Re: R: 2 or 3 levels: beware of the compromise effect

On Mar 26, 2004, at 10:33 AM, Roberto Castaldo wrote:
> Yvette:
> I remembered a case study from when I was taking management studies 
> classes
> and looked it up again (thanks Google!). This study showed that people 
> did
> not want to choose the extreme option and were more likely to choose 
> the
> second option if they could choose between 3 instead of 2 options. 
> McDonalds
> has known this for quite a while: letting you choose between small, 
> medium
> and large means more people will choose medium than without the large
> option. In literature, this is known as the 'compromise effect'.

MM:
In all of the cases I have seen involving organizations adopting 
accessibility policy, care has been taken to evaluate what among the 
guidelines is possible, given the affected content. The decision has to 
do with labor estimates and liabilities issues, not how many choices 
are offered. Governments and other organizations do not buy their 
accessibility standards with fries and a Coke.

> Roberto C:
> We all should do our best to guide people to apply all guidelines, but 
> let's
> check the worst case in both situations (2 or 3 levels):
>
> - 2 levels: I think that many users could choose to apply only level 
> one, so
> IMHO this case is quite bad and we should try and avoid it.
> - 3 levels: if many developers choose "medium" option, they will work 
> on
> each level 1 and level 2 criterion; it's not so bad, after all.

There are a couple of things I think need to be cleared up before we 
continue this discussion. First, the proposal I made, which was 
unfortunately associated (by me) with the word "optional," is:

Leave the three levels in place. But for _conformance levels_, offer 
level 1 and level 2 only (or, as I proposed a month ago, level 1, 1+, 
2, and 2+ -- or B, B+, A, A+).

The current document itself defines level 3 as a catch-all -- and even 
suggests that these are already optional. (Remember that not long ago, 
these were called "best practices".) Quote:
"(Level 3 success criteria) are additional criteria that go beyond 
Level 1 and 2 that _may_ be applied to make sites accessible to more 
people with all or particular types of disability."

The design of the document has been to take those things that 
evaluation tools can't test well and put them up into level 3. By 
creating a set of criteria that are not machine-testable, it should be 
made clear somehow that _no_ evaluation tool should offer a level-3 or 
AAA conformance approval. Ever. Substantially all current AAA claims 
are based on a poor (some could say dishonest, or negligent) 
interpretation of the output of an automated tool.

In order for a third standardized level of conformance to be genuinely 
useful, authors would need to run a usability test on their documents 
using several humans with a broad range of disabilities, and be able to 
provably demonstrate that their implementation of all level 3 criteria 
has been done, and has had a positive effect. This would take longer to 
validate than it would to produce the content in nearly all cases -- in 
many cases, testing would take several times the development time. It 
is my opinion that all those who will claim to have done so when WCAG 2 
is a Recommendation will be lying. We do not want to encourage liars: 
it's bad for business, and bad for users. Death to dishonesty.

-
m

Received on Friday, 26 March 2004 15:17:07 UTC