[x838nwy]

The point you made about boredorm (and implied listener fatigue) is very valid.

To stop yourself going mad, how about restricting to fewer sessions (say 10), then have a break and change the track to a different kind of music and doing another stint of 10, and so on.

Choosing the Lorde song as your only test track is interesting. That ablum is quite compressed and "produced" (not all in a bad way). I wonder how a different kind of sound would fare?
I really like her album BTW.

My original playlist was not usable as i stupidly included songs from a hd that i didn't bring home with me. So yeah, i'm a bit of a genius like that. Anyhow, i chose it as it's a track i'm familiar with and the album is nothing short of awesome. I would like to also include some live recordings and may be of classical music in the mix to actually hear real (non synthesized) instruments and see if that makes any difference.

I will be redesigning the test and hopefully will give it another try on Sunday. I may take turns with another person - like do 10 each in one sitting and see what we find....

For some reason, i cannot find "stock" cables anywhere... Can we take the Blue Jeans Cable RCA as a "stock" cable, may be? I'll keep looking, but i don't want to go too far and end up with something that doesn't actually work...

 

[bigshot]

9 out of 16 is still in the realm of random chance. You appear to have not been hearing a difference, just thinking you did. The tricks the mind plays are as useful a takeaway from this test as whether cables have a sound.

 

[x838nwy]

9 out of 16 is still in the realm of random chance. You appear to have not been hearing a difference, just thinking you did. The tricks the mind plays are as useful a takeaway from this test as whether cables have a sound.

Fwiw, it was 9 out of 15. But i understand what you mean.

 

[ab initio]

Fwiw, it was 9 out of 15. But i understand what you mean.

I think what you are doing is a great start. Currently, you really don't have enough data points to confidently say that you hear a difference. It is possible to have a 60% success rate and conclude there is a difference; however, it takes a lot of statistically independent samples before a >95% confidence level is reached. I encourage you to continue with more trials to increase your sample size and to report them here. We can help you analyze your results. I think it is helpful  and informative to the rest of the community to go through the experiment and analysis processes, and perhaps it will inspire more folks to try their own tests!
 

 
Cheers

 

[x838nwy]

 
I think what you are doing is a great start. Currently, you really don't have enough data points to confidently say that you hear a difference. It is possible to have a 60% success rate and conclude there is a difference; however, it takes a lot of statistically independent samples before a >95% confidence level is reached. I encourage you to continue with more trials to increase your sample size and to report them here. We can help you analyze your results. I think it is helpful  and informative to the rest of the community to go through the experiment and analysis processes, and perhaps it will inspire more folks to try their own tests!
 

 
Cheers

 

 
Thank you
 
Before work this morning, I re-arranged the set up and searched for a stock pair of RCA interconnects (without the Blue Heaven, the wife can't hear her TV so that sort of had to go back!). I've also had to move the Gungnir to another spot as the Oppo bdp105D now replaced it… The bad news is, in the new position, the Gungnir now hums something awful. So a slight set back, and a pain in the ar$e as we only discovered it after I've set everything up… This is harder than it first appeared...

 

[esldude]

I think you are chasing ghosts thinking, "hey, if i dropped the 1st three, I got 2/3 right."  Oh believe me, I know how strong and seemingly right that seems when doing such tests myself. (Hey I must have been getting better after the first few). But what most likely will happen is more test trials and it will get closer and closer to 50%.  Even then it is well within statistical likelihood to get 8 of 12 now and again. 
 
Just for kicks in a spreadsheet.  I just created with a random number generator 1000 trials.  Set up so half are right and half are wrong.  When split into test segments of 15 choices over 66 sessions of 15 trials then 14 times 9 or 10 of 15 were correct.  When the same results were split up into 12 choices over 83 sessions, 8 of 12 or 9 of twelve occurred 8 times.  Things like this make it easier to see results that appear to indicate something or feel like something is going on quite regularly occur in just those numbers when you know the results are purely mathematically generated randomness.
 
In my randomly generated results there was one time with 11 in a row correct.  Since I arbitrarily split those into 12 or 15 trial segments it happened to fall so one 15 trial result was 10 of 15 and that 12 trial segment was one of the 8 of 12 results.  Had it fell differently you might have gotten an 11 of 12 result.  And it all would have been just random variation.  If you had gotten 11 of 12 it would be very hard to convince you it was random (and it meets the 95% confidence level).  Of course repeating the test again if random you wouldn't have likely repeated your performance. 
 
Now when people ask for more trials or repeats, some folks say  nothing is enough and just feel in their gut it is so right that they hear something.  But if you know just some basic statistical facts you learn certain results mislead you if you don't have the full perspective of what is possible.  Try playing with synthetic random results in a spreadsheet.  It makes it easier to see how such results occur when actually purely random.  I have done this with various fake trials using 10,000 or 100,000 results.  The results follow the statistical predictions.  Every so often unlikely things occur simply by chance. 
 
Still I commend you on trying the test for yourself and honestly reporting the results.  It is a good and educational thing to do.  Congrats!

 

[x838nwy]

  I think you are chasing ghosts thinking, "hey, if i dropped the 1st three, I got 2/3 right."  Oh believe me, I know how strong and seemingly right that seems when doing such tests myself. (Hey I must have been getting better after the first few). But what most likely will happen is more test trials and it will get closer and closer to 50%.  Even then it is well within statistical likelihood to get 8 of 12 now and again. 
 
Just for kicks in a spreadsheet.  I just created with a random number generator 1000 trials.  Set up so half are right and half are wrong.  When split into test segments of 15 choices over 66 sessions of 15 trials then 14 times 9 or 10 of 15 were correct.  When the same results were split up into 12 choices over 83 sessions, 8 of 12 or 9 of twelve occurred 8 times.  Things like this make it easier to see results that appear to indicate something or feel like something is going on quite regularly occur in just those numbers when you know the results are purely mathematically generated randomness.
 
In my randomly generated results there was one time with 11 in a row correct.  Since I arbitrarily split those into 12 or 15 trial segments it happened to fall so one 15 trial result was 10 of 15 and that 12 trial segment was one of the 8 of 12 results.  Had it fell differently you might have gotten an 11 of 12 result.  And it all would have been just random variation.  If you had gotten 11 of 12 it would be very hard to convince you it was random (and it meets the 95% confidence level).  Of course repeating the test again if random you wouldn't have likely repeated your performance. 
 
Now when people ask for more trials or repeats, some folks say  nothing is enough and just feel in their gut it is so right that they hear something.  But if you know just some basic statistical facts you learn certain results mislead you if you don't have the full perspective of what is possible.  Try playing with synthetic random results in a spreadsheet.  It makes it easier to see how such results occur when actually purely random.  I have done this with various fake trials using 10,000 or 100,000 results.  The results follow the statistical predictions.  Every so often unlikely things occur simply by chance. 
 
Still I commend you on trying the test for yourself and honestly reporting the results.  It is a good and educational thing to do.  Congrats!

 
I hated stats back in my days, but yes, there exist a possibility that I get 15/15 through just guessing, but it's going to be something awful small, like 1/(2^15) or something. It'd be cheating for me to simply omit any result so I kindda left them all in, the thing about looking at the last part only was that I *felt* it easier to tell them apart towards the end. Now I'm not saying that it *was* easier, just it *felt* that way. One can view this in two ways, either a.) my hearing was "trained" to pick out the difference more effectively or b.) if I imagine something hard enough, it will seem more real to me even if it does not, in fact, exist. I'm on the fence at the moment on that.
 
Any how, 'NORMAL' RCA CABLES ACQUIRED!!! It's a 1.5m pair, so I hope you guys don't mind that it will be going up against a 1m pair of Wire World cables. There was a cheaper pair at something like $6 but they really did look like they won't actually connect so I splashed out a bit and spent $8 on this pair. I will try for 30 tests this weekend. It'll be my wife doing the electric toothbrush + sys switch routine so there WILL be breaks and a lengthy explanation as to why I'm actually doing this. Now IF she's willing - and it's a gigantic IF - I'll ask her to do some too.
 
Will keep you posted 

 

[bigshot]

One thing you might want to take note of is the direction you always lean when you stretch the interpretation. You definitely have expectation bias. Which is to be expected! That's why you put all the controls on the test.

 

[castleofargh]

 
Any how, 'NORMAL' RCA CABLES ACQUIRED!!! It's a 1.5m pair, so I hope you guys don't mind that it will be going up against a 1m pair of Wire World cables.
 

 I'm not sure if going from 1 to 1.5m would actually change much. but I would tend to think that cable length is actually important, when I really don't care for the rest as long as it's standard for it's purpose(impedance/diameter/stranded/braided/made of gold/etc).
I mean if you have to believe cables to be such bad guys that you need to get a 1000$ one, wouldn't it make sens to try and get the shortest one to reduce all the bad effects they might be full of?

 

[limpidglitch]

   
Any how, 'NORMAL' RCA CABLES ACQUIRED!!! It's a 1.5m pair, so I hope you guys don't mind that it will be going up against a 1m pair of Wire World cables. There was a cheaper pair at something like $6 but they really did look like they won't actually connect so I splashed out a bit and spent $8 on this pair. 
 
Will keep you posted 

 
Strictly speaking, if the material and configuration of the cable is expected to make a difference, so would the length of the cable.

So if your test turn out positive (there is a difference), you will have no way to know whether that was due to the materials or the length of the cables.

Conversely, if your tests turn out negative (there is no difference), you can't completely rule out the possibility of this being due to effects canceling each other.

This is why in these kinds of tests it is normal to only change one variable at a time.

 

[bigshot]

But of course none of it makes a difference.

 

[limpidglitch]

Naturally, which is why I prefaced it with "strictly speaking".
The outcome should be interesting none-the-less.

 

[x838nwy]

Strictly speaking, if the material and configuration of the cable is expected to make a difference, so would the length of the cable.


So if your test turn out positive (there is a difference), you will have no way to know whether that was due to the materials or the length of the cables.


Conversely, if your tests turn out negative (there is no difference), you can't completely rule out the possibility of this being due to effects canceling each other.


This is why in these kinds of tests it is normal to only change one variable at a time.

While i totally agree with you, but we could edit your post and change the words "the length of the cable(s)" to "the brand of connectors used" and it'd still be a valid point. IF differences exist, my test will not indicate their factors of origin(s) or how much each factor contribute to the result as a whole. It's be interesting to do the test between two models of cables from the same company (same length). Sadly I don't have those right now

I'll get some results up later today if all goes to plan.

 

[castleofargh]

  But of course none of it makes a difference.

 cable length does make a difference, I've read some complicated stuff about it from people who seemed to know what they were talking about. but ofc the cable they spoke of was a 100m long ^_^. 

 

[SilverEars]

 
 cable length does make a difference, I've read some complicated stuff about it from people who seemed to know what they were talking about. but ofc the cable they spoke of was a 100m long ^_^. 

  Yeah there are datasheets that shows the max range of copper cables such as Cat5.  So there! cables make a difference.