Comments on: On the Abuse of Chernoff Faces

By: Transforming Data

Transforming Data — Thu, 20 Sep 2012 21:22:34 +0000

[...] Critique of Chernoff Faces Baby Name Voyager [...]

By: econniff

econniff — Mon, 29 Aug 2011 03:27:45 +0000

I like how, in the death penalty map, if a state puts mostly young people to death via either lethal injection or electrocution, you get a smiling wink. Hahaha. My first impression was that those were the states where people on death row were least likely to actually get executed. Thus: *wink*.

By: Tina

Tina — Sat, 17 Jul 2010 23:59:38 +0000

I’m just echoing what Reid already said – that because we tend to view faces as a whole and not as individual components, I had to keep going back to the coding scheme to figure out what I was looking at. I’m not sure that saved me any more time as a reader than if I’d been synthesizing data from multiple bar graphs. The death penalty map has a second problem – if you’re ignoring what the actual codes mean altogether, it looks an awful lot like putting African-Americans or non-pasty-white folks to death makes folks awful happy down in Texas. There are happy white faces here, but since the white faces are the majority on the map, it took me a while to find them – the two happy darker faces stand out rather prominently against that backdrop. (At least Nebraska looks to be unamused by putting non-whites to death.)

Overall, the whole use of faces of different shapes and colors in the first example reminds me far too much of Nazi propaganda (for example, the graphic on p.21 of Robert Proctor’s “Racial Hygiene,” if you have a copy handy, in which the Slavic face becomes more and more threatening as it gets larger to indicate population increase). I generally try to avoid coding my data in such a way that makes the reader think of Nazis.

By: PyongYang

PyongYang — Thu, 01 Jul 2010 14:23:35 +0000

Hi there

I just wanted to say i like your blog very much and i would have participated to your contest if it hadnt been about the Minard Map. It was a “blockbuster-map” and i just couldnt try and re-do it. But i think the contest was a good idea, maybe you should try and give a more large theme, with which we could be free to try more diffrent things.

Anyway, thanks so much for your blog, i loveit ! Keep on going and keep up the good work.

By: Reid McCamish

Reid McCamish — Thu, 17 Jun 2010 17:32:56 +0000

To me, the core problem with misusing these faces is that the data easily become obscured by the non-alignment of the features with the content of the data. For example, on third item, the example coding scheme, the assembled face means nothing to me without a detailed review of the coding scheme. I think faces are usually perceived as a single unit due to human’s need to process facial expression quickly. Thus, it takes a lot more cognitive burden to notice that the nose is big and the hairline is receding that to notice that the face looks angry or sad.

The first map works rather well. For example, the one dark face tells me, upon a glance, that the population is mostly black (notably *not* non-white), poor, and distressed, which maps pretty well to the actual data. The other maps tell me very little at a glance.

By: Andrew

Andrew — Thu, 17 Jun 2010 03:16:13 +0000

I guess Minard has been vindicated. His map really is the boss. I was going to enter a map, but I truly could not think of a way to display the data in a better way…I toyed with the idea of making it more artistic and less graphical, but could not adequately do it with the tools I had.