|
|
Post by Arlon10 on Jan 16, 2019 14:01:17 GMT
If you're still curious about the normal distribution curve after all the wrangling and you want to see what one looks like you can just google "images normal distribution." A better way though is to build one yourself. That way you can watch the relationship between the data and the theoretical curve as the data builds up.
The best thing to measure is the toss of a coin since almost anything else is going to have "lurking" and sometimes more obvious variables distorting the curve.
To begin you get 11 columns labeled 0 through 10. You toss a coin 10 times and note the number of tails. If the number of tails is 7 you put a marker in column 7. Then you have to toss the coin 10 more times to get the position of the next marker. Perhaps you get 5 tails out of those 10 new tosses, then you put a marker in column 5. You have to toss the coin 10 more times for each marker, and you need quite many markers before you begin to see the curve develop. This can get very tedious, but you can use Python to simulate.
The point of all this trouble is to notice the developing picture. At first it might look nothing at all like a normal curve. The reason for that is too small a sample size, not lurking variables. later it will begin to very roughly approximate the normal curve. It would be much, much later before you could notice any lurking variables especially since with coin tosses there are virtually none.
*
* * *
* * *
* * *
* * * * *
00 01 02 03 04 05 06 07 08 09 10
Sample size too small
Getting better resolution requires more columns, for example 101 columns and 100 coin tosses per marker. Obviously no one wants to do that without a Python simulator.
I have an idea for an interesting experiment using this board's polling feature. I will make 11 choices available for zero to 10 tails. Each board member then tosses a coin ten times and selects the poll choice for the number of tails in their trial. That will give us a rough picture of the "quality" of the board. Of course it has some serious flaws. You can enter whatever number you like and some people might not enter the actual number they got. I suspect the number of responses will be quite far below that required (too small sample size) for a fair reading.
I considered making it possible for you to "vote" more than once. That way we could get a better picture with fewer members. However I do not know whether allowing more than one "choice" allows more than one vote per choice. In order to get accurate results you would have to be able to vote more than once for one choice. Does anyone know whether that is possible?
I will not add the poll here until I think enough people understand what is asked of them and are willing to participate. I think it would be best if the polling results are hidden until some trial period of about a week is completed. You might want to enlist friends to join this discussion board so that we can get a better picture.
|
|
|
|
Post by goz on Jan 16, 2019 21:30:17 GMT
Gee, that sounds like fun!
Are Australian coins OK?
Can I use a kip?
Do you have a system for the coins falling upside down in the Southern Hemisphere? (gravity can be a bit dodgy here)
When will I have time to do the washing?
Am I honest enough to get the results you want?
Can I enlist a supernatural force to that end?
Just askin'!
|
|
|
|
Post by Arlon10 on Jan 16, 2019 23:21:04 GMT
Gee, that sounds like fun! Are Australian coins OK? Can I use a kip? Do you have a system for the coins falling upside down in the Southern Hemisphere? (gravity can be a bit dodgy here) When will I have time to do the washing? Am I honest enough to get the results you want? Can I enlist a supernatural force to that end? Just askin'! It appears no one else is interested. I'm not surprised, but will try floating the thread a few days. What we would probably find is that this discussion board has very few members with very little honest participation in anything much. I suppose some people would rather we didn't make that obvious. People on boards like this like to think what they say matters. A bad curve would mean the board is "insignificant." Sunday is going to be a huge football day and that might thin the crowd here even more.
|
|
|
|
Post by goz on Jan 17, 2019 6:53:10 GMT
Gee, that sounds like fun! Are Australian coins OK? Can I use a kip? Do you have a system for the coins falling upside down in the Southern Hemisphere? (gravity can be a bit dodgy here) When will I have time to do the washing? Am I honest enough to get the results you want? Can I enlist a supernatural force to that end? Just askin'! It appears no one else is interested. I'm not surprised, but will try floating the thread a few days. What we would probably find is that this discussion board has very few members with very little honest participation in anything much. I suppose some people would rather we didn't make that obvious. People on boards like this like to think what they say matters. A bad curve would mean the board is "insignificant." Sunday is going to be a huge football day and that might thin the crowd here even more. Thanks for your erudite and explanatory answers.
|
|
|
|
Post by phludowin on Jan 17, 2019 11:00:16 GMT
The bell curve only appears if we let the sample size go towards infinity.
I doubt that we can manage this.
And since there's no poll here anyway to enter data, I'm not surprised that not many people have replied.
|
|
|
|
Post by goz on Jan 17, 2019 20:21:55 GMT
The bell curve only appears if we let the sample size go towards infinity. I doubt that we can manage this. And since there's no poll here anyway to enter data, I'm not surprised that not many people have replied. Don't be so 
|
|
|
|
Post by Arlon10 on Jan 18, 2019 10:35:30 GMT
The bell curve only appears if we let the sample size go towards infinity. I doubt that we can manage this. And since there's no poll here anyway to enter data, I'm not surprised that not many people have replied. No, we only need enough to see the shape of the "curve" of the data and how closely it resembles the theoretical model. That takes way less than an infinite set of data. On your computer screen you can see the shape of the curves in a person's face with far less than an infinite number of pixels. Digital audio would not be possible except that we do not need an infinite number of voltage steps to mimic analog audio. We can see a "smooth" curve with only a thousand points, but even with a few hundred points we can see where the data is going and its general shape. Even with a few hundred points we can see whether there are any strong irregularities. Smaller irregularities would not mean much anyway even if we could see them. Even with far less data we can see whether there is any skew in the data. There can be a poll as soon as people express interest in one. If a few people express interest, then probably many more will participate. Maybe we just need to invite people. If faustus5 , FilmFlaneur , @graham , gameboy , goz , rachelcarson1953 , or others participate and they invite others and those others invite others it would be more than enough to see the general shape of the curve. If 25 people participate we could see the general shape of the curve. Part of the problem of course is that most of you probably have yet to learn the mathematics involved to know what is going on to have any interest in it. |
|
|
|
Post by general313 on Jan 18, 2019 16:02:17 GMT
The bell curve only appears if we let the sample size go towards infinity. I doubt that we can manage this. And since there's no poll here anyway to enter data, I'm not surprised that not many people have replied. No, we only need enough to see the shape of the "curve" of the data and how closely it resembles the theoretical model. That takes way less than an infinite set of data. On your computer screen you can see the shape of the curves in a person's face with far less than an infinite number of pixels. Digital audio would not be possible except that we do not need an infinite number of voltage steps to mimic analog audio. We can see a "smooth" curve with only a thousand points, but even with a few hundred points we can see where the data is going and its general shape. Even with a few hundred points we can see whether there are any strong irregularities. Smaller irregularities would not mean much anyway even if we could see them. Even with far less data we can see whether there is any skew in the data. There can be a poll as soon as people express interest in one. If a few people express interest, then probably many more will participate. Maybe we just need to invite people. If faustus5 , FilmFlaneur , @graham , gameboy , goz , rachelcarson1953 , or others participate and they invite others and those others invite others it would be more than enough to see the general shape of the curve. If 25 people participate we could see the general shape of the curve. Part of the problem of course is that most of you probably have yet to learn the mathematics involved to know what is going on to have any interest in it. Why don't you just write a program, run it a hundred or a thousand times and post the results here? |
|
|
|
Post by goz on Jan 18, 2019 20:57:47 GMT
The bell curve only appears if we let the sample size go towards infinity. I doubt that we can manage this. And since there's no poll here anyway to enter data, I'm not surprised that not many people have replied. No, we only need enough to see the shape of the "curve" of the data and how closely it resembles the theoretical model. That takes way less than an infinite set of data. On your computer screen you can see the shape of the curves in a person's face with far less than an infinite number of pixels. Digital audio would not be possible except that we do not need an infinite number of voltage steps to mimic analog audio. We can see a "smooth" curve with only a thousand points, but even with a few hundred points we can see where the data is going and its general shape. Even with a few hundred points we can see whether there are any strong irregularities. Smaller irregularities would not mean much anyway even if we could see them. Even with far less data we can see whether there is any skew in the data. There can be a poll as soon as people express interest in one. If a few people express interest, then probably many more will participate. Maybe we just need to invite people. If faustus5 , FilmFlaneur , @graham , gameboy , goz , rachelcarson1953 , or others participate and they invite others and those others invite others it would be more than enough to see the general shape of the curve. If 25 people participate we could see the general shape of the curve. Part of the problem of course is that most of you probably have yet to learn the mathematics involved to know what is going on to have any interest in it. Way to go! What an impressive recruitment technique! Why don't we just do a graph of that? In Australia we count flies walking up a wall, that could be fun! Count you own local insect perambulating! Extra points if they  |
|
|
|
Post by Arlon10 on Jan 19, 2019 1:23:19 GMT
No, we only need enough to see the shape of the "curve" of the data and how closely it resembles the theoretical model. That takes way less than an infinite set of data. On your computer screen you can see the shape of the curves in a person's face with far less than an infinite number of pixels. Digital audio would not be possible except that we do not need an infinite number of voltage steps to mimic analog audio. We can see a "smooth" curve with only a thousand points, but even with a few hundred points we can see where the data is going and its general shape. Even with a few hundred points we can see whether there are any strong irregularities. Smaller irregularities would not mean much anyway even if we could see them. Even with far less data we can see whether there is any skew in the data. There can be a poll as soon as people express interest in one. If a few people express interest, then probably many more will participate. Maybe we just need to invite people. If faustus5 , FilmFlaneur , @graham , gameboy , goz , rachelcarson1953 , or others participate and they invite others and those others invite others it would be more than enough to see the general shape of the curve. If 25 people participate we could see the general shape of the curve. Part of the problem of course is that most of you probably have yet to learn the mathematics involved to know what is going on to have any interest in it. Way to go! What an impressive recruitment technique! Why don't we just do a graph of that? In Australia we count flies walking up a wall, that could be fun! Count you own local insect perambulating! Extra points if they If you don't have a video drone, it's something else to do. |
|
