Oh yeah? And speaking of coins and coin tosses, how about this?:
I find dimes - every day. (The US 10 cent piece). Not pennies, not nickles, not quaters - dimes. Find them at home, at work, on the street, at the beach in the supermarket, in bars etc.
And they are 9 times out of 10 heads up. Weird.....
I believe that late Franklin Delano Roosevelt is trying to tell me something. If only I knew what.
I find dimes - every day. (The US 10 cent piece). Not pennies, not nickles, not quaters - dimes. Find them at home, at work, on the street, at the beach in the supermarket, in bars etc.
And they are 9 times out of 10 heads up. Weird.....
I believe that late Franklin Delano Roosevelt is trying to tell me something. If only I knew what.
@ syn08,
please; you have to calculate the probability that by 45 participants, each doing 30 trials and answering by pure chance, not even one participant got more than 53% correct answers.
Additional information; it needs at least 30 trials, because 16 out of 30 is 53,33% -> 53%
It was reported by miliosz that no participant got more than 53% correct answers -> that means every particpant got in his 30 trial attempt between 0 - 53% correct answers.
@ janneman,
that´s just the beginning; one day we´ll talk about fairness coefficients, power and all the other really exiting stuff.
please; you have to calculate the probability that by 45 participants, each doing 30 trials and answering by pure chance, not even one participant got more than 53% correct answers.
Additional information; it needs at least 30 trials, because 16 out of 30 is 53,33% -> 53%
It was reported by miliosz that no participant got more than 53% correct answers -> that means every particpant got in his 30 trial attempt between 0 - 53% correct answers.
@ janneman,
that´s just the beginning; one day we´ll talk about fairness coefficients, power and all the other really exiting stuff.
Jakob2 said:@ syn08,
please; you have to calculate the probability that by 45 participants, each doing 30 trials and answering by pure chance, not even one participant got more than 53% correct answers.
Additional information; it needs at least 30 trials, because 16 out of 30 is 53,33% -> 53%
It was reported by miliosz that no participant got more than 53% correct answers -> that means every particpant got in his 30 trial attempt between 0 - 53% correct answers.
The probability in a 30 coins throw to NOT get any of the 14 15 or 16 heads is obviously 1-0.415 = 0.585
Now if you repeat 45 times the 30 coins flip, the probability to get AT LEAST ONE out of 30 experiments not rendering 14, 15 or 16 heads is 1 - the probability to get 14 15 or 16 heads in ALL 30 experiments. Which makes:
1 - 0.415^45 = 0.99999999999999999351120054308606
otherwise said, almost true for every practical use. What miliosz apparently said has a probability to happen of about 6.489e-18 If this is what miliosz stated, then that's absolutely baloney.
To be fair, milosz didn't specify the number of trials per subject. He simply reported that not one of 45 successfully identified the source at better than 3% above random chance. Obviously in a 'coin toss'-type scenario the statistical results merge to 50/50 with a high degree of confidence given enough trials. The proper question then is: given that no subject bested chance by more than 3% and assuming the results were completely random (ie. no audible difference existed in principle), how many trials per subject are required to reach that confidence level to a statistically significant degree? Your analysis suggests a pretty big number.
rdf said:
The probability to get in a N coins throw the results between 47% and 53% needs to be brought to about x = 0.925 (because 1-x^45=0.97)
The number of throws (experiments) per every of the 45 subjects to have the output into 47-53% with a probability of 0.925 is not easy to calculate (a combinatorial equation has to be solved) and such results are usually tabulated. I'm not in the statistics so I don't have tables at hand, but my wild guess is for at least several hundreds.
Re: Re: a perfectly good idea...
I finally read it (quickly). By its reasoning, the differences between CD and SACD can be expected to be quite clearly recognisable in longish listening tests due to the presence of "hypersonic" content, given suitable material and amps/speakers.
It also makes the suggestion of "audio blindsight" that I went on about earlier: "This finding suggests that these areas may not belong to the conventional auditory perception system." (although I've seen plenty of papers where the authors get all creative - some can pull off the image of credibility better than others)
I'm still no closer to knowing which way is up. Maybe I'll drag out the SACD player and set up some tests myself.
Jakob2 said:
I finally read it (quickly). By its reasoning, the differences between CD and SACD can be expected to be quite clearly recognisable in longish listening tests due to the presence of "hypersonic" content, given suitable material and amps/speakers.
It also makes the suggestion of "audio blindsight" that I went on about earlier: "This finding suggests that these areas may not belong to the conventional auditory perception system." (although I've seen plenty of papers where the authors get all creative - some can pull off the image of credibility better than others)
I'm still no closer to knowing which way is up. Maybe I'll drag out the SACD player and set up some tests myself.
syn08 said:...
etc...) b) the cause/effect relationship is weaker than we should care about (we don't consider the impact of the Heisenberg uncertainty principle in our audio amps, or should we?
Well the noise floor in an amplifier or microphone etc is determined by events on a subatomic scale, so it's not too much of a stretch of the imagination. Now it's just a noise floor so should be random and sound like "noise" - but it's not - it becomes a fixed signal which is always the same to everyone when played. Therefore it's going to be different from what was actually recorded every time it is played (because it's the same if you get what I mean). The significance of this data (as listened to / measured) is defined when it encounters temporal processing (frequency domain transformations - noise shaping, bandlimiting, ears) which is in the range of the original noise floor. Think double slit electron experiment (or use the modern alternative to thinking, ie Google). This should mean the data on a CD can be variable, because it is actually fixed.
That sounds mad enough to me. All this LHC stuff has triggered a form of web browsing insanity that is difficult to fight.
adx said:
LHC = Large Hadron Collider
(The first sign of insanity is replying to your own posts, but I've already beaten myself to it in this case.)
As a result of all this browsing, I decided I wanted to build a "small hadron collider". It could be buried about a foot under the grass in the front yard, or even straddling the fence with some neighbours. The hadron could go in a Halon bottle to save some labelling effort. Rather than liquid helium I would use liquid water, and make sure it was a coldish day.
Somewhat predictably I do actually have a lot of the stuff needed to do it, or at least make a cyclotron: High vacuum pumps, big RF amplifiers, massive power supplies for (non-cryogenic) magnets, CNC and stuff.
(And in case I really confuse someone who stumbles across this, no, "hadron" isn't a gas, it's a class of subatomic particles.)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.