- From: Gregg Vanderheiden <GV@TRACE.WISC.EDU>
- Date: Wed, 15 May 2002 12:05:58 -0500
- To: w3c-wai-gl@w3.org
- Message-id: <006601c1fc32$c89c37d0$b817a8c0@laptop600>
Hi All, At the last GL Teleconference call we discussed the levels of conformance and how to determine what goes in which. I promised to take a crack at capturing what has been discussed and done in this regard. Here is what it looks like so far Preamble 1 - We already decided that success criteria had to all be things that were testable. This could either be machine testing or Human testing with a high inter-rater reliability (HIRR). (i.e. most raters who understand issue and/or measure would all say the same thing.) 2 - We also decided that Levels 1 and 2 would have to be things that applied to all (types of) sites. Especially important for Level 1 - since no type of conformance could be claimed if without all level one items. And if we made it so that some sites could not reasonably meet a level 1, we would be making it so that some sites could make not claims of accessibilty. In the notes below MACHINE - means machine testable HIRR - means Human evaluate-able with High Inter Rater Reliability (meaning that raters would agree) POSSIBLE GOAL STATEMENTS FOR CREATING LEVEL 1,2,3 etc SUCCESS CRITERIA Criteria for Level 1 Success Criteria - Things that are testable (Machine or HIRR) - We want to recommend that these be done everywhere - Needs to be things that can be done on every site - no exceptions unless in the success criteria - Absolute essentials. If these aren't done - most everyone in a major disability group can't access the information. (e.g. no captions) Level 2 - Things that are testable (Machine or HIRR) - Things that go beyond Level 1 but that can still be done on all sites (if designed properly) - Important and readily achievable on all sites without great burden - These are things that we hope and encourage all sites to do - and want these to be doable enough that we can ask all managers to require (i.e. require Level 2 conformance) Level 3 - Things that are testable (Machine or HIRR) - Things that can be done on all sites but that would require extra effort and commitment. - Don't expect these to be done on most sites. But would be done on sites that really want to go the extra mile. - First level that would include things that require a certain level of control over the server or grasp of programming/technologies fall here? (ex. a typical hosting service won't allow you to do X or run Y script on their servers) Additional / Advisory Items - things that are not testable but are good ideas to address this checkpoint (good advice) - things that can't be done on all sites - things that might be done to target or tune a site specifically for a disability group who needs this checkpoint SHOULD THERE BE A LEVEL 4???? - only reason for a level 4 would be to put the Advisory Items into a Level - instead of moving them to the non-normative block of information (which would have to happen if they were not success criteria). Since some listing may ONLY list the success criteria (and not the non-normative advisory section) making these level 4 would keep them visible - to make the advisory items into level 4 items though, they would have to be played with to make them testable. And I'm not sure we can do that without making them harder to read and or less effective. NOTE: As you go down the levels - you will pick up more and more people who cannot use a site unless lower level items are implemented. Thus the lower levels make it easier for some people with disabilities, but they will make it possible for others, particularly with multiple or very severe disabilities. Oh yes. I also looked at the MUST, SHOULD, MAY and CAN'T USE, HARD TO USE, EASIER TO USE formats of the WCAG 1.0 guidelines - and couldn't make them work here. One thing that we found in our work in the GL group - is that items that make it easier to use for some people - are critical for others. So all the items end up being "MUST" or "CAN'T USE WITHOUT" for some people. Thus we decided that all checkpoints need to have Level 1 items -- even if they were hard to write and still make the Machine or HIRR testability criteria. So we end up with "the site reviewed the content with the objectives below in mind" or "The site asserts that they reviewed the content with the objectives below in mind" Or level 2 that are " the user asserts they considered this and did the best they felt was possible and appropriate." But we don't want to have checkpoints that don't have any testable measures that can be applied across sites (or we make some sites that cannot claim any conformance since L1 is minimum) AND - we don't want to take some checkpoints off the table for L1 (or they might not even show up on people's todo list). Your thoughts? Gregg ------------------------------------ Gregg Vanderheiden Ph.D. Ind Engr - Biomed - Trace, Univ of Wis gv@trace.wisc.edu --
Received on Wednesday, 15 May 2002 13:10:27 UTC