General Discussion
Related: Editorials & Other Articles, Issue Forums, Alliance Forums, Region ForumsYou can teach an entire semester of how to lie with statistics with the y-axis of this chart.
You can teach an entire semester of how to lie with statistics with the y-axis of this chart.
Link to tweet
unblock
(52,328 posts)ProfessorGAC
(65,191 posts)I was looking at it on phone. When I used my tablet with 5 times bigger screen, I saw it!
jberryhill
(62,444 posts)The OP and the Tweeter don't understand a logarithmic scale. But that's why the graph is roughly linear.
Since most people don't understand how to read a log scale, using it on the news is not the best idea.
ProfessorGAC
(65,191 posts)It's odd numbering for a log scale, though.
And, in my original comment, I said there's no lie. An odd way to present the data without further explanation, but there's no lie.
jberryhill
(62,444 posts)Is that the source had a properly graduated log scale, and then they interpolated additional y-axis graduations using a linear estimate.
ProfessorGAC
(65,191 posts)It's still an odd approach. It's nothing I ever would have done. And I typically was showing data to other technical professionals.
mahatmakanejeeves
(57,613 posts)I disagree.
It starts out with increments of 30, then it shifts to an increment of 50, then it shifts to an increment of 10, and finally it goes back to increments of 50 again.
This might not be apparent on a small screen.
They should have started over.
jberryhill
(62,444 posts)Then, the graphic department interpolated linear tick marks.
That's how they ended up with what they got.
That's readily discernible to the same overall slope at the one and and the other, and the way in which every other log graph of the spread works.
The line is right, but the y-axis was a log scale that someone fucked up by adding graduations without knowing it was a log scale.
hlthe2b
(102,376 posts)NAH... There is no ceiling.