearance sways perceived taste to some degree. I doubt you have the data but would be interesting to take a look if you did.
Triangle tests are easy to administer but not super sensitive as there are carry over effects. For example, it is typical that the first sample tested will taste stronger in flavor than the next. 2 AFC or duo trio are more powerful if memory serves me correctly, but take more effort to administer. These are neat experiments overall and I enjoy the content!
I think there's research that shows the colour of the cup affects perception of flavour? I realise it makes life harder, but if you don't mix up which colour has the different sample in it, then it might factor in? Although the amount of insignificant tests suggests it doesn't make a difference.
P < 0.05 means that there is at least 95% chance that your result is significant. You had an elegant experiment recently about beer finings: does it improve the taste? You had nearly 20 tests and 1 came back positive and the other negative. That translates to 5 % showed it is significant and 95% are non significant... or in other words p
What I love with Brulosophy is when they test a widely spread myth of brewing and it turns out it has no noticeable impact at all, even over several experiments performed and people still firmly insists it has made a huge difference for them in their brewing and certainly it must be Brulosophy doing something wrong and is not scientific at all.
A common statistical test, the t-test, was actually invented by guinness. I would also be interested in a meta-analyses to see if a specific colour of cup martin uses is more often chosen as the odd one out and if this varies with what is being looked for. I get that it looks better having different colours
So, in any test some people will pick the correct one just by accident but if 11 out of 20 picked the correct one it's less than 5% likely that these 11 picked the right one by accident? Meaning it's 95% likely they picked the correct one because they genuinely detected a difference.
Love the show and podcast learned a lot but I am surprised you use plastic cups as we know sight and aroma are part of the taste/experience. Maybe colored coasters underneath glass
Great explanation but you never talk about effect size and statistical power. Most of your experiments are low sample size. Any non-sig result could easily be a type II error. Remember absence of evidence does not mean evidence of absence. You’re only doing half the necessary stats.
True. I would suggest three cups of the same color, but labeled with three random three digit numbers. Also for proper triangle tests, you need a large set of tests, i.e. lots of people. At least a Hundred. There's going to be lots of bias in these tests as they run them.
I watch A LOT of "amateur" science stuff and I think Brülosophy is by far the best for scientific process... Absolutely superb appreciation for experimental design. Way better than a lot of multi billion dollar companies I've worked with professionally.
I absolutely LOVE this show! All aspects of it! Not a single boring video!
Love the in depth explanation. He's given it before on podcasts, but I'm glad it out here on UA-cam now.
Glad to see Marshall on this show!
Excellent summation! Thanks for presenting this.
Is there any data on what percentage of tasters pick certain colors? Technically speaking, the primary colors are red, blue and yellow. By doing red, blue and green instead blue and green are technically closer to each other than red on the color spectrum. So red would be the most different from a visual perspective and we know that color and visual appearance sways perceived taste to some degree. I doubt you have the data but would be interesting to take a look if you did.
Excellent video. Explains the science behind the 3 cup tastings.
very informative video about data collection overall!
Triangle tests are easy to administer but not super sensitive as there are carry over effects. For example, it is typical that the first sample tested will taste stronger in flavor than the next. 2 AFC or duo trio are more powerful if memory serves me correctly, but take more effort to administer. These are neat experiments overall and I enjoy the content!
Thanks for the Video! Still need to do a exBEERiments myself sometime in the future to see how hard it is for myself to find a odd beer out.
I think there's research that shows the colour of the cup affects perception of flavour? I realise it makes life harder, but if you don't mix up which colour has the different sample in it, then it might factor in? Although the amount of insignificant tests suggests it doesn't make a difference.
P < 0.05 means that there is at least 95% chance that your result is significant.
You had an elegant experiment recently about beer finings: does it improve the taste? You had nearly 20 tests and 1 came back positive and the other negative. That translates to 5 % showed it is significant and 95% are non significant... or in other words p
What I love with Brulosophy is when they test a widely spread myth of brewing and it turns out it has no noticeable impact at all, even over several experiments performed and people still firmly insists it has made a huge difference for them in their brewing and certainly it must be Brulosophy doing something wrong and is not scientific at all.
A common statistical test, the t-test, was actually invented by guinness.
I would also be interested in a meta-analyses to see if a specific colour of cup martin uses is more often chosen as the odd one out and if this varies with what is being looked for. I get that it looks better having different colours
So, in any test some people will pick the correct one just by accident but if 11 out of 20 picked the correct one it's less than 5% likely that these 11 picked the right one by accident? Meaning it's 95% likely they picked the correct one because they genuinely detected a difference.
Love the experiments, but drinking beer from a plastic cup seems wrong. Love the show 👏🏻😎
Love the show and podcast learned a lot but I am surprised you use plastic cups as we know sight and aroma are part of the taste/experience. Maybe colored coasters underneath glass
Great explanation but you never talk about effect size and statistical power. Most of your experiments are low sample size. Any non-sig result could easily be a type II error. Remember absence of evidence does not mean evidence of absence. You’re only doing half the necessary stats.
Color affects perception. So....
True. I would suggest three cups of the same color, but labeled with three random three digit numbers. Also for proper triangle tests, you need a large set of tests, i.e. lots of people. At least a Hundred. There's going to be lots of bias in these tests as they run them.