On Tuesday, September 27, 2016 10:37:53 PM Arne Babenhauserheide wrote:
> > -> Please lets not make this subject of the result discussion yet, it's
> > incomplete and potentially includes bogus data.
> 
> That is true, and I already wrote that that’s a possibility.
> 
> However that does not invalidate the more important result that using
> only one way to evaluate the poll results in bogus rankings, even if the
> correct data is used.
> 
> Each of the different ways to evaluate provides for different means of
> skewing the result, and from what I can tell, these means were used to
> some degree. From their difference between the evaluation methods we can
> at least tell how uncertain the result of the poll is.

Hmm yes, there are indeed some signs of attempts to skew the results by people 
voting for very extreme values. This may be fixed by the median already?
And yes, it would indeed be nice to have some way to deal with them beyond 
manually excluding people from voting/estimating
- if we do manually exclude certain estimates the resulting discussion will be 
lots of great fun :|

So overall, I am open to looking at those methods, just *after* trying with 
the original poll method first :)

Let's just shrug this subthread off as a misunderstanding please, ok? :)
I'll go to bed now and finish the main spreadsheets ASAP this week.
You'll also get raw CSVs then to experiment with, so if the original method 
does not cope well with manipulation you can try to prove it with data then :)

Good night & thanks for the clarifications!

Attachment: signature.asc
Description: This is a digitally signed message part.

_______________________________________________
Devl mailing list
[email protected]
https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to