E-mail List Archives

RE: Accessible Matrix Forms


From: Jukka K. Korpela
Date: May 12, 2006 2:40PM

On Fri, 12 May 2006, Andrew Kirkpatrick wrote:

> Radio buttons that aren't in a matrix have two labels also.

The point is that when you don't use a matrix, you can assign a label to
each radio button without difficulty. In a matrix, you would have to put
the label into the cell where the button resides.

> Using
> fieldset is one solution with the legend for the radio button prompt in
> the row heading.

I don't see how that would affect the label issue.

> I would make this without a table, using fieldset for the prompt and
> either use explicit labels that are moved offscreen or use the title
> attribute on the radio input.

We're more or less supposed to use a fieldset element for a set of radio
buttons, though in practice, a heading or just heading-like text is almost
as good as a legend - and causes fewer protests from visual designers.

Using something and moving it offscreen sounds trickery, and it _is_
trickery. So is the title attribute, which is supposed to be, and which is
in reality, an advisory title that may or may not be accessible to the
user and noticed by him. Such ideas are based on mental models that only
cover "normal" use of a graphic browser on one hand and some assumed
assistive technologies du jour on the other.

> Jim Thatcher will chime in soon, but the
> title attribute is read by default on form controls, irrespective of the
> user's verbosity setting for title attribute reading.

On some assistive technologies, yes. That covers about 1 % of users or
less. Many other users have accessibility problems too. In particular,
labels would be relevant to people who wish to use a graphic browser with
a mouse but have a minor motoric disability, so that it is difficult to
click exactly on a small button.

My point is that this is clumsy and unreliably trickery, aiming at making
an imitation of a paper questionnaire more accessible instead of designing
a form that works well in the web context.

Jukka "Yucca" Korpela, http://www.cs.tut.fi/~jkorpela/