It seems that you're using an outdated browser. Some things may not work as they should (or don't work at all).
We suggest you upgrade newer and better browser like: Chrome, Firefox, Internet Explorer or Opera

×
Community
General discussion (archive)help with SCIENCE! giveaway(132 posts)(132 posts)
(132 posts)
Pages:
1
2
3
4
5
6
7
8
9
This is my favourite topic
Posted October 01, 2012
avatar
dragonbeast: yay!
(ps i guessed 55,43,69 so i was dead on at the first one but far off at the third)
Out of curiosity: How do you tell single digit percentages by eye (even if it is just a guess)? I guessed very quickly (50,40,65) - I wasn't sure if kazmar was looking for first impressions - and I'm not sure how I could have narrowed it down that much. Best I could imagine is multiples of 5, or maybe 2.5 if it looks like "just a bit more".
Posted October 01, 2012
avatar
dragonbeast: yay!
(ps i guessed 55,43,69 so i was dead on at the first one but far off at the third)
avatar
etna87: Out of curiosity: How do you tell single digit percentages by eye (even if it is just a guess)? I guessed very quickly (50,40,65) - I wasn't sure if kazmar was looking for first impressions - and I'm not sure how I could have narrowed it down that much. Best I could imagine is multiples of 5, or maybe 2.5 if it looks like "just a bit more".
at the first one it was just a feeling, the second one i felt far 50 to much and 40 only a little short so 43 was the closest to that feel and the 69 was because i did not feel like typing 70
Posted October 01, 2012
For those who are interested, here is the first part of the experiment, it does not contain the data on game choice.

https://rapidshare.com/files/1363886772/gogproject.pdf

I'm sure it has quite a few errors in it. :P
Post edited October 01, 2012 by kazmar
Posted October 01, 2012
Congratulations to the winners and thanks for the Giveaway
Posted October 01, 2012
avatar
kazmar: For those interested the values were (rounded to 1.d.p from left to right) 54.6 42.9 & 62.0
I wasn't too far off then, with 50%, 40% and 60%. I'm interested, though, did the test reveal any findings from the favourite type of game?
Posted October 01, 2012
50%, 43% and 62% here. The first one was wrong by far :/
Thanks again and congrats to the winners, I hope you got the results you needed o/
Posted October 03, 2012
avatar
kazmar: Ok class pen's down!

with 99 people's results now collected the study is at the end.

Important thing first the winners are:

Dragonbeast with the 9.99 GOG
&
Thespian* with the 5.99 GOG

(Or if you guys want you can have some games from the square promo before it ends, 3 for Dragonbeast & 2 for Thespian*)
Thank you to everyone who helped!

For those interested the values were (rounded to 1.d.p from left to right) 54.6 42.9 & 62.0
And yes Austrobogulator you were very close, but not as close as ashwald.

I will write up a quick paper with the results and analysis and post it here shortly.

Thanks again.
wow ... I was a lot closer than I thought I'd be: 55%, 40%, and 60%

Good luck on your full project.

And congrats to the winners!
Posted October 03, 2012
avatar
kazmar: For those who are interested, here is the first part of the experiment, it does not contain the data on game choice.

https://rapidshare.com/files/1363886772/gogproject.pdf

I'm sure it has quite a few errors in it. :P
Please make a new post when the experiment has finished. I'm quite interested to see your results and I think others are too.
Posted October 04, 2012
avatar
kazmar: snipp
avatar
Dzsono: snipp
I'll try and post something this weekend.
Posted October 14, 2012
Much later than hoped but, here is the separated out data. I split the types of games into two types, reaction (FPS platform) and studying (puzzle, turn based). From this I then measured the error for each shape and then the average absolute error for all three shapes. The average total absolute error for each group was 9.4% for studying and 9.6% for Reaction.
Attachments:
gog.png (20 Kb)
Post edited October 14, 2012 by kazmar
Posted October 14, 2012
But probably more interesting in the use of rounding, by collecting the frequency of 2nd digits it shows a slight increase in the use of 10's as estimates by the reaction group, such to make every other 2nd digit appear less than compared to the studying group.

The frequency is then matched for both sets, with 0's being the most abundant, then 5's then 2's and 8's (assuming use as surrogate 2.5% and 7.5% but leaning on the floor for the lower and the ceiling for the higher.)
Attachments:
gogdigts.png (20 Kb)
Posted October 14, 2012
avatar
kazmar: Much later than hoped but, here is the separated out data. I split the types of games into two types, reaction (FPS platform) and studying (puzzle, turn based). From this I then measured the error for each shape and then the average absolute error for all three shapes. The average total absolute error for each group was 9.4% for studying and 9.6% for Reaction.
The difference isn't going to survive any tests for statistical signifiance, but it's nevertheless interesting, thanks for sharing. :) I've been curious how you'd handle the huge number of different game choices statistically; aggregating them into two groups seems indeed the best way of action. Of course, you could also run an explorative factor analysis across the data, and see which way of grouping the games fits best ... ;)
Pages:
1
2
3
4
5
6
7
8
9
This is my favourite topic
Community
General discussion (archive)help with SCIENCE! giveaway(132 posts)(132 posts)
(132 posts)