Fun with Statistics

by Tedra Osell on December 11, 2011

Old friend and former grad student buddy Lawrence White (not sure where he’s teaching these days–Lawrence, are you out reading? –pointed out that <a href=”http://www.businessweek.com/magazine/correlation-or-causation-12012011-gfx.html”>this</a> might be a useful teaching tool. Plus it’s just kinda fun.

Suddenly it all makes sense

Speaking of teaching, this is one of two seasons of the year in which I feel quite gleeful that I no longer have that responsibility. You people really shouldn’t be surfing the web unless your grading is done, you know.

{ 20 comments }

1

Cosma Shalizi 12.12.11 at 1:08 am

Further to the topic. (I can’t draw a curve where the top correlation is below 0.7.) Since I plan to use this in my class next semester, I feel no guilt whatsoever.

2

Tedra Osell 12.12.11 at 1:46 am

Oh wow, that is awesome!

3

Substance McGravitas 12.12.11 at 2:34 am

Further to the topic.

That is far too much fun for one link. Thanks.

4

Mike W 12.12.11 at 2:52 am

Global warming can’t be a hoax. Clearly, NSF funding is being used to cause global warming!

5

Popeye 12.12.11 at 3:25 am

1: I randomly got one that maxed at .6261 (and only returned 1 result)

6

John Quiggin 12.12.11 at 3:32 am

But, when would we have time to surf the web if not for urgent, hard-deadline tasks like grading, for which procrastination is so irresistibly necessary

7

Yarrow 12.12.11 at 3:59 am

I can’t draw a curve where the top correlation is below 0.7

Go vertically from 0 to 80, horizontally to the line marking the next year, vertically down that line to 0, horizontally to the next year line, vertically to 80, and so forth. Most likely you’ll get “No results found for Drawn Series”. (Still awesome, though!)

8

Meredith 12.12.11 at 4:21 am

The more administrative crapola to take care of or the more grading to be done in fall and winter, the more time spent web-surfing. In the spring, the higher the pile of papers to be read, the more work gets done in the garden. You could graph it. And yes, it’s cause and not just correlation.

9

praisegod barebones 12.12.11 at 6:24 am

10

sg 12.12.11 at 8:17 am

I managed to get a near perfect correlation with a curve that peaks in mid-2004 and then declines to a steady state. It corresponds with any product released in mid-2004 (e.g. the “dell truemobile 1300”).

That link is awesome fun. The one where you type in a search and see what it correlates with is cool. Why does “koizumi junichiro” correlate at 0.6448 with “things fall apart achebe”?

11

J. Otto Pohl 12.12.11 at 10:29 am

The obruni exams are not due until 30 Dec. and I have finished grading all of them. Indigenous exam marks are due much later. I got eleven graded so far this morning. I only have a couple hundred left.

12

Belle Waring 12.12.11 at 12:07 pm

The link in 1 is amazing.

13

marcel 12.12.11 at 1:04 pm

About that link at 1: I get nothing, no images.
The page I get looks (largely like this):

Google correlate

Documentation | Search by Drawing
Comic Book |
FAQ | Please sign in to use this tool.
Tutorial |
Whitepaper |

Correlate Labs |
Search by Drawing |

At first I thought it might be my browser, since I use Opera, but I get the same thing in IE and Firefox. Any suggestions about what’s going on? Or should I be the one to say that Emperor Shalizi has no clothes?

(NB: The vertical lines in the preview below are not lined up, but they should be. Not sure how to do that).

14

Neil 12.12.11 at 1:23 pm

Sign of the times? Searches for “global warfare” are showing an upswing not seen for a long time:

http://www.google.com/trends/correlate/search?e=id%3AUbfDeBVQ6Wu&e=global+warfare&t=weekly

Marcel, I don’t know what’s going on with your browser. I can’t use the tool on my iPad, but even there I can see it.

15

Freddie 12.12.11 at 2:07 pm

“Correlation is not causation” may be my least favorite blog comment these days. It’s thrown out there constantly like it’s an argument ender. But actual science depends enormously on finding correlation, and in practice it’s often hard or impossible to prove causation. That’s okay, because often correlation is sufficient to create practical benefit. You can’t package that attitude into a zinger, though.

16

Leinad 12.12.11 at 2:18 pm

17

zrichellez 12.12.11 at 2:32 pm

One of my favorite recent graphs via Andrew Gelman as “worst graph of the year” is of “Military Considerations” which was reportedly used to train FBI agents. Sweet.
Still a few weeks left to challenge…but here it is
http://andrewgelman.com/2011/09/worst-graph-of-the-year/

18

Cosma Shalizi 12.12.11 at 2:50 pm

13: You have to be signed in to a Google account to play with it. (I have no idea what nefarious purpose Google has for insisting on this.)

19

Andreas Moser 12.12.11 at 4:56 pm

I use this to try to get people to memorize that correlation is not causation: http://andreasmoser.wordpress.com/2011/04/05/if-you-dont-read-my-blog-you-will-die/ – to no avail, unfortunately.

20

dsquared 12.12.11 at 5:02 pm

For two series of equal degree integration, correlation (in the sense of stationary residuals from a regression of one on the other) does imply (Granger-) causation.

Comments on this entry are closed.