For the answer, you have to wait until the poll is completed and closed
But it would be much better not to post listening impressions for now.
What was the point to say "ULTIMATE listening test" or "test your... AUDIO CHAIN" here? I don't understand. Should I pull out the best of my gears, and why? Because with 1 inch computer speaker I have already a preference. But what if the preference changes when I use different system?
I'm really confuse here. But okay, I will vote based on my 1" computer speaker at work...
There's always the possibility he's lying. People still have to be careful they can really hear a difference to avoid getting suckered.
From the start post:
"just vote which sound file you prefer (or no difference heard)"
Doesn't matter Jan - just a matter of preference, it's no real double blind
After being told that there is a difference (and after being told by one poster it is HUGE), nobody in his right mind will dare to say he didn't hear a difference. So now I have to select A or B - where do I base it on? No idea if I am biased or actually DO hear a difference. And preference cannot be argued anyway. Results are pointless.
Jan
He only said there was a difference in hardware.
And to prevent biases we have the ABX comparator tool in foobar. See post 7 in this thread.
Btw I did a phase reversal test and that showed differences about 30dB below the original levels. Saying that that is a huge difference is a bit far stretched...
And we all know about masking don't we, so audible differences could be quite difficult to hear in an ABX test.
Suggestion: We could show the results of the ABX test we did. Foobar produces a text file that can be posted here.
Okay, then. I have no idea what this poll is about... I will figure it out at home by playing the files on my speakers.
Which is still untrue; a socalled forced choice paired comparison is an A/B test which asks if "A better" or "B better".
And it is known to be a discrimnation (based on preference) test, quite often used in consumer testing.
In a 2-AFC test there is no tie (means no third answer option like "no difference" or "no preference")
If there is a third answer this test is called a 2-AC .
If the 2-AFC protocol causes a bias it will in no way favour a false positive as listener still have to guess if they did not detect a difference.
While that might lead to reduces power of the test it avoids the internal criterion problem of any listener, and therefore this is called a criterion free procedure.
The 2-AC test version provides more data but the statistical handling is not as well developed as it is for the 2-AFC protocol.
In the 2-AFC case there exist three traditional possibilities to handle the "no preference/difference" votes:
-) exclude them from the analysis
-) split them equally and allocate to the both other answers
-) allocate the votes to the both other answers under the assumption that
these listeners would have voted like the majority already did
after this, proceed with exact binomial tests for the result.
Recent discussion led to another approach in which a negative control (means the same stimulus presented twice) is used to get a socalled "identicality norm" for the population and to analyze the other results against this underlying "norm".
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.