The new iPhone was unveiled this week with the usual frenzied discussion about its merits, or lack thereof. Omitted from the media coverage was any conisderation of the phone's environmental impact. Unsurprising perhaps, when there is the sexier fare of talking assistants and errr, debating the model number, but correspondants can't be excused by a lack of data to go on.

Following a run-in with Greenpeace a few years ago, Apple now publish an environmental report for every new product they launch. The reports are not particularly detailed, but they do include estimated greenhouse gas (GHG) emissions for a product over its entire life cycle (i.e. including those caused by using the phone).

Here is the data published for the new iPhone 4S *pie chart klaxon*:

Pie charts are, at best, overused as a data visualisation technique (at least this one isn't 3D). This  does quickly convey some useful information: while production unsurprisingly accounts for the biggest chunk of emissions, customer use makes up almost a quarter of the total. The presence of recycling is also of note, a handy reminder that breaking down and resuing electronics is not an energy-free, or even safe, activity.

So we have a mildly interesting breakdown of the emissions sources alongside a total figure of 70kg of GHGs. Overall, I go away with a relatively good feeling about the product's environmental credibility, mainly as the pie chart seems to be made of wood.

What is lacking from this display is a comparison over time, in particular we have no context for the total GHG figure which is the data central to any environmental analysis. Apple have already published comparable data for the previous two iPhone models, it would have been very easy to include them all in one place. It only took me about three minutes to do the bar chart below, which is very revealing. Indeed, if one were cynical maybe it would be a bit too revealing for Apple to publish themselves:

Compiling the data from three pie charts in separate reports into one bar chart shows that across most parts of the lifecycle, iPhone emissions are on the rise, with a 27% jump between the iPhone 4 and 4S.

Not good.

What's more the seemingly relentless rise in production emissions was only offset on the iPhone 4 by a huge drop, of almost 50%, in those estimated from customer use. Sadly, there isn't much detail published in the methodology for these figures. I have no reason to doubt them, other than to point out that such a large change is... well... surprising.

So what has this simple exercise told us?

1. Pie charts can convey a very limited amount of information and, like other data visualisations, are easily sullied by daft colour schemes. Bar charts almost always trump pie charts.
2. Context is key: by adding in comparison over time we get a much better understanding of the numbers, even for pretty abstract stuff like GHG emissions.
3. Apple's environmental reporting might be a (small) step in the right direction, but poor visualisation and manipulation of the data is obscuring an increasingly disturbing trend in their environmental record.

[Thanks to Darragh Browne from Carbon Calculated, whose initial spot prompted this post.]

1. Representing exactly the same data in two different ways is a waste of time, space and ink. Make a decision on the best way to present your data and stick with it (if you have to do this several times within a report, it also helps to be consistent with your techniques where possible).

2. As we go on about ad nauseam 3D pie charts are never a good idea, and in this case the 'lie factor' (that is, the difference between the pixels shown on screen for each pie slice and the numbers they represent) is huge. 

The story here is that traffic lights are resposnible for many times more carbon emissions than traffic signs. The message comes over more powerfully (and accurately) in the table.

As Nicola remarked on our last training day, pie charts can be just about acceptable when there are two or three sections. More than that is a definite no-no....

The table also changes the unit of measurement. Rounding to 'effective digits' is a useful rule of thumb when presenting data; only present the numbers which are necessary to display the differences between categories. Here, this is done by rounding to one decimal place (after changing the unit of measurement to kilotonnes rather than tonnes).

A good example of how clarity rather than quantity of information provides much imporved results for the reader.