Brix refractometer == Atago

#1: Post by **bbobbert3** » February 13th, 2020, 3:52 pm

I haven't taken data on a VST refractometer. From what I understand, it's similar in performance as an Atago. I couldn't find a statistical difference between the cheap Brix refractometer and an Atago for measuring TDS for espresso.

#2: Post by **Marcelnl** » February 13th, 2020, 4:27 pm

I would not expect a difference given that refraction is the same regardless of the manufacturers name on the device, yet without a description of the methods used to be able to compare the two the graps in my view are meaningless.

I mean, nothing against what you did, but I would also not be able to find a statistically meaningful difference between a car and a bike measuring TDS simply because my working knowledge of stats is ; call a statistician.

#3: Post by **jpender** » February 13th, 2020, 6:49 pm

Looks like the "cheap Brix" reads higher on average relative to the Atago.

But even if that weren't true... so what? It says nothing about the precision. I could make the same comparison between my 1g resolution scale and my 0.01g resolution scale.

#4: Post by **Mosjev** » February 18th, 2020, 10:35 pm

I love that you did this. Looks largely linear to me with perhaps a slight bias at higher TDS.

What is the hypothesis here? To me it seems like you are asking: Is the cheap Brix refractometer as accurate as an Atago?

If the answer is yes, then no reason to spend the extra money. If the answer is no (who could do a t-test if you so choose) then I have to decide whether the increase in accuracy is worth it.

From the plot you show here, it appears there is a slight difference between the two machines. However, we don't know which is correct (or at least closer to being correct).

What I'd really love to see is the accuracy of a serial dilution. Take a super high TDS sample and split it in half. Use one half to measure and dilute the other half by 2x. Repeat, repeat, repeat.... Then repeat the whole process again 2 more times so you have 3 independent dilutions. Probably 8 samples at this 2 fold serial dilution will span the dynamic range of both assays. If you then plot the measured TDS against the log2 of your dilution factor you should get a straight line. The degree to which the individual points deviate from this line and the degree to which the slope dedicated from 1 will tell you which assays is more accurate.