testing the theory that required alt = bad

I have been thinking about this issue and come up with a  study that may
provide some indication of the correctness of the theory that forcing
authors to include the alt attribute results in an increase in null alt (on
images that should not have it) and also the frequency of bogus values being
inserted in the alt attribute.

alt being required has been in the (X)HTML specs for some time (10 years?).
If a random sample valid (X)HTML pages is assessed and the alt values are
analysed, it could be discerned what number of images have incorrect uses of
alt="" or alt="bogus" .

This could be compared against  the alt or lack of alt and the correctness
of the alt values on a random sample of non valid pages,

This would provide (i think) a measure that indicates whether forcing
authors to insert an alt attribute actually increases/decreases the
frequency of bogus/null alt values.

If the theory of requiring alt= bad is correct then an increase in incorrect
uses of alt in pages that validate should be apparent.

any thoughts?

-- 
with regards

Steve Faulkner
Technical Director - TPG Europe
Director - Web Accessibility Tools Consortium

www.paciellogroup.com | www.wat-c.org
Web Accessibility Toolbar -
http://www.paciellogroup.com/resources/wat-ie-about.html