I always read the Guardian‘s ‘On The Road’ car reviews on a Saturday. They’re not detailed enough to inform a purchasing decision but I like car reviews that aren’t all Top Gear about things. I’ve always been intrigued by the reviews’ Eco Rating so I decided to graph the data.
The Eco Rating is given as a number from 1 (or 0?) to 10. Quite high numbers are given to cars that consume a fair amount of irreplaceable resources which always seemed odd to me. (And let’s ignore the eco-ness of owning a large metal and plastic thing in the first place.)
So I set about creating a graph that compares the cars’ fuel consumption and CO2 emissions to the Eco Rating, to see what the correlation is like. Here are a couple of images, but you get a better feel for it by playing with the interactive version on bl.ocks.org.
I created this with D3.js using data from a couple of years’ worth of reviews (all now in this Google Sheet or this JSON file). I think this is a “ladder graph”, a variant on a Slopegraph — read more about both in Charlie Park’s blog post.
If the Eco Rating bore close relation to the fuel consumption and emissions we wouldn’t expect to see many lines crossing over. But there are. For example cars with a 7/10 rating range from the Mini Clubman Cooper S All 4 with 38mpg to the Mercedes E-Class that does 72mpg. Their CO2 emissions are also quite different.
One might initially assume that if everything matched perfectly then the lines would all be horizontal… but the angle depends so much on the domain chosen for the left-hand axis that I think that’s not necessarily true. Less variation in angle of slope, and fewer lines crossing, would be good indicators of accuracy though.
It’s not like I’m shocked — SHOCKED! — that the Eco Rating is probably plucked out of thin air rather than being a rigorously-tested Which?-style score. But it’s nice to have my assumptions confirmed.