site banner

Small-Scale Question Sunday for September 10, 2023

Do you have a dumb question that you're kind of embarrassed to ask in the main thread? Is there something you're just not sure about?

This is your opportunity to ask questions. No question too simple or too silly.

Culture war topics are accepted, and proposals for a better intro post are appreciated.

2
Jump in the discussion.

No email address required.

I'm doing a fun little assignment where we're supposed to find a few misleading graphs, infographics, or other data visualization and talk about how it could be misleading. For example, I remember this fun one from the CBC in Canada that deliberately misstated the proportion of public funding. Anyone have any other good examples that spring to mind? (Not that I'm trying to freeload, it's not that kind of difficult assignment, but I'm curious if there are any that have just stuck in your head like that one did for months or years?

As might be expected, Soviet ingenuity has come up with some good ones. This lives rent-free in my head.

Occasionally I'd browse Defense Charts. Behold! Axis scaling is for chumps! There's also this gem.

There's also its more bitter cousin, Lib Dem Graphs, which I found was walled off by Twitter.

At least the original source of the CBC graph has been banished.

The FSM graph about pirates and global warming.

The Enliven Project false rape accusation infographic. Even Amanda Marcotte found problems with it. But the thing she didn't point out, which makes it obviously an intentional lie: the figures get smaller from the left side of the graph to the right side.

What the hell? What narrative were they expecting to sell with that trick?

The irony is that the data kind of does say what they want it to, in that the numbers seem to rise after the stand your ground law, but they just built an utterly ridiculous chart to show it.

I'll always point to ProPublica's Machine Bias as the example of lying with statistics and stories. Compare their pair of graphs "Black Defendants’ Risk Scores" and "White Defendants’ Risk Scores" to the Washington Post's article's graph "Recidivism Rates by Risk Score".

Maybe don't include that example in your assignment, though.

Could you share this instead as pictures or whatever? Wapo has a paywall, and the wayback machine does not include images.

If I’m reading this right, it looks like the risk scores are pretty well calibrated.

I assume ProPublica was using their graphs to argue the difference in histograms was a systemic bias?

Nothing nearly that sophisticated I'm afraid. Here is one of Propublica's senior writers accusing crime prediction algorithms of being racially biased for predicting more crime in black neighborhoods.

/images/16945411023725226.webp

And so PredPol continues to make crime predictions that are incredibly unevenly distributed by race. Take these two neighborhoods in Plainfield, N.J. – where 11 crimes were predicted in the White neighborhood and 1,940 in the Black & Brown neighborhood.

Try searching for "Misleading Liberal Democrat Bar Graphs" and you'll find a bunch of examples. It's become an internet meme.

To anyone who just thinks you’re injecting culture war, no, the Twitter account really is called https://twitter.com/LibDemGraphs

I like this one. The scale hides the 50% growth in private schools over the graphed period.

Is that from one of Tufte's books?

I found it online, in a discussion of Darrell Huff's How to Lie with Statistics, but which I don't have a copy of any longer. It wasn't sourced, so I'm not sure who the originator was.