The Tufte count

One of the things I picked up from Tufte is the horrible habit of counting the amount of data on a chart.  This is part of the info gathering to estimate the data-ink ratio (amount of data divided by the amount of ink used to depict them).

Leon B, a reader, left this in my inbox, months ago it turned out.  This is the British government's way of informing people how energy-efficient their homes are.  As Leon said:

these charts might be a great example of governments going overboard with colours, bars, letters and numbers and lines for something that really only has four data points.

In addition, I find the use of two different scales to be confusing and unnecessary.  If it is decided that scores in a particular range can be grouped as A, B, ..., G, then the original scale should be discarded.  52 is E and 70 is C.  (This is especially so since the score ranges are not intuitive, like 69-80 = C ?!)

Even worse, what's the point of citing the 0-100 scale without explaining what is the metric?

A table presentation does a far better job in a fraction of the space:

Source: Home Information Pack, UK Government.  Graph from Wikipedia.

 

PS. This post set off a torrent of emotions (see the comments).  Another version that I discarded was the simplest table possible.  In my view, there is still way too much distracting "junk" in the original design.  No one has yet explained why the 0-100 scale should be emphasized, or what it means!

Comments

[Ken]

This is an example where a graph is wasted, all the information is in the ratings, and plotting them adds so little it is pointless.

[Barthox]

These charts are not just used for houses in the UK.

Here in Belgium, I've seen them on cars (principally) at dealerships.

And I believe I've seen them for house appliances,etc ...

I think that it comes from an EU directive related to energy saving, that must have set the way the chart looks like. (there is a mention of that EU directive on your pictures)

[Dan H]

As Barthox says, these charts are an EU design. They're most often seen on household appliances such as fridges, where they have to be displayed so consumers can make an informed decision before they buy. Most of them don't have "current" and "potential", don't have 100-point scales, and give the 'raw' data as figures in little boxes beneath the main chart (as units/day, IIRC). The design may not be efficient, but it is very clear even to lay-persons. If they instead used your little table, the average householder would have no idea what information was being given to them. Old ladies would start fishing for their spectacles. As things are, the chart isn't only immediately obvious, it also supplies a vocabulary by which manufacturers can advertise their "A rated" fridges and have consumers know what they are on about.

The scheme has, in fact, been tremendously successful in promoting energy efficiency. Almost all new appliances are A or B rated now, and I imagine there will soon be talk of introducing an A* grade or moving the whole scale. By putting the energy efficiency where everyone can see and understand it, they've given manufacturers a reason to make their goods energy-efficient, and given consumers a simple tick-the-boxes factor to consider when shopping.

[John S.]

In the U.S. appliances have labels with a high data-to-ink ratio, but I'm not sure they're any better. Here is an example:

http://www.dom.com/customer/efficiency/bus/images/guide.gif

The stickers are essentialy displaying four pieces of data. The EU stickers may go a bit overboard with colors, but I do think the idea of giving letter grades is a good one, and much easier for the layman to understand.

[Joe Harris]

As a Tufte fan I almost always agree with you, and the little ticks with numbers are very much out of place.

In this case though the point of this graphic generally is force manufacturers (or home sellers) to display a factual, comparable indication of the product's energy efficiency.

The letter grades are designed to evoke an emotional response, rather than a rational one, and they have been very successful in this respect.

I personally would like to see something similar displayed on the windows and advertising for cars. Might make it a bit harder to sex up those SUV ads.

Joe

{BTW Love the blog, keep it up!}

[Mrweatherbee]

I understand the US Energyguide sticker, but I don't quite get how this EU label works for appliances and the like. How is "potential" determined, and is it meant to be interpreted as "another manufacturer can make a similar device this much more efficienct, so therefore this product could be made the same"? With the potential clearly falling well below the "A" grade, I don't quite understand how the "A" grade is determined if it is indeed an unattainable ideal? I agree with the general comment that there is more data here than necessary, and I would argue that the data is not only excessive but also contradictory (or at least unclear).

[Robert Kosara]

I agree with what the others have said about this not being about efficiency. The original shows that there is a lot of room for improvement, but your redesign doesn't show me that - at least not in an emotional way. Also, this is not about cramming as much information into as little space as possible, but about slapping a large sticker onto each appliance (or SUV ;) that people pay attention to when they make a purchase decision. So in that sense I think that this design works very well.

[Dan]

The criticisms of both the original and the redesign are valid. The solution here is to either add data to the original so that it means something or omit the chart entirely.

Drawing a turd with less ink doesn't make it less of a turd.

[Bernard Lebelle]

The notion of "current" and "potential" seems to be a British subtlety. In France, you would only see the Grade from A to G (assembled as coloured bar lines, A being the smallest and G the largest) with a sticker (kind of left pointing arrow that would be black & white repeating the Grade’s letter). Additional information can be obtained by looking below the chart for quantitative information.

Though it can be look as an overkill (the colouring from green to red, the growing bar lines), the aim of this graphic design is not to reach Tufte’s data-ink ratio excellence level rather but rather providing a common screening grid for “the masses” on home appliances energy consumption... this graphic design also work for people with bad eye-sight :)

[jokjus]

Curious note: There is a fundamental difference between how Americans and Europeans compare energy efficiency. US energy sticker is based on comparison to other products, ie. COMPETITION and EU design is based on standard, ie. GOVERNMENT REGULATION.
American model motivates producers to match the competition, but European model sets standard that producers should reach. Both has their benefits, maybe we could combine these two approaches?

[Jens]

Two points:

For our American friends: This is, for example, how all appliances are usually labelled: http://www.sei.ie/uploadedfiles/Education/appliance_label_new.gif

In general: I actually disagree with the criticism of the label. Let me explain why. I think that a label has only a few things it must achieve. First, the reader should be able to understand it without any additional information. Second, the lable must be really easy to understand. Third, this process should be very fast.

So, I look at this label and I just get it. Within seconds. I don't understand why the bottom one has a different colour coding from the top one. Yes, the numbers look strange, but to be frankly honest, they are so small, I had to fetch my glasses to be able to see them... So, yes, they should be removed since they are too small and lend nothing to the chart. BUT:

Apart from that, I actually think this is a label that actually works. I understand this house is sort of expensive to run now, could be run cheaper, but will never be brilliant. I also understand that it is quite bad for the environment, but could be brought up to a decent level.

I understand none of that with any of the revised labels.

[TH]

The current status vs. potential is something that is specifically used in the charts for houses. This is because one purpose of this is to encourage people to fix their houses to be more energy-efficient.

I do agree with the original post that there is very little data in these graphs, and for the purpose of houses or even cars these more elegant designs would indeed be preferred.

I think the data to ink -ratio as a guiding principle somewhat fails in the context of house appliances, however, just as jokjus pointed out. For anyone who finds this information relevant, shopping for a new fridge quickly demonstrates the benefits of the redundant design - just a quick glance around in the salesroom points gives you an idea which appliances are energy-hogs and which ones are efficient, as these ratings are stickered to the front of every appliance.

The arbitrary scale steps is another matter, but as this is based on an EU-directive it didn't surprise me at all. :-)

[Axel]

As far as I know, the "EU Label" originally was a general label for a variety of mostly household aids, ranging from light bulbs to ovens, washing machines or ice boxes. Now, also home energy performance ratings charts are mandatory and they use the same logic with the seven categories ("colored arrows"). For a detailled description of the various elements of the EU label for washing machines in English see http://tinyurl.com/2hj4qu

It's all part of EU's long-ranging programs like SAVE or the Energy Action Plan For Energy Efficiency or the European Climate Change Programme, promoting energy efficiency and encouraging energy-saving behaviour to cut carbon dioxide emissions by at least 20% from 1990 levels, by 2020.

For ordinary customers, the main information is the character from A to G and the colors of the arrows. These seven categories are always used, whereas A stands for most and G for least energy efficient. How the range for every category is determined is a bit tricky. In short, it's the result of political negotiations with companies and interest groups and international standardizations. That's the reason why the metric is always different.

Because the aim of the label is to give companies and consumers incentives to produce or buy more efficient products it's based on the absolute consumption and not the relative performance in comparison with other products.

As an example, in the case of washing machines, a standard industry test is used to estimate the energy consumption (cotton, 60 degrees Celsius). A consumption of equal or less than 0.19 kilowatt-hour per kilogram washings qualifies for an "A", 0.19 < consumption <= 0.23 for a "B" etc. More than 0.39 is the criterion for a "G".

[jokjus]

What I have noticed lately here in Finland when browsing fridges and ovens is that most of them -if not all - are nowadays rated as A. And here's the pitfall of EU-system. Why should companies bother to improve their energy efficiency if they already get the highest rating? From this point of view the American system of comparing existing range of products with each other is better in challenging companies to push the envelope.

[Axel]

joktus,

it depends. "Classical" light bulbs are typically rated "E". And since July 2004 the existing energy efficiency rating category of A for household fridges and freezer appliances has been divided into 3 new categories (A, A+ and A++)...

[Timo Elliott]

(1) The graphic is destined to be an instantly-recognized consumer logo first, and a data-indicator second, so the information-to-ink ratio isn't applicable...

(2) A big problem I have with Tufte and other information purists is that they ignore the importance of "selling" the info. Here in the real world, "superfluous" ink may slightly obscure the information, but if it results in the person actually looking at the chart when they otherwise would skip over it, the net information transmission goes up.

BI Questions Blog

[backSLIDER]

When I was a kid my father would sometimes ask me to do something I didn't want to do too much. Being an uppity little brat I would so that I didn't want to or not move or something like that. One of his little tricks to get me to brush my teeth or clean up something was to say "I'll time you". So Maybe the chart is crap. It doesn't matter. It isn't trying to tell you anything. You don't need to have your toaster and your blender be the same respective relative efficiency. So give people a little gold star and help then do what they logically want to do but forget about. I hate when my father did that...

[Mr. Harrison]

Okay, this thread has been dormant for a while, so forgive me for reviving it. But I just recently discovered this, so here goes.

The reason the standard Tufte data-ink analysis doesn't seem to tell the whole story in this case is that each of these graphics gives you one data point (or one small set of data points, depending on how you view it) in a larger picture that you create as you travel through time and space. The assumption behind a lot of ET's work is that the goal is to create a single visualization that conveys a big picture. In this case, however, the goal is not to draw a graph that illustrates the energy efficiency of all homes and appliances; it's to convey something about the one you're looking at. The design is admittedly over the top for such a small amount of data, but, as noted above, it serves the purpose of facilitating recognition. It's not only more visible, but you instantly know that the information you see plays by the same rules as the other ones you have run across. (Whether those rules are well-conceived or useful, as some have questioned, is another issue, but not one for which the designer of the graphic should be held responsible.) The design works well to convey this information in the way it will be consumed: bit by bit, in busy and unfamiliar surroundings, by people who may not be absorbed by the task of understanding it.

[HIP Hoper]

As someone who is going through the process of buying and selling a house, I can honestly say that I have paid no attention whatsoever to the energy efficiency report in the Home Information Pack. If I like the house I am going to buy it. So my point is the design of the chart to me is irrelevant, I would just get rid of it all together.

[Mr. Simpson]

This thread has been dormant for a while again, but in reading the comments I think that the story of the graphic has been missed. Most of the comments treat the A to G as data, but it is the scale. The Tick is the data point.

The story is about the scales. From school we know letter grades, A is best, therefore G is worst, from traffic lights and game shows we know red is bad and green is good. The length of the bar shows how much energy the product would use in a relative sense.

Each of these reinforces each other. So - A Green Smallest Energy used = Best, G Red Largest Energy used = Worst.

The tick mark is the data point it provides the graphic position on the A to G Scale. It has a data label to provide a numerical value to compare other products if the values are closer - a finer grain comparison. The actual value does not matter as much (it is from 0 to 100, and is a score), but rather as compared to another product. The scale of higher = better is also reinforced by the vertical position of the tick.

The second chart uses the same story but changes the color scale from Grey to Light Blue - and while the gradation change may be less than ideal, from reading the chart title we can see that the scale reflects the CO2 generated and the grey being a dirty sky and the light blue being a clear sky.

The problem with chart analysis is that this is really one data point as Mr. Harrison notes.

The problem with ink to data analysis and the tables proposed is that they lose information.

The table presented loses the context of how much energy the appliance / home uses (relative) and how much it contributes to green house gases. As well it loses the fine grain information of the number - for comparison to other products. I.e. two products might have the same letter grade, but have different numerical values.

As for the potential column - that escapes me with out research into how they calculate it.