Seeing theory
January 1, 2019 9:53 AM   Subscribe

It's one thing to know statistics; it's another to see statistics.

"Seeing Theory was created by Daniel Kunin while an undergraduate at Brown University. The goal of this website is to make statistics more accessible through interactive visualizations"
posted by ChuraChura (7 comments total) 71 users marked this as a favorite
 
This website makes me feel physically unwell. I am not trying to be churlish but I wonder if it is age or just something that is a little unusual like people that get seizures watching certain video.
posted by Pembquist at 10:37 AM on January 1, 2019 [1 favorite]


Bookmarked for throwing it at the “Coincidence? I think not!” tinfoil hat crew.
posted by farlukar at 10:38 AM on January 1, 2019


Should really consider a "roll 1000 times" or larger button. 100 rarely converges in a way that answers more questions than it raises.
posted by abulafa at 12:06 PM on January 1, 2019


Everything I know about statistics, I learned from this book: How to Lie With Statistics by Darrell Huff.

(Personal story: in 6th grade, I had an assignment to write a book report on a non-fiction book... ANY non-fiction book. In my search for a short book in the school library, I found that, with the bonus of a provocative title that was a perfect writing prompt. But as I tried to do a 'quick read', it made me think more and more and ended up going far to inform my philosophy of "positive cynicism".
posted by oneswellfoop at 12:24 PM on January 1, 2019 [2 favorites]


If you like this site, you may also like the much more comprehensive Random. (Check out the "apps".)
posted by aws17576 at 12:33 PM on January 1, 2019 [3 favorites]


These are gorgeous. The explanation of conditional probability is particularly beautiful (though the "rain" effect might confuse a little since they're also talking about "raining" as one potential outcome; the "raining" in the animation is just showing where random events fall). Super, super impressive work from an undergrad also.

I also completely defend the choice to show 100 rolls instead of 1000; part of getting an intuitive sense of probability is realizing that even though 100 repetitions of a simple dice roll sounds like a lot, sometimes it actually takes much longer to really converge.
posted by en forme de poire at 1:14 PM on January 1, 2019 [4 favorites]


It's a lovely project. I also like the choice of the "roll 100 times" button, which is easy to press many times in a row.

I do physics experiments where we look for part-per-billion asymmetries. The experiments are hard: from proposal to completion is usually more than ten years. When people ask me about it I frequently ask them to imagine deciding whether a coin is fair or not: how many flips does it take? Suppose I tell you a coin will come tails 51% of the time, instead of half --- how many flips do you have to do to believe me? Everybody guesses a hundred. The answer is tens of thousands, depending on the statistical confidence that you want. The visualization here is very good.

And yes, if you want to see a part-per-billion asymmetry, you need something like 1018 events. Some of my experiments use neutron beams. That many neutrons, 1018, have a mass of a microgram. I kind of boggled when I first realized that I had built a machine that had absorbed a microgram of pure radiation.
posted by fantabulous timewaster at 7:39 AM on January 2, 2019 [6 favorites]


« Older Cilia later   |   Queer Kids Wish You a Happy New Year Newer »


This thread has been archived and is closed to new comments