thoughts

disseminating coffee research(ing)

The SCAE Gold Cup Research programme, which I have discussed previously, is a timely piece of work, but also a massive pain in the arse. I applaud the SCAE for undertaking this endeavour, as it is an ungainly, painful operation to run. It has been conducted 4 or 5 times at various locations around Europe now, I’ve been involved with 2 of those, and it’s no easy task. You have limited time to dial in, sometimes with unfamiliar equipment. Not to mention the difficulty in deciding on one coffee which should represent all coffee.

The scope of the research is to determine taste preference across 5 different levels of extraction, with strength being constant. That in itself would be good to know. However, it isn’t designed to cover the entire landscape of coffee brewing. It would be utterly impractical to do so, given the limitations of resources and time afforded to the study.

What if there was an easier way?

This kind of study needs several things, first off data, lots of it. The more data acquired the more powerful the results. It needs sensory evaluation of coffee made to a known recipe and of a known strength (extraction can thus be inferred).

There are thousands of users of VST refractometers around the world. Most probably brew coffee everyday. If each of these users submitted only one measurement to this study we would have a significant, useable sample size. If some of these users submitted multiple measurements over the course of a few months, we could have a huge, powerful sample size.

So, here’s what I am proposing. Anybody with a refractometer can (and should) submit data to this study.

Using a simple form, submit the recipe, the TDS/strength, and a score (your evaluation of the brew).

This approach has its pros and cons.

  • It gets around the need to pick one coffee. Now it is every coffee or any coffee.
  • It potentially allows for a much wider coverage of the brewing chart.
  • It could allow for a large number of samples to be acquired in a short amount of time, with minimal effort.
  • There will be inherent variability between users, in how they brew, and how they measure, and how they score. There will inevitably be some bad data. However, given a large enough sample size you would hope this would even out.

I would envisage this running for about 6 months, at which time the data would be published and freely available for anyone to use/analyse.

The following is the suggested protocol (I thank Vince Fedele for his help on this).

  • Measure (by weight) and record your ground coffee and your brew water.
  • Transfer a small sample (4-5g) from the brewed (& filtered) coffee to a cool glass/cup.
  • Draw sample into pipette/syringe, expelling as much air as possible, leaving the tip submerged in remaining sample. Do not measure at this time.
  • Evaluate coffee over your normal range of drinking temperatures, choose score based on scale below.
  • Remove pipette, discard a few drops, then transfer to sample well and measure as normal.

If anyone has thoughts on improving this experimental design please leave a comment below. The basic intention of the steps here is to allow the participant to taste the coffee without being influenced by the measurement. Therefore it is key to try to prevent erroneous readings due to evaporation over time.

The form can be found here (on this blog) – or here standalone.

I suggest iPhone users follow this guide to add a shotcut to their homescreen for frequent use.

The scale used for evaluation is a 9 point hedonic scale similar to this:

Advertisements
Standard

26 thoughts on “disseminating coffee research(ing)

  1. Michael mc Laughlin says:

    Having been involved in the actual research for three out of the five of these I can certainly see how this would help, getting dialled in and keeping brews consistent over the whole time can be very time consuming and frustrating, great idea

  2. Pingback: Dispersed Discovery « MARCO ÜBER PROJECT

  3. Mike Mierz says:

    This is very cool and I really wish I had a refractometer so I could participate in this. In the mean time is there anyway to see the chart as people are entering in data?

    Thanks for taking the time to put this together, best.

    • Hi Mike – I’m not 100% opposed to some sort of live chart – I would say it is doable. My only concern would be that it might influence participants…?

      Thoughts?

      • Joona Suominen says:

        I think it’s best to only release the data after the test is done.

        I believe if the averages are known the risk of experimenter’s bias increases. There might also be some social favorability connected to certain values.

        Keeping the data hidden during the test controls this variable quite well and as far as I know it’s also the standard test procedure in social sciences.

  4. Rob Smyth says:

    Like Michael I’ve been actively involved with this on 3 of the 5 occasions & would love to see something useful come from all the effort that’s been put in so far, so I applaud your attempt to drive this on with a different approach. I’ve a refractometer and measure a lot of my brews after tasting, so I’ll certainly submit readings when it’s practical to do so.

    My immediate concern is we’re now limiting the research to coffee folk and in particular a smaller subset of the speciality coffee industry that have refractometers. While we may see a pattern emerge within this subset, we’re excluding the broader coffee community and the general public who may have very different taste preferences to ourselves. This is unavoidable though for conducting the research in this way, but it may still throw up some interesting data.

  5. Dave Hart says:

    I in food research, and worked in the research group of a major coffee company, before coming to food research.

    It will be a good idea to get people to check their refractometers with, say a 10% sugar solution to start with so you know the instrument is reading correctly first.

    From the work I’ve been doing when (the lab is empty) on the ‘modern style’ roasts I think the ‘brew graph’ needs to be completely re-drawn. I drinking SM’s amazing Kenyan Tegu a while back, and when I did the maths it was more than 2% solids, with no hint of it being too strong.

    @Mike Miertz: I’ve been messing around a bit here and have had a sixth form kid in doing a project looking at this, and we’ve been getting some pretty good results using a Bellingham and Stanley e-line hand-held refractometer. It costs about £40, and with an excel spreadsheet that the kid devised comparing RI against solids determined gravimetriclly you can calculate all the same numbers that MoJo gives you, for a tenth of the cost.

    • Hi Dave – I think that is wise. Though I imagine a reference solution closer to the target range ie 1-2% might be better? VST do sell these, though the price might be off-putting for some.

    • AndyS says:

      Dave: If you want to use one of those handheld analog refractometers to determine people’s preference for coffee strength, you could use one and get some marginally accurate data. But to go the next step (as David intends) and calculate extraction yield, those cheap e-line refractometers won’t do the job. For yield calculations you want an instrument with resolution of 0.00001, accuracy of 0.00005 or better, and good precision. The Bellingham and Stanley e-line refracts are nowhere near that — they’re at least ten or twenty times worse.

  6. Chris Capell says:

    Just a thought – I tend to weigh dose and final beverage instead of dose and brew water. I believe this reduces issues of variance of water retained by grounds, or by varying brew methods. Would be interested to hear your thoughts on this.

    • I haven’t thought too much about it, except for one time I was testing a brew method that retained a lot of brew water, and I had to go that route. My question back at you would be – what is the variance retained by grounds?

  7. Awesome! Will be glad to pitch in.

    Funny thing, was just writing on my own blog today about how scientific research in coffee is still budding, and how we really do rely on Twitter and the Internet to get information out to people, and to retrieve information.

    Thanks for proving my point, hours later!

  8. Chris Capell says:

    My reply would be, I don’t know. 🙂 That’s why I worry about calculating yield by using a constant in the equation for water retained by grounds. I weigh the beverage, measure the TDS as a percentage, multiply that by beverage weIght to get the mass of solids in the cup, and then divide by the original dose to get the yield. I suppose the amount of solIds in the water retained in the grounds or withheld by the brewer could be of a significantly varying TDS and/or total weight as to invalidate my calculation, but I don’t *think* that’s usually the case. Anyway, just wanted to throw it out there as a possible way to reduce variance in the results you’re collecting, as well as to see if you had any more thoughts about it. I’m far from certain about it all.

    Regarding calibrating refractometers, I just use distilled water, which is cheap and easy to find, and follow the calibration instructions that came with the refractometer. I do it before every refracting session.

  9. Navin says:

    Cool idea. But since you are encouraging people to submit responses from their iPhones, you may want to make sure the submission form displays correctly on an iPhone web browser. (Right now, in portrait orientation, the options for “Score” are cut off so that I can only see buttons for 1 to 6 — this could certainly lead to a bias in your results!)

  10. Great idea. I suggest trying to collect as much data as possible with the forms – what about a field for ‘roast date’ (if known), to avoid errors as dislike being due to bad tasting coffee rather than brew strength. It could be useful to also collect data on coffee type, origin – e.g. a washed bourbon may taste better at a certain strength whilst a natural processed bourbon may have a different sweet spot. The more data the better. Best of luck with this project.

    • Hi Kevin, I had some thoughts along those lines. My thinking in not including such fields was keeping it as simple as possible was more likely to encourage use. The more fields, potentially the more confusing, and time consuming… That said, if people chirp in in favour of such a design I would be interested in listening to their arguments.

  11. Pingback: Random Brew Recipe Generator | the other black stuff

  12. Hey David, when I bookmarked the form on my iPhone, the formatting is a bit screwed up. Apart from that, count us in, we will get you data on the coffee we drink.

  13. Pingback: paper form for research | the other black stuff

  14. Pingback: phase 2 | the other black stuff

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s