Power Cable Test – Volunteers Needed

Previously mentioned on another ZG thread but in the interests of a little further cross-forum promotion for those who may not be aware..

ZG's Wadia Miester (Tony of Coherent Audio) has kindly offered to host a blind listening test with a selection of aftermarket hi-fi power cables.

Volunteers are needed to take part. If wish to get involved then please read the volunteer tester requirements and method described here first:

http://www.hifiwigwam.com/forum1/655-1.html


Post on this thread (or HFWW) with Saturdays in September / October you would be able to attend should you wish to volunteer. The location is Gloucestershire.

An updating list of volunteer testers can be seen here:

http://www.hifiwigwam.com/forum1/820.html


At the time of posting two further volunteer testers are required.

Thanks.

 

If there's still a place I'll be happy to take part (I'm a 'C').

I could make Saturday 24th Sep, 1st Oct, 8th Oct, 15th Oct, 29th Oct (ie. not 17th Sep or 22nd Oct).

Please let me know asap if and when I'm wanted.

 

Thanks Steve. You're on the list.

http://www.hifiwigwam.com/forum1/820.html

As soon as a date becomes apparent I'll inform all by PM at the earliest possible opportunity (and post on the above thread).

Richard

ONE more?

Although volunteers still accepted on a provisional / reserve basis after the first four spaces have been filled.

 

Results now posted

http://www.hifiwigwam.com/forum1/1104.html

 

very interesting. Good news for Russ and Black Rhodium

 

Yes indeed; in each case two out of three thought it sounded better than the stock cable. There's also good news for the stock cable though; two out of three thought it sounded better than itself.

 

I found the Russ Andrews Yello cable made the biggest change to my components when I changed from the stock cables - and it was quite a pleasant improvement.

Upgrading to better cables in the range didn't make anywhere near the same improvement, and the Nordost Shiva was in my mind a complete waste of money and not as good as the RA cables.

Also I found that the cables to the amp made more difference than to the cd player.

It's nice to see the cheaper cables doing better than the high priced ones.

 

Well done to all who took part and to those that organised the event.......

If I might offer a thought or two for any future event........

Try and get a bigger listener sample and perhaps a smaller selection of items to test.

Its impossible to get a hundred listeners ....unless your a Uni but thats what your going to need to get any meaningfull statistically.....

there mustbe some hifi loving uni students doing stats that would jump at the chance to organise something like this isn't there?

 

it sounded better than itself? how impressive of it

 

I did have a thought when you were announcing the test dates, but since I wasn't volunteering I kept my gob shut. Surely it is too ambitious to try to choose a cable you prefer?

FWIW my experience with cables is limited. Some appear to change the sound slightly, but its small beer compared to other changes. I have never investigated further than that, since I have the rest of my life/house/hifi to organize first.

What I am trying to say is - suppose you accept the hypothesis that cables do change sounds - there is nothing in that statement to suggest that you should prefer one sound to another, to be able to rank them.

I would have thought that proving people can repeatably hear changes between A and B etc. is the best that could be 'proven' statistically.

 

AIUI there was no requirement to rank the cables, it was in order to say 'they sound the same', or 'they sound different and I have no preference'.

It's interesting that the differences Tony thought 'plain' were not obvious to the panel of professed believers. I don't understand why he wasn't on the panel.

(as a cynic I wasn't allowed to take part...)

Paul

 

That is what I tried to suggest to Mosfet even before the test. However, he think it would be too difficult to explain to listeners how to do this but I never understand why that is should be the case.

 

zanash, your comment about the relation between the number of test items (k) and judges (n) doing the test is incorrect. this can be seen as a quasi-experimental design, in which n does not need to be 100, it can be very much smaller for results to have statistical power. Similarly, k does not need to be small - indeed, with k large, a better handle can be had on the 'reliability' of the judges.

the use of the control A-A in the design fits within this.

the summary data given does not allow me to say more than this.

 

Statistically it is not a large enough sample. 4 listeners does not constitute a sample by any reasonable means. You really need a much larger sample than that to get a convincing and meaningful result. No matter how mathematically correct, everyone is going to say that the sample was too small, and it's still within the bounds of possible co-incidence. Get a hundred people to see it with a convincing percentage getting the same answer, and you're doing better, a thousand and maybe the world will start to listen!

 

there are two types of generalisation one wants to make in quasi/actual experiments:
1) internal
2) external.

The first is aimed at about being sure that the result (no difference, better, worse) is more than could reasonably be expected from chance. the reliability of the judges matters, and needs to be measured. the setting also needs to allow any possible differences to be discerned - the better the system & the recording, then the more likely that the 'marginal' benefit of a cable is discernible.

The second is aimed at generalising to 'all' good systems/recordings, and to 'all' judges.

The first is much easier than the second.

Classic example of an experiment is testing one strain of wheat against another, or one nuturient against another - using objective measures of growth. Here we must use a subjective measure - hence the judges, but these have reliability problems. Hence the reason for using the test items to assess that reliability, and so wanting the number of such items (k) to be large.

I regret its a long time that I have done this stuff professionally, but its do-able.

 

Hi,

Amusing.

First, may I ask if any controls where done to validate the test-setup and participating listeners agains "known audible" stimuli?

If not, the results are of no consequence, as we are unable to ascertain if any modest but audible difference would have been detected, had it been present.

So, based on the test I would say that not only can the test hypothesis (Mains cable induced differences are imagined) not be rejected with any certainty, but no other value can be attached to the test either, as no effort was made to illustrate what could be actually identified relaiably under the test conditions.

Past that, once we have evened out the chances of type a and type b statistical errors we find that we have insufficient data to draw any conclusion with any reasonable certainty.

Ciao T

 

I've just read 'I deliberately did not record who made which observations because I was anxious that this should be a test of cables, not of people'. Shame, as this reduces the power of the test, and so the number of judges (n) would likely have to be larger than 3.

(essentially, the judges are the measurement instrument, and as such contribute to total variance; the identity of the reliability of each judge would have to be confidential of course!!)

 

I concur. so the next step for those interested is to set up a quasi-experimental design which could detect what we wish to detect:

a) [no] discernible differences between out-of-the-box & specialist cable
b) improvement due to specialist cable

I also wonder whether applying the specialist cable to the CDP, and not the amp, is most favourable to discerning a difference - especially since the bel canto amps are reputed to be susceptible to effect of dirty mains.

but credit where credit is due - this was attempt to make objective summary/test of subjective judgements, and so should be encouraged.

 

Most importantly - were any of sloppies cables included?

Well with 4 listeners and multiple tests you can make significant conclusions - but it needs to be exceedingly obvious from the results - which this isnt. You could also try using ranking tables which compare every cable with every other and require less data for coarser conclusions. For instance with half a dozen listeners and a dozen cables it might work quite well. Been years since I did this kind of stats though so I would need to brush up.

 

I still think the people involved need to be congratulated for their efforts...not criticised on their methodology.

Though certain aspects could be improved in future experiments, lets try to be positive about this and not overly critical.