I was bored, so I decided to put together a survey and get people to answer very large numerical questions. Some are more accessible (populations, etc.), while others are more abstract. I'm hoping to see if the distribution of answers change depending on the familiarity of the subject matter... and whether the distributions are anywhere near the right answer.

If you have a couple minutes and wouldn't mind filling out the survey, I'd appreciate it. Please take it seriously. I encourage you to look up the answers after you're done, but please don't post them in here for others to see before they take the survey.

https://docs.google.com/forms/d/1LubG1dxwM-TEw9Xzl2lvIo0mcoEiSI5jyZ6hwD9yfSc/viewform

I'll come back and share the results after a couple days. I figure by then I'll have gotten all the participants by then and won't be skewing anyone's response with data.

4/12/2014 8:15:36 AM

what grade did I get?

4/12/2014 8:59:20 AM

what grade did I get?

4/12/2014 8:59:20 AM

it was embarrassing how far off I was on the NC and Moon questions.
several orders of magnitude

4/12/2014 9:19:05 AM

So far I have learned that some people are horribly misinformed about the population of the world... or have no idea how many zeros to make a large number.

4/12/2014 9:30:22 AM

the inability to demarcate with commas can be a serious issue. Like giving phone numbers to europeans. The numbers are all pretty much the same, but just the cadence in which they are delivered is enough to completely crushfuck how you store and file the information

I wonder if thats part of your study too. If the observation of the effect of elimination of digit boundary marks on a person's rigid calculation ability isn't part of your study, you could tack that on there with a fancy footnote citing "His Eminence, Fermat d'Warren County"

I humbly request 2 credit hours for my work in this field.

[Edited on April 12, 2014 at 10:06 AM. Reason : sig]

4/12/2014 9:41:35 AM

The only one I knew was the US population.

The rest where ballpark guesses. Fucked up most of them.

4/12/2014 10:24:47 AM

what grade did I get?

4/12/2014 10:43:10 AM

I, too, wanna know the answers

4/12/2014 12:09:51 PM

i think i was within an order of magnitude for all but 1 or 2. Not necessarily close, but at least within an order of magnitude, haha. I would say that I was only close on 1 or 2.

4/12/2014 12:30:34 PM

"I wonder if thats part of your study too."

No, just Google's shitty inability to process commas for a numerical inputs. And I do think it fucked some people up... unless people really think there are like 31 billion Americans.

4/12/2014 5:26:37 PM

LOL

4/12/2014 6:36:59 PM

My fermi estimates (some are probably way off)

310,000,000
375,000
100,000
50,000
5,000
300
500,000
75,000
2,000,000
300,000

4/12/2014 7:11:40 PM

you should totes make an "answer page" that peeps can go check out

4/12/2014 7:26:00 PM

I have the answers, but want to wait to give them out publicly so that people will not see them before taking the quiz. If anyone has taken the quiz and wants the answers, shoot me a PM and I will send them to you.

4/12/2014 7:52:24 PM

I did well on all of them. Moon one was probably the worst -- I thought it was a 3 second radio delay to the moon but it is only 1 second.

4/12/2014 8:07:28 PM

I was pretty close to the moon one. Not because I knew from any science class or calculation, but rather whenever someone talks about a car with a shitload of miles on it, they normally mention whether it could be driven to the moon and back (knew it was about 1/4 million)

[Edited on April 13, 2014 at 2:24 AM. Reason : ]

4/13/2014 2:24:29 AM

The only one i felt way off was the roads in NC and words in Tolstoy

4/13/2014 2:39:07 AM

i think i put 17million as the miles of road in NC

4/13/2014 1:07:35 PM

I'd love to see how I did.

4/14/2014 3:01:52 PM

Im trying to crunch some of the data, but I ran into an issue trying to plot the histograms. There are some HUGE TAILS on the answer distributions... to the point where a majority of the answers end up bunched in a single bin when I plot it in Minitab.

I can spread it out a little if I take the log, but it's still heavily grouped in the middle and I feel like it loses some of the meaning. Any other ideas for ways to meaningfully extract out the tails in the distributions to focus on the meat in the middle... without just arbitrarily removing outliers until the histogram "looks good".

Maybe do a t-test on all the values and kill the ones that fail? Don't know if that would translate into a pleasingly informative histogram or not. All I know is... there are either some people who just don't fucking have a clue about numbers... or didn't take it seriously and just hamfisted the zero key.

[Edited on April 14, 2014 at 5:20 PM. Reason : ]

4/14/2014 5:20:08 PM

How does the average of each answer compare to the correct answer?

My guess compared to real answer:

Population of US: under by 10%
Earth to Moon: under by 20%
Miles of Roads: under by 37%
Babies born daily: under by 72%
Google employees: under by 69%
Cokes per second: over by 857% (order of magnitude off)
Mass of Empire State Building: over by 311% (order of magnitude off)
War and Peace word count: under by 57%
Species of Insects discovered: under by 97%
Seconds of Dr Who: over by 34%

Looks like I mostly under predict, but I was pleased to get the right order of magnitude for a lot of them.


[Edited on April 14, 2014 at 7:00 PM. Reason : numbers]

4/14/2014 6:42:01 PM

that's not what "order of magnitude" means. You're within an order of magnitude on all of them.

4/14/2014 7:05:44 PM

Number of Cokes, weight of Empire State, and insect species identified were an order off.
Cokes per second: 100,000 vs 10,450
Empire State: 1,500,000 vs 365,000
Insects: 30,000 vs 1,000,000

Guess I should have put the other term as the denominator in my percentage calculation.

(me-real)/me instead of (me-real)/real

[Edited on April 14, 2014 at 7:29 PM. Reason : or something like that]

4/14/2014 7:27:43 PM

Alright... I crunched the numbers! Spoilers: It's a ton of log normal distributions.

Responses: http://goo.gl/m1rdsE

Answers and Histograms: http://goo.gl/LwEJy1

Some of the humongous answers threw off the means. I didn't care enough to go back and calculate the mode.

[Edited on April 14, 2014 at 7:34 PM. Reason : ]

4/14/2014 7:34:17 PM

"publis roads"?

4/14/2014 7:38:57 PM

I had three that I was within like 3% of. IPopulation, moon, and war and peace...

I was surprised I was within 10% of Dr who, I've never watched it and know very little about it.

Totally fucked the coke one, Google employees, and roads.

[Edited on April 14, 2014 at 7:44 PM. Reason : doesn't look like I'm in the spread sheet]

4/14/2014 7:39:42 PM

Quote :

"Answers and Histograms: http://goo.gl/LwEJy1 "

some people are seriously stupid

4/15/2014 4:09:05 PM

what would be useful is if you could annotate the correct answer on the histograms, maybe as a thin line in another color or something.

4/15/2014 10:51:11 PM