Creating accessible online surveys

When creating online surveys it is important to consider the needs of disabled users and ensure that your survey is as accessible as possible. Importantly, an accessible survey will be easier for all users to complete and not just those users with a disability.

Accessibility is not just a technical issue. Creating an accessible online survey involves careful survey design in addition to the technical delivery of a survey. A poorly designed survey may be inaccessible no matter how it is delivered.

This document focuses on some of the design and technical issues associated with creating accessible online surveys.

Survey design issues

Use simple language and question structure

Using the clearest and simplest language possible has benefits for all users and not just those with a disability. You should consider the following to ensure that your surveys are easy to understand:

• Use simple language and short sentences
• Use sections and notes to give your survey structure and provide helpful information
• If you need to present a large amount of text consider breaking this up into a bulleted list
• If you use acronyms expand these the first time they are encountered on a page
• Provide an indication of how long the survey is and/or how much the respondent has completed
• Make sure the text used for possible answers is clear and unambiguous

One common mistake is to ask respondents to rank items using a numeric scale which is defined in the question text. A user with a short-term memory impairment may not be able to remember what the options relate to and will have to keep referring back to the scale for each question.

A better approach would be to avoid using a separate scale and instead use explicit wording for each answer. The user would then be reminded of the possible responses on a question by question basis:

Images, audio and video

When using images, audio and video within surveys authors must provide a text equivalent for this content.

For images, authors must provide an accurate text alternative. The survey software may prompt authors to provide a text alternative for an image but it's the author's responsibility to make sure that the information entered is appropriate.

If a survey author is including audio or video clips within their survey then they must provide a text equivalent (such as a separate transcript or captioning).

Question formatting

Many survey authoring tools like Bristol Online Surveys (BOS) provide some formatting options which, when used correctly, can make it easier for users to complete surveys.

For example, you can chose to present multiple-choice answers horizontally or vertically.

For a simple 'yes/no' question, a horizontal format might work well. However, if you have many answers, it's often easier for respondents with a visual impairment to read the different options if they are arranged vertically.

BOS allows survey authors to provide emphasis (text usually displayed in italics) and strong emphasis (text usually displayed in bold). This can be used to provide emphasis for particular words but should not be used on whole sentences or all answers as this makes the text harder to read. For a similar reason we recommend you avoid using BLOCK CAPITALS.

Complex question structures

Many survey tools, including BOS, allow survey authors to create grids of one or more questions.

Grids are very flexible but this flexibility can be used to create extremely complicated questions (both from a design perspective and from a technical perspective in terms of explicitly associating the question text with the relevant form control). Consequently grids can pose significant barriers for disabled users.

A further issue is that some authors create very large grids. We have seen grids containing five rows and fifteen question columns. This results in 5 x 15 = 75 individual input elements (check boxes etc.) and it is doubtful that any user, irrespective of disability, would be able to complete this easily.

To ensure that your survey is as accessible as possible you should not use grids to format your questions. Since grids are just a way of combining/presenting other question types we recommend separating out the individual questions and allowing users to complete them individually.

Technical delivery of surveys

The technical delivery of surveys differs between survey tools, the following section focuses solely on Bristol Online Surveys.

Form controls

Bristol Online Surveys explicitly associates labels with their associated form controls for all question types with the exception of questions in a grid format. As previously discussed, when creating accessible surveys we recommend that you do not use grids as the complicated structure can cause significant issues for some user groups.

Surveys delivered using Bristol Online Surveys should be operable without the use of a mouse and respondents can tab between questions and submit surveys using keyboard input only.

Similarly, if a browser or other user agent does not support JavaScript then this should not present a barrier to the completion of a survey. The date question type does use a JavaScript "date picker" but these questions also provide a description of the response format required (e.g. DD-MM-YYYY) so respondents can enter text in the correct format if the JavaScript date picker is not available.

Semantic structure and headings

The Bristol Online Surveys tool is designed to produce semantically correct and valid XHTML. However it is possible for survey authors to add their own HTML using the "note" feature so the survey output can be affected by survey authors.

Similarly, BOS produces surveys that have a correct document hierarchy (H2 following a single H1 etc.) but survey authors can accidentally leave out some heading levels (by not using sections or other titles as intended within their surveys) or introduce other headings using "notes". In general these changes from the default output of BOS should not pose significant accessibility issues but authors are encouraged to pilot their surveys with as many different user groups as possible before launching.

Colour schemes and text

If a survey is delivered using BOS's default colour scheme (as opposed to a custom theme) then the colour scheme of the headings and survey components adheres to the AA contrast level set out in the Web Content Accessibility Guidelines version 2 (WCAG2) and tested using the Snook Colour Contrast Checker.

The ratios are as follows:

• Large text (bold text of 120% and 130% in size) has a contrast ratio of 3.66:1 (i.e. greater than 3:1).
• Section headings, question text and possible answers have contrast ratios greater than 4.5:1 as required by the WCAG2 guidelines. Specifically the ratios are:
  • Section text: 5:1
  • Question text: 10:1
  • Possible answers: 10:1

Please note, these contrast values apply to the default colour scheme used to deliver BOS surveys. If your account has a custom colour scheme tailored to match that of your organisation's web site or corporate colour scheme then you will need to check the contrast ratios separately.

In terms of contrast, high contrast can aid users with a visual impairment but, conversely, high contrast can pose problems for people with dyslexia. Consequently colour schemes used to aid one particular group of people can cause problems for another.

In the case of the standard colour scheme for surveys delivered through BOS we have aimed for a balanced approach. The contrast between the background and the various text used in a survey remains high but similarly we don't present question text or possible answers on a white background and this should aid dyslexia sufferers. Most importantly, BOS surveys can be completed with style sheets disabled or with a user specified style sheet applied. So, whilst we've tried to achieve a good balance, it is also possible for respondents to over-ride the colour scheme completely should they need to.

We have adopted a similar approach with the text size used in surveys. We have aimed to provide a good balance between attractiveness and usability but have also made sure that users can increase the font size using the functionality within their browser should they need to.

We have not provided a separate tool to allow users to increase/decrease font size within the survey page as this would merely replicate functionality already available within the browser. Further information on this specific topic can be found in the following FAQ: Can I change the font size in my surveys?

Conclusion

When creating online surveys it is important to consider a range of different factors and also to test/pilot your survey with as many users as possible. Authors can perform some tests themselves (such as ensuring the survey can be completed using only the keyboard) but making sure your survey works with particular groups of users is best achieved through testing by representatives of those groups. In addition, testing should identify any issues related to the text used in the survey and the survey can then be changed prior to launching the final survey.

Survey accessibility is something that's very important to the BOS team and we'd very much welcome feedback .