A few weeks ago, we decided to take a poke around at some presidential candidate’s websites. What we found was nothing short of surprising, so we wanted to share some of what we found.
To perform this analysis, we used Tenon.io (that’s us, of course) to analyze all public facing URLs of each candidate’s site. As always, we must caution that there’s only so much that automated testing can do. Our view is that automated testing is best as a diagnostic tool, not as a compliance measure. Still, it is interesting to see the results to compare the sites against each other. I think we can say that given that each site was tested against the identical set of criteria, a relative judgment of performance can be made.
| Candidate | Average Errors | Density |
|---|---|---|
| Ben Carson | 2 | 76% of pages have 1-10% Density |
| Bernie Sanders | 9 | 73% of pages have 11-20% Density or less |
| Chris Christie | 14 | 98% of pages have 21-40% Density |
| Donald Trump | 6 | 53% of pages have 1-2% Density; 96% of pages have 30% density or less |
| Hillary Clinton | 27 | 33% of pages have 81% Density or higher |
| Marco Rubio | 19 | 77% of pages 31-60% Density |
| Martin O’Malley | 23 | 85% of pages 30% Density or better |
| Ted Cruz | 27 | 91% of pages are 31-40% Density |
| Rand Paul | 32 | 72% of pages are > 100% Density |
| Mike Huckabee | 38 | 81% of pages are 81% Density or worse |
About the findings
The first thing to note is that the number of errors listed is the average among each page of each site. Given that most of these sites had hundreds of pages, the average error count is a pretty good number to compare against. But a much better indicator of performance is issue density. Issue density basically measures saturation of accessibility issues and it is with density that we really found differences.
Anaysis
Given the above data, we were pretty shocked at what we found, because the findings really flew in the face of our own political biases. We were pleasantly surprised by some and disappointed in others. The positive news, overall, is that most of the presidential candidates perform better than the global average. Across all pages tested by Tenon, we find an average issue count of 51 issues per page. Each candidate performs much better than that. Unfortunately a few did very poorly when it comes to issue density.
The winners
Ben Carson and Donald Trump were the big surprises from this testing. Conservatives are not well known among the accessibility community for having a lot of empathy for people with disabilities so for these candidates’ sites to outperform all of the other candidates by such a large margin was pretty surprising. Both of them have significantly better issue counts and issue densities than the rest.
The losers
Mike Huckabee, Rand Paul, and Hillary Clinton performed especially poorly. Although their issue counts are lower than the global average, their issue densities are terrible. 72% of Rand Paul’s tested pages are more than 100% issue density. There are pages on Hillary Clinton’s site that have over 200 issues. Even that can’t top Rand Paul, however. With 72% of the pages on his site having over 100% issue density, his site would be a contender for some of the worst sites ever tested.
The rest
The rest of the candidates fared pretty well. Martin O’malley, Chris Christie, and Bernie Sanders stand out a bit as “almost winners” for having lower-than-average issue densities. Even though their issue counts were much higher than Carson and Trump, their issue densities were still quite good especially when weighed against the losers.
A caution
Not all issues are equal. Taking a closer look at some of the issue results we noticed some patterns that may artificially skew the results to look more negative than they are. This, again, is evidence that automated testing is best used as a diagnostic tool and not as a judge. One important case-in-point: “Link text with identical title”.
A link text with identical title looks like this:
<a href="/path/to/link" title="Home">Home</a>
This type of issue accounts for 33% of the issues found on Hillary Clinton’s website. This is an artifact of old thinking still in use by many WordPress theme developers. There was a time when this pattern was perpetuated in WordPress core, but thankfully that is no longer the case. Still, many WordPress themes still carry on the practice for some unknown reason. This practice is unquestionably useless and a possible source of annoyance for users of screen readers. But, the pragmatist in me says that among the many other ways people can make their website inaccessible this isn’t even close to the worst. This doesn’t give the Clinton team a “bye” on their site’s accessibility. We still found 4542 images without alt attributes, 1061 form elements without labels, and more.
Conclusion
Like most other websites out there, Presidential candidates’ websites are hit-or-miss, with mostly hits. Unfortunately, they still exhibit the same kinds of silly development patterns that could be resolved by closer attention to semantics and frequent testing.
For an analysis of a broader array of accessibility topics of the candidates, check out Thomas Logan’s post Accessibility 2016: Where Do the Candidates Stand?