IT Accessibility Policies and Practices in Higher Education: Research Findings Terrill Thompson University of Washington [email protected] @terrillthompson http://terrillthompson.com The Article • Thompson, T., Comden, D., Ferguson, S., Burgstahler, S., Moore, E. (2013). Seeking Predictors of Web Accessibility in U.S. Higher Education Institutions. Information Technology and Disabilities E-Journal, 13(1). • Available online at: http://itd.athenpro.org/volume13/number1/t hompson.html Research Questions 1. What is the current state of web and PDF accessibility at higher education institutions in the United States? 2. How many higher education institutions in the United States have policies related to web and/or IT accessibility? 3. How do institutions compare on their level of conversation related to web and technology accessibility? 4. Which independent variables are the best predictors of web and PDF accessibility? The Population (n=3251) • • • • • • 1204 Associate's Colleges 623 Master's Colleges and Universities 594 Baccalaureate Colleges 518 Special Focus Institutions 283 Doctorate-granting Universities 29 Tribal Colleges • Includes only institutions located in the 50 U.S. states • Individual institutions with a shared parent institution and web template (e.g., ITT Technical Institute, Devry) were evaluated as one institution Research Sample by Google Search • Top 10 HTML pages at each institution • Top 10 PDFs at each institution • A count of number of results searching for "web accessibility" and "technology accessibility" (Conversation) • A list of URLs searching for "web technology accessibility policy" Auto-evaluated each web page (n=31,701) • • • • • % of <img> elements with alt attributes % of <input> elements with "labels" Page has HTML headings (Y/N) Page has a lang attribute (Y/N) Page has ARIA landmark roles (Y/N) Auto-evaluated each PDF (n=28,395) • One measure: Is PDF tagged? (Y/N) • Searched for string "/Marked true" – A reasonably reliable method of identifying tagged PDF – 100% reliable with manual test of 100 PDFs (50 tagged & 50 untagged) Accessibility Links on Home Pages • Automatically checked all links on institutional home pages for the word "Accessibility" Manually Reviewed Policy Results • A team of reviewers generously flagged all Google results that might be considered a policy • Generally reviewed only top 10 search results at each institution, but were given latitude to follow obvious leads one level deep • A panel of three experts conducted a more thorough analysis of all policies that survived the cut. Four Categories of "Policies" • • • • Formal-standalone Formal-incorporated Standards or guidelines General statement Results for Research Question 1: What is the current state of web and PDF accessibility at higher education institutions in the United States? Overall Web Accessibility • • • • • • • Headings – 77.9% (83.6% for Doctorate) Alt text – 60.4% (66.1% for Doctorate) Labels – 39.8% (44.2% for Doctorate) Lang – 37.3% (45.6% for Doctorate) ARIA – 3.3% (4.2% for Doctorate) Overall (Web) – 43.7% (48.8% for Doctorate) Tagged PDF – 33.8% (20.7% for Doctorate) A Few More Observations • California State University, State of Texas, and State of Illinois institutions significantly higher than average overall • Slight but significant difference between highest-scoring region (Pacific) and lowest (Southeast and Southwest) • In all of the above groups, very high withingroup variance On PDF Accessibility • Highest scores were Tribal Colleges (40.7%) • Second highest were Associates (39.7%) • Explanation? – Microsoft Word 2010 for Windows – Automatically creates tagged PDF – Larger institutions are creating more complex PDFs using a wider variety of tools • A manual inspection of a few revealed that some really are accessible, but many are not Results for Research Question 2: How many higher education institutions in the United States have policies related to web and/or IT accessibility? Policies Found • • • • • • • 8.4% of institutions (274 policies) 26.1% of Doctorate (74 policies) 14.8% of Masters (92 policies) 6.3% of Associates (76 policies) 3.5% of Baccalaureate (21 policies) 2.1% of Special Focus (11 policies) 0 Tribal Colleges Types of Policies • Doctorate: – – – – 47.3% are formal-standalone (35 policies) 24.3% are formal-incorporated (18 policies) 14.9% are general statement 13.5% are guidelines and standards • Masters: – Equal numbers of standalone and incorporated (39.1%, 36 policies) • General statements were the most common type of policy among other types of institutions Results for Research Question 3: How do institutions compare on their amount of conversation related to web and technology accessibility?? A Huge Range of "Conversation" • Approx. 1000 institutions have 0 results for "web accessibility" or "technology accessibility" • Approx. 100 institutions have over 1000 results • Six institutions have more than 10,000 results • Five of these have "accessibility" links on their home pages (67.4% of these = "Accessibility") • The one that doesn't: Penn State • 17 of the top 50 institutions do not have a home page link Results for Research Question 4: Which independent variables are the best predictors of web and PDF accessibility? A policy's effect on conversation Institutions with a policy are: • Significantly higher on alt text • Significantly higher on labeled input fields • Not significantly different on other web accessibility measures (headings, lang, ARIA) • Significantly less likely to have tagged PDF Effects of type of policy • Effect size are small, with high variance with groups. However… • Formal standalone policies are related to greater accessibility • Formal standalone policies are highly related to higher levels of conversation. This effect is especially strong when there is no Accessibility link on the home page. Stepwise Multiple Regression Analysis • Great overall web accessibility scores are associated with: – – – – – – More web accessibility conversation Having an accessibility policy of any type Being a Masters or Doctorate Institution Being located in California Not being located in the Southeast or Southwest Having an accessibility link on the home page • Together, all of these predictors only accounted for 3.3% of total variance. Which means… The best predictors of web accessibility were not measured in this study.