Web Survey Considerations
by David Kimmel
If youve spent much time surfing the Web recently, you may have noticed that Web surveys are on the rise. By nature, all surveys, whether they are in print or online, have two competing objectives:
From a usability perspective, this conflict should be transparent to the user, but many
times it is not. This article focuses on three major factors that influence the user
experience of a Web survey.
Chunk Survey Content
You are undoubtedly familiar with the concept of chunking informationthe process of breaking down large amounts of information, which is often very complex, into small manageable pieces. For long technical documents, the benefits of chunking are palpable. But why is chunking important for a Web survey?
The answer can be derived from a combination of the overall number of questions, the average length of the questions, and how closely the questions relate to one another. The survey architect must group the questions in a way that is meaningful to the user, both in the order they are presented and in their proximity to one another. If possible, related questions should be located on the same page of a multi-paged survey. Also, survey pages should require little if any scrolling. The pages should also provide a way for users to orient themselves throughout the survey.
Choose Appropriate Controls
One quantitative measure of survey success is its completion rate. The design of the questions, from word choice, to question length, to the actual controls that are used to represent the users' responses, play a pivotal role in the completion rate of the survey. Anything that can be done to enhance completion rate will enhance its success. For example, if you need to represent a list of possible responses to a question, you may have a choice between using a drop-down list box or a series of radio buttons.
When deciding whether to use a list box or radio buttons, you should consider the number of responses and the length of the individual responses. Radio buttons works especially well if there are only two possible choices or the individual responses are longer than a few words. List boxes are effective for long lists of items that can be quickly scanned and arranged alphanumerically. List boxes become more cumbersome as the length of the individual responses increases because the list box must grow to the length of the longest response. On the Web, list boxes may also require more mouse clicks to complete a response when compared to radio buttons because one click may be required to open the list box and another to make a selection. When using a list box, provide plenty of space around it to prevent occluding the text beneath it when it is opened.
Consider the spatial proximity and alignment of radio buttons to prevent users from
providing unintended responsesread Florida butterfly ballot. Both
controls must provide salient feedback to the user that the system has accepted the
response they intended.
Implications of Fitts Law
As you design the questions, it is important to consider the implications of Fitts Law. Fitts Law states that the time it takes for the user to hit a target is a function of the distance to and the size of the target.
In Figure 1, a list box may present a larger target than a cluster of radio buttons. Any benefit realized by its larger size may be lost if it takes the user longer to open the list, scan the responses, and complete the question. Radio buttons may require greater precision, depending on the distance between the individual buttons and the surrounding text (including text labels).
Keep Users Oriented
The size of the survey window is critical to the users perception of the survey. Often you will need to decide whether to use a small window with very little scrolling that allows you to keep the primary Web site in view or a large window that would require no scrolling but obscure the primary Web site and the desktop. The goal is to strike a balance between the overall window size and the formatting of the individual questions while limiting scrolling and reducing the total number of pages.
Another aspect of the design to consider is how to keep the users oriented as they complete the survey. On a paper survey, an obvious way to do this is to number the individual questions. However, if you have a lengthy survey, the number of questions may adversely affect the completion rate. One possible solution is to number each page of questions (assuming you have a series of small pages) and provide an indicator that shows the user where he or she is in the sequence of pages, i.e., page 1 of 5, page 2 of 5, and so forth.
This article has provided a glimpse into some of the issues to consider when developing a survey that is to be delivered through the Web. The primary responsibility of the survey architect when designing a Web survey is to bolster the completion rate by creating effective interaction so the user can respond accurately and efficiently. Key factors to remember when designing a Web survey include:
The design of Web surveys is an area for information architects to flourish as
companies face tougher scrutiny to show that their Web site and their products are meeting
users' expectations while ultimately contributing to the bottom line.
Berkun, Scott (2001, June 22) Fitts Law Applied to the Web. See http://uiweb.com/issues/issue09.htm.
Graham, Jeffrey (2001, June 22) Best Practices for Building Web Surveys. See www.clickz.com/mkt/emkt_strat/article.php/820651.
Kirakowski, Jurek (2001, June 22) Questionnaires in Usability Engineering FAQ, 3rd Ed. See www.ucc.ie/hfrg/resources/qfaq1.html.
Rapp, Ethan (2001, June 22) The Catch-22 of Online Survey Research. See www.researchsig.com/sig_article_catch22.htm.