WebAIM: Screen Reader User Survey #3 Results

Some main points:

-JAWS (still most popular reader) is becoming less popular among those who take the surveys, while NVDA use has increased.
I wonder if the increase in VoiceOver is from the increasing popularity of iWhatevers (who all come with it pre-installed)

-IE is still the most popular browser among survey respondents.
(note from me: Some of the free readers, like Gnome Orca, only work with Firefox at the moment (webkit is being worked on).
I hope a good majority of the "Others" users are Opera users)

-The overwhelming majority of respondents have Javascript enabled (98.4%, and an increase from previous surveys).
This should pretty much put to rest the idea that screen readers can't or don't work with Javascript (it's still just being run by the browser).

(note from me: What developers need to be careful with are widgets (like drag-n-drop sliders, calendars, etc) and content being dynamically added/ajax (ARIA attributes can help here). Scripts can interfere with a reader, and the user may not know how to turn Javascript off for that page (happened fairly recently on a mailing list where a script gave regular updates to a small scrollbox with calendar... the reader kept refreshing the page). Testing your scripts in a screen reader is a good idea if you can.)

-Mobile use is up significantly! Nokia is still the most popular (using Nuance as the reader for Symbian systems). About two-thirds use a mobile with screen reader regularly

-Respondents are generally optimistic about future accessibility of the web, though they don't think much improved in the last year

-Respondents believe more accessible web sites, rather than better AT, would have the biggest impact on web accessibility

-While many respondents used social media sites (twitter, facebook, etc), the overall impression is that they are only somewhat accessible.
(me: I wonder if the variety of responses is affected by available plugins and specific features of these sites rather than the basic services.
(note for twitter users: there is Accessible Twitter : web accessibility for the Twitter.com website application, an app that uses the twitter API... very cool))

-While nearly half of the respondents didn't know what to think of HTML5, those who were familiar with it were positive about it

-Awareness of ARIA landmarks has increased since the last study, but usage is uneven

-Page navigation is increasingly relying on headers (h1-h6 tags).
To me, this suggests web pages are also using headers more than they used to (making the technique more useful than in the past).
Where the survey mentions "Find", I don't know if they mean a reader find function or the one built into the browser.

-There's a slight decrease in the use of skip links since the previous survey. But in general usage is still spread around pretty good.
I still include skip links where appropriate, not so much for screen reader users, but for sighted keyboarders who may appreciate it.

-Some respondents try to find and use the mobile version of a site if one exists (generally those with less proficiency with their readers).
I recently saw this suggested on the Orca mailing list by an Orca user as a "tip". I think this adds more fuel to the call by some web standardistas to rethink the complexity of the "desktop" versions of sites we build. The version for mobile is more likely to be a version that does everything users want it to, without the bloat.

-Text-only versions? Responses are mixed. Those with disabilities were more likely to use them than those without, but there was about 30% who rarely or never use them

-Two h1 tags, one for the site name and one for the particular page title, was most popular! If just one is available, users preferred the document/page name to be in the h1 rather than the site name (makes sense)
(from me: If your site name is in your <title> tag, that is the first thing announced when a page is being loaded. Google also looks at the <title>, so using an h1 for the site name on every page really has little value compared to the name of the actual page.)

-Longdesc, which was going to be removed from HTML5 (was not in the earlier specifications, but apparently is back in talks?) has most respondents being positive about it, with a large chunk not knowing much about it at all

Awesome study, keep them coming guys!