Surveys are increasingly being conducted online, and it is pertinent to establish clear guidelines for presenting self-administered survey items on a computer screen. Goals should include reducing respondent burden and measurement error. Web survey designers often need to decide how to best present long lists of information and where to place navigation buttons (i.e., next and previous). We evaluated the usability of web-based National Survey of College Graduates (NSCG) questionnaire prototypes. In the first round of testing (n=8), respondents had difficulties proceeding through the survey because the ‗Next‘ button was on the left side of the screen and the ‗Previous‘ button was on the right. In the second round of testing (n=30), four versions of the survey were tested to assess usability of the ‗Next‘ and ‗Previous‘ buttons based on their placement on the screen (right or left side) and additionally, to assess the display format of a long list of occupation options (one-column versus two-columns).

Dependent measures included participants‘ comments from a think aloud protocol, self-reported ratings of satisfaction with the survey, responses to qualitative debriefing questions, time required to complete the job code item, and eye-tracking data focusing on which button (i.e., ‗Next‘ or ‗Previous‘) and which half of the job code list participants looked at first and more often. Qualitative and quantitative results revealed that participants preferred and performed quicker when the long list of occupation options was displayed in two columns rather than one and when ‗Next‘ was displayed to the right of the ‗Previous‘ button rather than vice versa. Our findings support usability best practice guidelines that recommend eliminating excessive scrolling on Web sites, and following reading conventions (i.e., looking to the right to move forward, as if turning a page in a book).