Communications of the ACM

Figure 2 shows the difference in adherence between initial and final submissions for the 465 papers accepted for publication at CHI 2014, where authors were given specific details on the accessibility barriers of their respective papers. In four of five recommendations, accessibility of the papers increased 5% to 10% based on authors receiving feedback on accessibility. Unclear is why adherence to one recommendation (tab order) decreased slightly. There may be cases where authors had to update their final submission based on feedback from the publication vendor and forgot to reapply the accessibility changes.

Note that 30% accessibility of pub-
lished papers or even 60% accessibil-
ity is not ideal. The goal, as spelled out
by the SIGCHI Accessibility Commu-
nity, is 100% compliance. However,
accessibility is a multi-pronged effort,
and paper accessibility gets attention
because it is an easy-to-measure met-
ric; equally important are many other
details we have discussed here (such
as having accessibility chairs at each
conference, proper information flows,
and accessible physical locations).
For instance, in choosing the site for
the CHI 2019 conference—Glasgow,
U.K.—accessibility criteria were specif-
ically taken from the city’s proposals,
as well as from on-site walkthroughs,
which led to one city with a fully acces-
sible conference venue being chosen
Giving authors individual notification
of their papers’ accessibility between
acceptance and camera-ready submis-
sion in 2014 clearly increased the level
of accessibility compliance. While ac-
cessibility of papers did increase, 16%
to 26% is still not ideal, with a long way
to go. As a comparison, we analyzed the
accessibility of published papers from
the ASSETS 2015 conference, though
the sample size for ASSETS papers was
31, much smaller than the number of
CHI papers in any given year. ASSETS
generally uses two different approach-
es that have not yet been attempted by
the CHI conference: The first is that
authors are required (not just encour-
aged) to make their papers accessible
and the second that SIGACCESS, spon-
sor of the ASSETS conference, specifi-
cally requires the company that is con-
tracted for publishing, Sheridan, to
manage the accessibility process and
check for accessibility. We do not know
the specifics of what is required in its
contracts with Sheridan, and it is pos-
sible Sheridan is required to check for
different accessibility features than in
our evaluation. Given identical criteria,
compliance for ASSETS 2015 papers
was much higher than for CHI papers
(in any given year) but still not at the
100% goal. In 2015, 74.1% of the AS-
SETS papers had alternative text and
table headers, 93.5% had generated a
tagged .pdf file, and 90.3% had default-
language information included in the
.pdf, but only 51.6% of ASSETS 2015 pa-
pers had a correct tab order.

Figure 2. Difference in adherence among the 465 accepted papers for CHI 2014 between submitted and final versions (%). The bars here are likewise covered in patterned fill, rather than colors, to make the graphs more inclusive for colorblind readers.

0

5

10

15

20

25

30

Alternative Text Table Headers Tagged PDF Default Language Tab Order

Submitted Published

Figure 1. Percentage of published papers that adhered to each of the five recommendations (%), 2010–2016. The bars here and in Figure 2 are covered in patterned fill, rather than colors, to make the graphs more inclusive for colorblind readers.