- From: giorgio brajnik <brajnik@uniud.it>
- Date: Tue, 02 Jul 2013 12:28:15 +0200
- To: public-wcag-em-comments@w3.org
- CC: evelleman@bartimeus.nl
Complete Processes: This is a vague notion b/c it relies on “process” which is itself not defined, and difficult to define I would try to redefine it in terms of goals that users try or may want to try to achieve. All pages (from a starting point) that have to be visited so that a certain goal, out of a given list of goals/use cases, have to be audited. Templates: The auditor has not enough info regarding which templates were used, if any. Website: self enclosed? the web is by definition not self enclosed Web page: this does not say anything of what can be inside this container. For something like gmail, I need to test many different states in which the “page” can be. How does the methodology help me decide which states to consider? Conformance Evaluation Procedure: in a usability report one usually starts by saying which method was used. Here you are assuming that there are more than 1 method, but you really do not list/present them. Step 1.a: second “where possible, it is recommended to use formalizations such as regular expressions or ..” – this looks to me useless. Why is a list of page titles and their URI’s not sufficient? Step 3.e: why 5? why 5%? Where does it come from? How does one select a random sample? Literally that would mean to first list all (ALL!) the pages and then using a random number generator, select the sample. A variation could be a random stratified sample, where pages are first grouped into categories (like product description pages) and a random sample is selected from each group. Another way is a sample generated through random walks in the site (i.e. randomly selecting one link among the available and not yet visited ones) Notice that even a pure random sample of pages has to cope with “complete process”. And hence has to be extended. The size of the sample depends on the kind of results that are needed. For statistical significance, the size depends on the spread/variance of the measured variable. Hence size cannot be decided a priori. Other auditing goals might call for other criteria, like a coverage criterion that aims at capturing all the possible types of problems. Step 4: Introduction: Very general question: in reality what do people do? For example the Germans that follow the BITV, what methodology do they employ? Or the Knowbility.org people that run the accessibility rallies? Is this methodology connected to those? Because there’s no bibliography in this document I think it is not. And it should not be such. Step 4.a: if one enumerates the use cases then the sample of pages may change, as per “complete process” requirement. Hence this step should come first. Step 5.a (Basic Report): Rubin, in this book on usability testing, gives a neat description of different goals that usability investigations might have. Here something similar could be done. Step 5.c: Performance score? Who’s performance? In which sense? Are you talking of accessibility metrics, or expected performance by users belonging to certain categories and trying to achieve certain goals? -- Giorgio Brajnik ______________________________________________________________________ Dip. di Matematica e Informatica | voice: +39 (0432) 55.8445 Università di Udine | fax: +39 (0432) 55.8499 Via delle Scienze, 206 | email: brajnik@uniud.it Loc. Rizzi -- 33100 Udine -- ITALY | http://www.dimi.uniud.it/giorgio
Received on Tuesday, 2 July 2013 10:28:45 UTC