Talk:MaxDiff

Copright Issue?
The text in this article is copied from the technical documentaion of the software product MaxDiff from Sawtooth Software. The original text can be found at http://www.sawtoothsoftware.com/download/techpap/maxdifftech.pdf —Preceding unsigned comment added by 217.110.160.34 (talk) 12:25, 21 November 2008 (UTC)

Big update (Oct 15)
A lot of new material has just been added to this page by Terry Flynn. While some of it is no doubt useful and worthy of inclusion, there are a number of issues. In particular: I have tagged the page accordingly --ChrisSampson87 (talk) 14:25, 1 October 2015 (UTC)
 * The introduction is clearly far too long and much of its content needs moving to other sections in the article.
 * The article does not seem to have a neutral point of view, with far too much reference to the apparent benefits of BWS (and no discussion of its limitations) and also too much focus on the recently published BWS book (of which Terry is an author).
 * I have added references to additional groups' research and toned down and moved and where appropriate deleted description of the book to address NPOV concern. It is difficult to radically reduce the appearance of me, Louviere and Marley in the description since so much of the theory and methods were published by us but I hope this helps. Much of this has all been moved to the Best-Worst Scaling page, where it should really have been from the start (my bad). The maxdiff page is now much shorter but makes clear the two uses of the term, rather than the misleading redirect of BWS to maxdiff (presumably done by a company who uses maxdiff as its name for BWS). TerryFlynn (talk) 11:08, 2 October 2015 (UTC)

Large edit 2 October 2015
The material I added yesterday should have been on a separate page for Best-Worst Scaling. This has been rectified, and the automatic redirect from Best-Worst Scaling to the maxdiff page has been deleted. I have attempted to address the NPOV point in the material now on the BWS page.

The maxdiff page now makes clear the two uses of the term: (1) The academic interpretation, with a long history - this may need additions from someone like Marley; (2) The commercial use of the term in North America.

Thus the second point made - about the structure and overly long introduction - should hopefully no longer apply. I am happy to edit/remove mention of the commercial company - it is not clear in any case whether they do in fact implement maxdff estimation in their named maxdiff procedure.

TerryFlynn (talk) 10:10, 2 October 2015 (UTC)


 * I've removed the tags, as the issues indicated have now been addressed.--ChrisSampson87 (talk) 12:30, 2 October 2015 (UTC)
 * Thanks :-) TerryFlynn (talk) 16:41, 2 October 2015 (UTC)

This article seems to omit the algorithm itself
Strangely, this article discusses everything about the MaxDiff polling technique except the algorithm used to score the responses, which is the heart of why it works. Why the omission, or is it here and I just didn't see it? David Spector (talk) 16:49, 15 October 2020 (UTC)