I admit freely that I’m still learning every day when it comes to how best to visualize data and that means that I look at other visualization much more in detail than I would have done in the past. There are many best practices around how to visualize data and when a data visualization doesn’t keep to them I can’t help but go into analyzer mode (yes, even my own and I’m in no way perfect!).
One of those that caught my attention this week was the following.
This was part of a very nice infographic-style data visualization highlighting the attendee demographics for an event that I came across on someone else LinkedIn profile. I’m not showing the full infographic as I don’t want to make it into a shame exercise but just screen printed one element to highlight where I think some improvements could have been made. My apology for the bad quality, I didn’t have access to the original so this is a screenprint of someone else screenprint.
The above, at first glance, does look nice. I like the color use and simplification of the donut graph. But…. Non-quantitative graph types like donut charts can confuse and this one proves it:
- Benelux at first glance seems to be the biggest category based on sheer width as well as volume of the part (it spans out outside the donut chart)
- but Benelux isn’t mentioned as such in the legend…
- What is mentioned are The Netherlands and Belgium (two of the three countries making up Benelux) with a total percentage of 7% + 6% = 13%
- Mm… that does seem to be a lot less than Germany which is listed in the legend as representing a whopping 45% of attendees
- So either Luxembourg (nr three of Benelux) had more than ~32% attendees which wouldn’t be logic as that would make it the number two country without it even being mentioned in the legend or…
- The size of those parts have nothing to do with the actual numbers
- This bytheway also applies to the Nordics part of the chart which I guess would normally consist of Sweden (2%) + Norway (7%) + Danmark (2%) + Finland (not mentioned). In theory it is of similar size with Benelux (13% vs 13% based on the numbers in the legend) but in reality the visual part is closer to half that of Benelux…. And
- “Other” which should represent everything else normally would account for 29%, making it bigger than all the others except for Germany but is the smallest of all parts…?
- And on that note… Why all this grouping of nationalities without doing so too in the legend? Assuming everyone knows what countries belong to “Nordics” and “Benelux” is a slippery slope…
- And lastly one more things that really irks me is that when I add up all the numbers in the legend I get to 83% Where is the rest?
Ok, I’m sure there is a lot more to say about this graph but the above was enough to make me cringe already so I’ll stop at that.
Consistency, clear and concise labeling, correct and complete referencing as well as simple checks to make sure a data visualization is representing what it says it does are basics that everyone should adhere to. The above clearly doesn’t and I can’t help but feel very strongly that someone simply took an existing image (not an actual chart) and pasted on some names and values to represent their statistics instead of making a real graph. What do you think?
One thought on “How not… data visualizations”
I think this sort of post is an excellent reminder that the ease today with which we make graphics (or copy and repurpose) can sometimes lead to sloppy or misleading impressions. We all need to look at our data visualizations and ask what point we are making, whether the visualization gets across that point, and whether the data supports the visualization. This seems to fail on at least the latter two, as I am not sure what point was intended so I can’t judge that.
Comments are closed.